EN中文

Journal of Geographical Sciences >

2016 , Vol. 26 >Issue 2: 171 - 187

DOI: https://doi.org/10.1007/s11442-016-1261-8

Orginal Article

Model-based analysis of spatio-temporal changes in land use in Northeast China

  • XIA Tian , 1, 2 ,
  • *WU Wenbin , 1, 2 ,
  • ZHOU Qingbo 2 ,
  • Peter H. VERBURG 3 ,
  • YU Qiangyi 2 ,
  • YANG Peng 2 ,
  • YE Liming 4, 5
Expand
  • 1. Key Laboratory for Geographical Process Analysis & Simulation, Hubei Province / College of Urban & Environmental Sciences, Central China Normal University, Wuhan 430079, China
  • 2. Key Laboratory of Agri-Informatics, Ministry of Agriculture / Institute of Agricultural Resources and Regional Planning, Chinese Academy of Agricultural Sciences, Beijing 100081, China
  • 3. Institute for Environmental Studies, VU University Amsterdam, de Boelelaan 1085, 1081 HV Amsterdam, The Netherlands
  • 4. CAAS-UGent Joint Laboratory of Global Change and Food Security / Institute of Agricultural Resources and Regional Planning, Chinese Academy of Agricultural Sciences, Beijing 100081, China
  • 5. Department of Geology and Soil Sciences (WE13), Ghent University, 9000 Gent, Belgium;

Author: Xia Tian (1981-), Associate Professor, specialized in remote sensing monitoring and simulation of the influence of the global changes in agriculture. E-mail:

*Corresponding author: Wu Wenbin, Professor, E-mail:

Received date: 2015-05-25

  Accepted date: 2015-09-16

  Online published: 2016-02-25

Supported by

National Natural Science Foundation of China No.41201089.No.41271112

The Fundamental Research Funds for the Central Universities, No.CCNU15A05058

National Nonprofit Institute Research Grant of CAAS, No.IARRP-2015-28

Copyright

Journal of Geographical Sciences, All Rights Reserved
Fold

Abstract

Spatially explicit modeling techniques recently emerged as an alternative to monitor land use changes. This study adopted the well-known CLUE-S (Conversion of Land Use and its Effects at Small regional extent) model to analyze the spatio-temporal land use changes in a hot-spot in Northeast China (NEC). In total, 13 driving factors were selected to statistically analyze the spatial relationships between biophysical and socioeconomic factors and individual land use types. These relationships were then used to simulate land use dynamic changes during 1980-2010 at a 1 km spatial resolution, and to capture the overall land use change patterns. The obtained results indicate that increases in cropland area in NEC were mainly distributed in the Sanjiang Plain and the Songnen Plain during 1980-2000, with a small reduction between 2000 and 2010. An opposite pattern was identified for changes in forest areas. Forest decreases were mainly distributed in the Khingan Mountains and the Changbai Mountains between 1980 and 2000, with a slight increase during 2000-2010. The urban areas have expanded to occupy surrounding croplands and grasslands, particularly after the year 2000. More attention is needed on the newly gained croplands, which have largely replaced wetlands in the Sanjiang Plain over the last decade. Land use change patterns identified here should be considered in future policy making so as to strengthen local eco-environmental security.

Key words： land use change; CLUE-S; Northeast China; two-way simulation

Cite this article

XIA Tian , *WU Wenbin , ZHOU Qingbo , Peter H. VERBURG , YU Qiangyi , YANG Peng , YE Liming . Model-based analysis of spatio-temporal changes in land use in Northeast China[J]. Journal of Geographical Sciences, 2016 , 26(2) : 171 -187 . DOI: 10.1007/s11442-016-1261-8

1 Introduction

Human land use practices have transformed a large proportion of the planet’s land surface (Wu et al., 2007), and have major consequences for the environment through the alteration of the structure and functioning of ecosystems, and influences how ecosystems interact with the atmosphere, aquatic systems and surrounding land (Foley, 2005). Thus, the land system science community has sought to understand land use dynamics over time, space and scale (Turner II et al., 2013; Verburg et al., 2013). Considerable researches have investigated and analyzed past and present geographic distributions of land use and its dynamics with the aid of remote sensing, statistical methods and a combination of these (Gong et al., 2013; Gutman et al., 2013; Sleeter et al., 2013; Verburg et al., 2011; Wu et al., 2008). However, these studies are often labor intensive and time consuming, especially when applied to large areas. Land use changes are driven by a number of biophysical and socioeconomic factors (Lambin et al., 2001). When these factors are properly identified from past and present observations, and the causal relationships between these factors and land use changes are well constructed, changes in land use can be modeled using these derived relationships (Rounsevell M D A et al., 2006). Spatially explicit modeling techniques thus recently emerge as an alternative to monitoring and mapping approaches, as they are capable of representing land use change and its possible directions (Yu et al., 2012).
Many studies have used a modeling approach to evaluate land use changes and their effects over multiple scales (Letourneau et al., 2012; Rounsevell et al., 2003; Schaldach et al., 2011; Veldkamp et al., 1996; Verburg et al., 2002; Verburg et al., 2006). The methods used, which often require modification for local conditions, are normally either process-based (e.g., an economic or agent-based model) or pattern-based (e.g., a machine learning model) (Brown et al., 2013). Process-based models are often derived from micro- or macroeconomic theory, or use a multi-agent simulation approach (Verburg et al., 2004). They normally start from the viewpoint of individual landowners who make land use decisions with the objective of maximizing expected returns or utility derived from the land (Lambin et al., 2006). The challenge is to obtain sufficient data at the individual/household level to develop a well-parameterized and validated model of decision making. The behavioral assumptions are then valid for micro-level analysis, which restricts these models only be used in a small area. Conversely, pattern-based models, such as the CLUE-S model (Verburg, 2002), GEOMOD2 (Pontius Jr et al., 2001) and LOV (White et al., 2000), usually rely on empirical-statistical approaches to analyze the spatial structure of land use change. The underlying assumption is that the observed spatial relations between land use types and their driving forces represent currently active processes, and will remain valid in the future. This flexible approach is helpful in identifying key processes, and thus facilitates an understanding of the dynamic behavior of land use patterns on a large scale, or for some “hot-spot” areas (Bakker et al., 2012; Li et al., 2012).
Northeast China (NEC), located at relatively high latitudes (38°40°N-53°34°N) (Figure 1), is such a “hot-spot” area with vast importance for Chinese food and ecological security. It covers a total territory of 791,800 km2 and comprises Heilongjiang, Jilin, and Liaoning provinces. Most areas in NEC have a humid or semi-humid monsoon climate with annual temperatures ranging from -6 to 11°C. The average annual rainfall in this region ranges from 400 to 1,000 mm and the spatial pattern of rainfall decreases from the southeast to the northwest, with most rain falling in July and August (Li Z. et al., 2012). Among the major grain production regions in China, NEC is considered as the region most susceptible to global environmental change. Continuous measurements from meteorological stations indicate that NEC has experienced a warming trend, with an average surface temperature increase of 0.38℃ per decade over the past five decades (Gao et al., 2011; Wu et al., 2014). Previous studies in NEC also show that a large proportion of the land surface has been converted from one land use type to another. For instance, conversions from grassland and wetland to cropland (Liu et al., 2005), together with modifications within cropland, have resulted in a high growth in rice production, particularly over the past 30 years, during which period it has increased from 3% to 13% of China’s total rice production (Xia et al., 2014). Furthermore, NEC has also experienced great socioeconomic changes, e.g., land tenure reforms and widespread adoption of agricultural mechanization and intensification (Yu et al., 2013).
Figure 1 Land use in Northeast China in 2000.
Some studies have analyzed past land use changes in NEC mainly by using remotely sensed images (Liu et al., 2005; Liu et al., 2003). Unfortunately, detailed analyses of the patterns of land use changes in NEC are scarce. Understanding and modeling these land use changes is critical, as this can provide important information about the scope and impact of land use changes, and can support the analysis of the vulnerability of ecosystems and provide insight into how land use might respond to a range of environmental changes. The objective of this study is thus to analyze the spatio-temporal changes in land use in NEC over the period of 1980-2010. To do so, a classic pattern-based modeling approach was used to identify the possible determinants of land use distribution in NEC, and to simulate the dynamic changes in land use over the period of 1980-2010 at a resolution of 1 km, and further to capture the overall patterns of land use changes.

2 Data and methods

2.1 Data

A considerable amount of input data, including both spatial and nonspatial data, were used in this study (Table 1). Spatial datasets included satellite-based land use, a digital elevation model (DEM), temperature, precipitation, soil, roads (levels 1 to 3, China highway classification), rivers, residential areas, population and GDP. Most of these data were collected around the baseline year of 2000. Satellite-derived land use data in 2000 were used as a baseline map to analyze the spatial relationships between land use types and 13 selected biophysical and socioeconomic factors. Two additional land use maps in 1990 and 2005 were used for model validation. Nonspatial data mainly comprise the yearly land requirement for individual land use types. This was compiled from China Statistical Yearbooks and aggregated at the regional level, and then used as the land use demands in a spatial allocation model. All spatial data were converted into GIS grid data with a cell size of 1 km by 1 km in a standard GIS software environment (ESRI ArcGIS 9.3). DEM data were processed to generate aspect and slope. FAO soil data were further divided into 20 soil types according to the soil classification, and each soil type was used as a driving factor in spatial analysis (The names of soil classes are shown in Table 3, designated as Soil_1 to Soil_20) (Ye et al., 2008).
Table 1 List of input data used in this study
Variable Description Resolution Source
Land use data Remote sensing of land use patterns in 1990, 2000 and 2005 (including cropland, forest, grassland, water body, built-up area, wetland and unused land) 1 km Institute of Geographic Sciences and Natural
Resources Research (IGSNRR), CAS
DEM Spatial data, used to generate the aspect and slope 1 km IGSNRR, CAS
Temperature Years average temperature; annual accumulated temperature ≥0℃; annual accumulated temperature ≥10℃ 500 m China Meteorological
Administration
Rainfall Years average rainfall 1 km China Meteorological
Administration
Soil Map; soil type (subclass) distribution FAO soil classification
Road Level 1~3 traffic network distribution National Fundamental GIS
River River water distribution National Fundamental GIS
Residential area Residential distribution National Fundamental GIS
Population Demographic data distribution map, 1 km grid, population: people/km2 1 km IGSNRR, CAS
GDP Data distribution diagram, 1 km grid, GDP unit: million yuan/km2 1 km IGSNRR, CAS
Land use
requirements		 7 land use types, provincial level, 1980-2010 / China Statistical
Yearbooks

2.2 Model-based analysis

2.2.1 CLUE-S model
The well-known CLUE-S (Conversion of Land Use and its Effects at Small regional extent) model was adopted for this study. CLUE-S is a spatially explicit land use model that simulates the dynamic competition between different land use types. The model functions through the spatial allocation of aggregated demands for different land use types to individual grid cells across multiple time and space scales. The computational core of CLUE-S is a top-down allocation module, which allocates the aggregated land claim to the cells year by year in an iterative way. These land use demands impose the quantity of land use change for each land use type. The allocation module incorporates various mechanisms that determine the distribution of land use types appropriate for any given landscape. These mechanisms are parameterized as three categories of model settings and inputs, i.e., location suitability, spatial policies and restrictions, and conversion settings. These three categories together define the preferences and constraints from which the allocation routine determines an optimal solution and assigns a land use type to each grid cell. Detailed descriptions of CLUE-S can be found in (Verburg, 2002; Verburg et al., 2006).
Model simulation
In this study, CLUE-S was run over a period of 30 years (1980-2010), taking the year 2000 as the baseline due to reliable data sources. To ensure realistic and consistent simulated results for land use change analysis, a forward simulation was run for the period of 2000-2010 and a backward simulation was run for the period of 1980-2000. The forward simulation was a simple application of CLUE-S, while the backward simulation was different from the normal application, with some settings and adjustments modified as described below.
Land use requirements (demand)
Land use requirements define the aggregated demands for different land use types, which were input into CLUE-S for spatial allocation. Generally, aggregated land use demands can be either extrapolated from historical data or projected for complex economic models. In this study, seven land use types, i.e., cropland, forest, grassland, water body, built-up area, wetland and unused land, were included in the analysis, and their yearly land demands at the regional level were obtained from China Statistical Yearbooks.
Spatial policies and restrictions
Spatial policies and restrictions indicate the specific areas where land use changes are restricted. For instance, in the cropland protection zone in NEC, croplands are firmly protected, and it is prohibited to convert them into other land use types. Moreover, the primary forests in the Great Khingan and Lesser Khingan Mountains remain as permanent nature reserves following the regulations of the Khingan Mountains ecological nature reserve. All these areas were restricted to their current land uses in the simulation process.
Land use conversion settings
Conversion elasticity and transition sequences are set for individual land use types. Conversion elasticity defines the reversibility of land use changes, while transition sequences control the various types of land use conversions by a transition matrix. In this study, since the CLUE-S model was run both forward and backward, these two parameters were set independently for forward and backward simulation. Table 2 shows the elasticity coefficients used in this study. It can be seen that built-up areas can rarely be converted into other land use types in the forward simulation, but they are easily converted from others in the backward simulation. To reflect this, the elasticity coefficients were set as 0.9 for forward and 0.5 for backward, respectively.
Table 2 Settings of conversion elasticity
Simulation Cropland Forest Grassland Water body Built-up area Wetland Unused land
Backward 0.6 0.8 0.7 0.8 0.5 0.7 0.8
Forward 0.6 0.8 0.5 0.9 0.9 0.7 0.4
Location characteristics
Location characteristics describe the local suitability for individual land uses under a given selection of location factors. In this study, a binary logistic regression model was used to determine the driving factors for each land use type in 2000. Binary logistic regression is very often used to explore the casual relationship between predetermined variables and changes in land use (Geoghegan et al., 2001; Serneels et al., 2001) using the following formula.
Logit(pi) = ln(pi)/(1-pi)=β0+β1X1+β2X2+…+βnXn, (1)
where pi is the probability of a grid cell being land use type i and Xi are n variables selected as the driving factors, which could be either physical or socioeconomic. Coefficients (β) are estimated through binary logistic regression using the grid-based land use data as the dependent variable and the selected driving factors as independent variables.
In this study, a total of 13 driving factors (including 11 biophysical factors and two socioeconomic factors) were included to statistically analyze the driving factors of land use changes in NEC (a detailed explanation of driving factors is presented in the Results section). The goodness of fit of the logistic regression models was evaluated using the receiver operating characteristic (ROC), which ranges between 0.5 (completely random) and 1.0 (perfect fit) (Pontius Jr et al., 2001). The most important factors influencing the spatial distribution of land use were identified from the logistic regression as those with a confidence level greater than 95% (ɑ≤0.05). These significant factors and coefficients from the logistic regression analysis were combined to determine the location suitability of the seven land use types, which were then used in the spatial allocation.
2.2.2 Model validation
In this study, model validation was performed in two ways. First, CLUE-S model simulation results were compared with Null model simulation results in 1990 and 2005 (Pontius et al., 2004). When the CLUE-S model has higher simulation accuracy than the Null model, the model simulation of land use changes is judged to have scientific implications (van Vliet et al., 2011). Second, the satellite-derived land use datasets for 1990 and 2005 were employed as reference data for validating backward and forward simulation results, respectively. The degree of similarity between simulated and measured values at a regional level indicates whether the CLUE-S model simulation follows the patterns identified by satellite observation. As there is no generally agreed method for evaluating the goodness of fit of a simulation model, a quantitative method developed by Costanza R (1989) was used here. Since measurement at one resolution is unable to effectively determine the performance of a model in estimating complex land use patterns, this method determines the overall accuracy of a model by looking at its simulation accuracy at multiple resolutions. A multiple resolution method can yield additional information that is not contained in single resolution methods, and is necessary to adequately evaluate the performance of complex ecological models (Costanza, 1989).
2.2.3 Spatial analysis
Simulated results for the period of 1980-2010 were overlaid to analyze the general spatio-temporal changes in land use in NEC. The period was divided into three stages: 1980-1990, 1990-2000 and 2000-2010. The overlaid maps were further used to identify the area and spatial changes in individual land use types in that period. Moreover, some regions with typical land use changes were identified based on a detailed analysis.

3 Results and analysis

3.1 Statistical regression in 2000

Table 3 lists the statistically significant driving factors of land use based on logistic regression. In general, the spatial distributions of the seven land use types were statistically well explained by the selected factors, as indicated by high ROC values (ranging from 0.82 to 0.98).
Table 3 β values of location factors in regression results related to each land use type; significant coefficients (P<0.05) are listed
Cropland Forest Grassland Water
body		 Built-up
area		 Unused
land		 Wetland
Constant 47.08 18.38 90.53 -181.19 -2.55 -111.15 -82.06
Aspect 0.0015 -0.0045 0.0028 0.0051 -0.0020 0.0020 0.0038
Elevation (DEM) -0.0029 0.0025 -0.0004 -0.0076 n.s. 0.0040 -0.0036
GDP 0.0008 -0.0033 0.0034 0.0021 0.0126 -0.0006 -0.0034
POP 0.0076 -0.0044 -0.0043 -0.0061 0.0023 -0.0042 -0.0022
Rainfall -0.0003 0.0006 -0.0008 n.s. -0.0004 -0.0011 -0.0004
Distance to residential area 0.0006 0.0007 0.0022 -0.0019 -0.0030 -0.0024 0.0020
Distance to river 0.0004 0.0007 0.0003 -0.0034 n.s. 0.0030 -0.0016
Distance to road n.s. n.s. n.s. 0.0004 0.0004 n.s. n.s.
Slope -0.0295 0.0565 -0.0231 -0.0310 -0.0213 -0.1249 -0.0724
Annual accumulated
temperature ≥0℃		 -0.0539 -0.0848 -0.0772 0.3433 n.s. 0.3600 -0.3730
Annual accumulated t
emperature ≥10℃		 0.0066 0.0627 -0.0100 -0.1623 n.s. -0.2558 0.2983
Annual mean temperature 0.0271 n.s. 0.0301 -0.0694 n.s. n.s. 0.0218
I-Bb-U-C (Soil_1) 0.7175 0.3696 -1.0162 n.s. n.s. 3.3157 1.2644
AO13-3bc (Soil_2) -0.1009 1.3624 -0.5527 0.3061 0.9438 4.2715 n.s.
Ge63-2/3a (Soil_3) 0.1418 0.6060 2.5921 n.s. n.s. 1.8103 n.s.
A080-2bc (Soil_4) 1.0788 n.s. -0.5358 0.6740 -1.1014 2.4519 1.0878
I-Lo-2c (Soil_5) 0.2481 0.4874 1.1863 n.s. n.s. 2.9629 n.s.
Kh1-2b (Soil_6) 0.5820 0.9661 n.s. n.s. 7.8537 6.0074 n.s.
GL (Soil_7) -0.8421 1.6314 2.3325 n.s. n.s. n.s. -5.0107
Be87-2ab (Soil_8) -0.3598 0.8706 2.8974 n.s. n.s. n.s. -3.4772
Gm20-2/3a (Soil_9) n.s. 1.1289 0.1798 0.3058 n.s. 3.7156 0.6415
I-K-2C (Soil_10) -2.7090 1.8606 -1.8423 n.s. 1.3232 n.s. 3.3136
I-Y-2C (Soil_11) 0.4994 0.4537 n.s. 0.5158 n.s. 7.6462 1.1321
I-BH-U-C (Soil_12) 0.5178 0.6890 -0.4127 0.7754 n.s. 4.0914 0.9727
Hg6-2/3a (Soil_13) 0.3548 -0.7586 0.9411 0.4694 n.s. 6.4765 0.5742
J2-2a (Soil_14) 2.3345 n.s. 0.9976 -1.9142 n.s. 2.5272 -1.7082
Lg1-2b (Soil_15) n.s. -18.2998 n.s. 1.0302 n.s. n.s. n.s.
Yh2-1b (Soil_16) n.s. n.s. 0.8014 1.2989 -0.3218 6.4905 1.4371
I-Bh-2c (Soil_17) n.s. 0.3667 n.s. n.s. n.s. 5.9017 1.9701
WAT (Soil_18) -1.0158 -0.6522 -3.3175 n.s. 10.0656 n.s. 2.0486
I-Xl-2c (Soil_19) n.s. n.s. 1.4065 1.3493 1.1067 6.4157 0.8550
I-B-U-2c (Soil_20) -0.9316 -1.2753 -1.3537 3.9771 2.5932 5.5994 n.s.
ROC 0.93 0.95 0.82 0.94 0.98 0.97 0.90

Note: n.s., not significant at the 0.05 level
Soil_1 to Soil_20 show the code for the FAO soil classification, e.g., the FAO Soil code (soil_number)

The highest ROC values were for built-up areas (0.98) and unused land (0.97). Not surprisingly, built-up areas are largely influenced by socioeconomic factors, such as higher GDP, more population, rural residential areas and proximity to highways. Rainfall and slope are also important factors in the selection of built-up locations. Unused land is mainly located in areas that have a long distance from cities and in unsuitable environments with higher elevations (DEM). High to moderate ROC values were found for forest (0.95), water body (0.94) and cropland (0.93). It can be seen from Table 3 that forest in NEC is positively correlated with higher elevation, lower GDP, sparsely populated locations and more rainfall. Distribution of water bodies is positively correlated with lower elevation, lower population and annual accumulated temperature ≥0℃. The regression results show that southward location, low altitude, low slope, annual accumulated temperature ≥10℃ and higher population are the major factors that influence the spatial distribution of cropland. Soil type also has a certain influence on cropland, which agrees with previous studies (Chen et al., 2013; Yao et al., 2015; Zhang et al., 2011). Finally, wetland (0.90) and grassland (0.82) have relatively low ROC values. The factors for wetland are quite similar to those of unused land, as wetland is normally considered to be a specific form of unused land in China. Grassland in NEC is broadly disseminated and highly mixed with other land use types. It is also relatively difficult to describe its spatial characteristics by using location factors.

3.2 Model validation in 1990 and 2005

Figure 2 compares analysis between the model simulation and the satellite observation for year 1990 and 2005. It can be seen that although there were some places where the model simulation deviated somewhat from the satellite observation, in general the simulated and observed land use maps were similar to each other. A comparison indicates that the overall accuracy of CLUE-S simulation is 0.92 and 0.89 for 1990 and 2005, respectively. The Null model has an accuracy of 0.90 and 0.81 for 1990 and 2005, respectively, i.e. lower than that of the CLUE-S model. This suggests that the CLUE-S model has the potential to simulate land use changes. The comparison also shows that the backward simulation has a relatively higher accuracy than the forward simulation. This is due primarily to the difference in the spatial resolution of the reference data used for validation. From the evaluation results described above, it can be concluded that the localized CLUE-S model appears to be adequate for the purpose for which it was designed, and it appears to be useful for the analysis of long-term land use changes in NEC.
Figure 2 Comparison between model simulation (left) and satellite observation (right) in 1990 and 2005

3.3 General patterns of land use changes in NEC

Table 4 presents the general trend in changes in aggregated areas for each land use type in NEC for the overall period of 1980-2010. It can clearly be seen that in general, all land use types showed an obvious change in total area, but there were noticeable differences among the seven land use types for the three sub periods.
Table 4 Conversion matrix between seven land use types (km2)
From
To		 Cropland Forest Grassland Water
body		 Built-up
area		 Wetland Unused
land		 Sum of
decrease
1980
-
1990		 Cropland - 0 0 0 12 0 0 12
Forest 5468 - 0 11 0 0 5 5484
Grassland 4394 0 - 0 214 0 0 4608
Water body 758 0 0 - 0 0 0 758
Built-up area 0 0 0 0 - 0 0 0
Wetland 1231 6 7 0 9 - 0 1253
Unused land 739 11 0 0 18 0 - 768
Sum of increase 12590 17 7 11 241 0 5 -
1990
-
2000		 Cropland - 5630 680 529 152 921 78 7990
Forest 9619 - 1003 208 8 752 113 11703
Grassland 5896 283 - 58 58 35 2 6332
Water body 1415 92 24 - 0 11 0 1542
Built-up area 0 0 26 0 - 0 2 28
Wetland 1015 31 2 17 1 - 15 1081
Unused land 2632 196 0 0 4 0 - 2832
Sum of increase 20577 6232 1735 812 223 1719 210 -
2000
-
2010		 Cropland - 17140 5277 0 5711 360 1833 30321
Forest 7774 - 5275 0 4469 298 637 18453
Grassland 6324 13757 - 0 907 0 4703 25691
Water body 1428 162 280 - 0 588 26 2484
Built-up area 22 1 6 0 - 0 3 32
Wetland 9406 3068 4068 2 71 - 218 16833
Unused land 1218 162 10181 0 669 110 - 12340
Sum of increase 26172 34290 25087 2 11827 1356 7420 -
From 1980 to 1990, given the rapid development of agriculture in NEC, land use changes featured an obvious expansion of cropland. This expansion largely occurred in the western part of NCE, as shown in Figure 3. In total, cropland increased by 12,590 km2, of which 5468 km2, 4394 km2 and 1231 km2 were converted from forest, grassland and wetland, respectively. Very little cropland was lost during this period. A very small amount of grassland was converted into built-up areas due to the population increase.
Figure 3 Spatio-temporal changes in cropland during the period 1980-1990 (a), 1990-2000 (b) and 2000-2010 (c)
During 1990-2000, cropland continued to expand, with the conversion of a further 20,577 km2, but it also lost 7990 km2, resulting in a net increase of 12,587 km2. It can be seen from Figure 3 that cropland expansion was mainly located in the northern and eastern parts, rather than being widespread throughout NEC. Similar to the first period, forest and grassland were the major sources of cropland gain, with 9619 km2 of forest and 5896 km2 of grassland converted into cropland during this period. Figure 4 shows that a part of the forest in the Changbai Mountain region and grassland near Nuluerhu Mountain were converted into cropland. Moreover, some areas of water bodies and unused land in Songnen Plain were exploited and used for cropland. Loss of cropland was largely due to reforestation. About 5630 km2 of cropland was converted into forest mainly in the central part of NEC and in the west of the Changbai Mountain region.
Figure 4 Simulated land use maps in 1980, 1990, 2000 and 2010 for five regions (A-E)
During the recent period of 2000-2010, land use changes were even more profound. Cropland in NEC expanded quickly, in particular in Heilongjiang Province. Surprisingly, these increases largely came from wetland (9406 km2), followed by forest (7774 km2) and grassland (6324 km2). Wetland was most reduced in the Sanjiang Plain, where it was converted into rice fields. During this period, a large amount of NEC cropland was replaced by forest (17,140 km2), grassland (5277 km2) and built-up areas (5711 km2). Figure 4 shows that there were some areas of cropland in the Lesser Khingan Mountains and the Changbai Mountain region that were changed to forest, and conversion from cropland to grassland occurred in the Nuluerhu Mountain region and the Songnen Plain. Contrary to the previous two periods, the total area of cropland in NEC experienced a net decrease of 4149 km2. Forest area increased substantially during the period 2000-2010 at the expense of grassland, cropland and wetland. Built-up areas also increased rapidly in NEC due to urbanization and industrialization, with cropland and forest being the two major contributors to increased construction areas. Wetland and unused land decreased, being replaced by other land use types.

3.4 Land use changes in specific regions

Five representative regions (A-E) in NEC are shown in Figure 4. These were selected for in-depth analysis of land use changes during the period of 1980-2010. The Lesser Khingan Mountains region (region A in Figure 4) is abundant in forest resources, most of which are primary forest. The simulated results show that there was little land use change between 1980 and 2000, yet since 2000 a large amount of grassland and wetland surrounding the existing forest areas has been converted to forest. This is largely related to regional policy. The State Forestry Administration launched a new Forest Protection Project in 1998, and this started to take effect in 2000, which caused increases in forest resources in this region. Meanwhile, some sparse forests were converted into cropland in the transitional zone between forest and cropland. Region B in Figure 4 is the Songnen Plain region, one of the most important food producing regions in China. In addition to the major land use types (i.e., cropland, grassland and wetland), unused land, mainly covered by saline and alkaline soils, is also widespread (Yao et al., 2015). Commencing in 2000, the local government invested heavily in an attempt to solve this land degradation, with several ecological construction projects implemented (Ye and Van Ranst, 2009; Ye et al., 2008). As a result, large areas of unused land were transferred to grassland between 2000 and 2010, which is shown by the simulated results in Figure 4. Another trend in land use change in this region is the rapid development of urbanization, driven by a series of new policies for economic revitalization. Under these conditions, built-up areas increased significantly, mainly at the expense of surrounding cropland and grassland.
The Sanjiang Plain (region C in Figure 4), where the Heilongjiang, Songhua and Wusuli Rivers are confluent, is currently one of the most important grain production bases in China. As the lowest plain in eastern Heilongjiang Province, it embraces the most concentrated and widely scattered wetlands in China. It can be observed from Figure 4 that large areas of wetland and forest have been converted to croplands since 1980 to increase food production. In particular, since 2000 in the northern part of the Sanjiang Plain, rice fields have replaced wetland, as this is an easy, low-cost change. The Nuluerhu Mountain region (region D in Figure 4) is located in the western part of the Liaohe Plain and is bordered by Liaoning Province and Inner Mongolia Autonomous Region. Cropland in this region expanded extensively into grasslands up until 2000. However, after 2000, cropland was gradually converted to forest due to the implementation of the Grain for Green Project. Meanwhile, some grassland was also converted to forest to increase forest resources. The Changbai Mountain region (region E in Figure 4) shows similar trends to the Nuluerhu Mountain region. It experienced an obvious increase in croplands between 1980 and 2000, forest and grassland being the major sources of cropland expansion. After 2000, croplands in the hilly and sloping regions were converted to forest under the Grain for Green Project. Moreover, in response to the blooming of tourism in this region since 2000, some small or medium-sized cities developed rapidly, and built-up areas increased substantially. These newly built-up areas were mainly conversions from cropland and forest surrounding the cities.

4 Discussion

4.1 Two-directional simulation strategy

The CLUE-S model was used to capture overall patterns of land use changes in NEC over a period of approximately 30 years (1980‒2010) at a resolution of 1 km. To do so, the year 2000 was selected as the baseline year due to the lack of land use data and driving factor data for the year 1980. This study is thus characterized by the fact that the CLUE-S model employs a strategy of two-directional simulation. To ensure realistic and consistent simulated results of land use change analysis, a forward simulation for the period 2000-2010 and a backward simulation for the period 1980-2000 were carried out. This is different from previous studies, which normally used the CLUE-S model for forward simulation only. This application of backward simulation in this study is thus a good experiment. In fact, backward simulation differs from forward simulation in model parameterization. The transformation matrix and conversion elasticity had to be set independently according to different characteristics of the two simulation periods. For instance, conversion from built-up area to cropland is very unlikely in the forward simulation, but it frequently occurs in the backward simulation. The validation results showed a fairly high accuracy in both forward and backward simulation, suggesting that the strategy of two-directional simulation can enrich CLUE-S model applications in the future when input data are limited. However, uncertainty also remains due to the hard classification scheme used here. Although it is widely used for mapping land use types, it largely overlooks the representation of land systems (Turner II, 2013; Verburg, 2013) because the landscape is normally a mosaic, and should not be simply designated as one land use type, particularly at a coarse resolution of 1 km2. Future studies should introduce the land system concept into model simulation (Václavík et al., 2013; van Asselen et al., 2013).

4.2 The causes of land use changes

Logistic analysis results showed that biophysical and socioeconomic factors have a great effect on land use changes, but different influences among individual land use types. Changes in cropland, forest and grassland are mainly affected by climate change, topography and soil conditions, while changes in built-up areas are more influenced by the terrain and economic factors (Ye et al., 2013, 2015). The distribution of wetland and unused land are largely negatively correlated with GDP, population, land aspect and slope. Using these major driving factors, the distribution probability of individual land use types can be determined for specific locations. The factors with high correlation coefficients are used by the probability function for the corresponding land use types, while the factors with low significant are left out. Moreover, land use policy also plays an important role in land use changes. During the period of 1980-1990, NEC experienced slow economic progress, and the government was more concerned about food security issues. As a result, land use changes were not frequent, and occurred mainly in the southern part of NEC, where some forest and grassland areas were converted into cropland. The economy of NEC started to grow quickly after 1990. Increasing demand for food caused a large number of scattered forest, grassland and unused land areas been converted into cropland in the middle of NEC between 1990 and 2000. At the same period, the overexploitation of cropland led to ecological problems, i.e., land degradation and deterioration (Bindraban et al., 2012; Ye and Van Ranst, 2009). Consequently, new land use policies, such as returning cropland to forest, were implemented, which caused significant conversion of cropland into forest in NEC. During the period 2000‒2010, rapid economic development was the top item on the local government agenda. The increasing demand for built-up areas caused the disappearance of a large amount of cropland and forest surrounding the cities. In order to achieve a dynamic equilibrium in terms of total cropland, some wetland and forest areas were transferred to cropland in the northern part of NEC. This study included some limited areas of changing such as national nature reserve park to express the policies influence. However, it is hard to fully consider all the influence of policy factors. It should be thus acknowledged that although the selected factors can effectively explain the changes in the distribution of various land use types, some other causes, e.g., policy change, technological development and social preferences, which may also have a great impact on land use changes, were not considered in this study. This suggests some uncertainties in the final simulation results. This study only identified the major factors driving the land use changes, but the relative contribution of each factor should be explored further in the future research.

4.3 The implications of land use changes

This study showed that the NEC experienced an obvious land use changes over the period of 1980-2010. In particular, cropland constantly expanded northward during this period. The loss of cropland due to urbanization and industrialization in the southern part of NEC triggered more cropland conversion in the north, where there is more potentially available cropland. Moreover, global warming has had a positive impact on cropland expansion in the North, particularly in the Sanjiang Plain, where rice cultivation has expanded over the past 30 years at the expense of wetland and grassland (Xia, 2014). Although the new cropland could substantially increase national and regional food security, uncertain impacts on the environment and potentially detrimental system feedback could undermine future food production. For instance, the rapid decrease in wetland areas could damage the whole ecosystem and devastate the overall environmental resources of NEC, as wetlands have multiple roles to play as water resources and in terms of biodiversity protection, animal-bearing capacity and other ecological functions. Effective laws and regulations are required for wetland conservation. Future land use management should take these trade-offs into consideration. The land use systems in general must become fully sustainable, which will require new approaches to address the interwoven challenges of food production and environmental protection (An et al., 2007).

5 Conclusions

This study used the CLUE-S model to analyze the spatio-temporal changes in land use in NEC over the past 30 years. The model implemented both forward (2000-2010) and backward (1980-2000) simulations, taking the year 2000 as the baseline. Biophysical and socioeconomic factors could impact land use changes, but with different influences on individual land use types. Land use policy also played an important role in land use changes in NCE. The changing patterns in land use types differed. Distribution of cropland and forestland in NEC showed a clear X-shaped pattern. That is, cropland spanned the areas from the northeast (the Sanjiang Plain) to the southwest (the Liaohe Plain), while forestland ranged from the northwest (the Lesser Khingan Mountains) to the southeast (the Changbai Mountains). Increases in cropland area accounted for 33,167 km2 in NEC, and were mainly distributed in the Sanjiang Plain and the Songnen Plain during 1980-2000; and new croplands were mostly converted from forestlands and grasslands. After the year 2000, although new cropland was still gained from wetland in the Sanjiang Plain, a net loss of cropland area of 30,321 km2 was characterized. Changes of forestland area showed an opposite pattern to that of cropland. An area of 17,183 km2 of forestland was found lost in the Khingan Mountains and the Changbai Mountains between 1980 and 2000, followed by a small gain of 6249 km2 from neighbouring croplands and grasslands. Findings from this paper improved our understanding over the causes, locations and consequences of land use changes, and provided important support in land use planning and policy making to ensure sustainable management and use of land resources.

The authors have declared that no competing interests exist.

References

Publishing order | Descend order by publishing year | Descend order by cited within
1
An S, Li H, Guan Bet al., 2007. China’s natural wetlands: Past problems, current status, and future challenges. Ambio, 36(4): 335-342.

2
Bakker M, Veldkamp A, 2012. Changing relationships between land use and environmental characteristics and their consequences for spatially explicit land-use change prediction. Journal of Land Use Science, 7(4): 407-424.

3
Bindraban P S, van der Velde M, Ye Let al., 2012. Assessing the impact of soil degradation on food production.Current Opinion in Environmental Sustainability, 4(5): 478-488.Continuing soil degradation remains a serious threat to future food security. Yet, global soil degradation assessments are based on qualitative expert judgments or remotely sensed quantitative proxy values that suffice to raise awareness but are too coarse to identify appropriate sustainable land management interventions. Studies in China and Sub Saharan Africa illustrate the considerable impact of degradation on crop production but also point to the need for solutions dependent on location specific agro-ecological conditions and farming systems.The development of a comprehensive approach should be feasible to better assess both extent and impact of soil degradation interlinking various scales, based on production ecological approaches and remote sensing to allow disentangling natural and human induced causes of degradation. A shared common knowledge base cataloguing hard-won location-specific interventions is needed for successfully preventing or mitigating degradation.

DOI

4
Brown D G, Verburg P H, Pontius Jr R Get al., 2013. Opportunities to improve impact, integration, and evaluation of land change models. Current Opinion in Environmental Sustainability, 5(5): 452-457.Land change modeling supports analyses, assessments, and decisions concerning land management by providing a platform for both encoding mechanisms of land-change processes and making projections of future land-cover and land-use patterns. Approaches have ranged from pattern-based methods, such as machine learning models, to structural or process-based methods, such as economic or agent-based models. Selection of the appropriate modeling approach for a given scientific or decision making purpose is essential. Additionally, we argue that more needs to be done to develop and disseminate methods for evaluating land-change models (LCMs). The profession needs better data to support the use of LCMs, integration of models that operate at various scales, and combinations of models that address both positive and normative aspects of land use and land cover patterns and dynamics

DOI

5
Chen X, Fan R, Shi Xet al., 2013. Spatial variation of penetration resistance and water content as affected by tillage and crop rotation in a black soil in Northeast China. Acta Agriculturae Scandinavica Section B-Soil and Plant Science, 63(8): 740-747.

6
Costanza R., 1989. Model goodness of fit: A multiple resolution procedure. Ecological Modelling, 47(3/4): 199-215.Quantitatively evaluating the goodness of fit of ecological simulation models is difficult, and no generally agreed upon method has evolved. This paper presents a method for quantifying the goodness of fit of spatial and/or time series data and models based on measuring the similarity of the patterns, and the idea that measurement at one resolution is not sufficient to describe complex patterns. The method yields indices that summarize the way the fit changes as the resolution of measurement changes. An expanding ‘window’ is used to gradually degrade the resolution of the comparison. Lack of fit can be partitioned into ‘registration’, ‘resolution’ and residual components. This allows a better understanding of the underlying patterns and type of correspondence. Multiple resolution methods yield additional information not contained in single resolution methods that is necessary to adequately evaluate the performance of complex ecological models.

DOI

7
Foley J A, 2005. Global consequences of land use. Science, 309(5734): 570-574.Land use has generally been considered a local environmental issue, but it is becoming a force of global importance. Worldwide changes to forests, farmlands, waterways, and air are being driven by the need to provide food, fiber, water, and shelter to more than six billion people. Global croplands, pastures, plantations, and urban areas have expanded in recent decades, accompanied by large increases in energy, water, and fertilizer consumption, along with considerable losses of biodiversity. Such changes in land use have enabled humans to appropriate an increasing share of the planet's resources, but they also potentially undermine the capacity of ecosystems to sustain food production, maintain freshwater and forest resources, regulate climate and air quality, and ameliorate infectious diseases. We face the challenge of managing trade-offs between immediate human needs and maintaining the capacity of the biosphere to provide goods and services in the long term.

DOI PMID

8
Gao J, Liu Y, 2011. Climate warming and land use change in Heilongjiang Province, Northeast China. Applied Geography, 31(2): 476-482.This study explores the relationship between climate warming and rice paddy expansion in Heilongjiang Province of China. It is found that paddy fields more than quadrupled from 3479 km(2) in 1958 to 14 564 km(2) in 1980, and increased further to 21,940 km(2) in 2000. The newly gained paddy fields originated chiefly from dry fields (46.35%), swamps (30.22%), and primary forest (nearly 10%) during 1958-1980. During 1980-2000 paddy fields expanded at the expense of dry fields (70.50%), swamp (16.59%), and grassland (10.13%). Analysis of climate data shows a warming of over 2 degrees C from the 1960s to the 2000s in most places. All 28 meteorological stations except one experienced a warming trend. Spatially, the expansion of paddy fields coincided closely with the spatial distribution of annual temperature. These fields were located mostly between the isolines of 2-3 degrees C. Sowing area of grain increased at a modest rate during the 1970s and the 1980s when >0 degrees C area expanded rapidly. However, sowing area of rice rose in the 1990s and 2000s at a rate twice higher than that for sowing area of grain in the preceding decades. Thus, the expansion of paddy fields at the expense of other land covers was made possible owing to climate warming in the preceding decade. On average, it takes about 20 years for agricultural practices to adapt to the warmer climate. (C) 2010 Elsevier Ltd. All rights reserved.

DOI

9
Geoghegan J, Villar S C, Klepeis Pet al., 2001. Modeling tropical deforestation in the southern Yucatán peninsular region: Comparing survey and satellite data. Agriculture, Ecosystems & Environment, 85(1-3): 25-46.

10
Gong P, Wang J, Yu Let al., 2013. Finer resolution observation and monitoring of global land cover: First mapping results with Landsat TM and ETM+ data. International Journal of Remote Sensing, 34(7): 2607-2654.

11
Gutman G, Huang C, Chander Get al., 2013. Assessment of the NASA-USGS global land survey (GLS) datasets.Remote Sensing of Environment, 134: 249-265.The Global Land Survey (GLS) datasets are a collection of orthorectified, cloud-minimized Landsat-type satellite images, providing near complete coverage of the global land area decadally since the early 1970s. The global mosaics are centered on 1975, 1990, 2000, 2005, and 2010, and consist of data acquired from four sensors: Enhanced Thematic Mapper Plus, Thematic Mapper, Multispectral Scanner, and Advanced Land Imager. The GLS datasets have been widely used in land-cover and land-use change studies at local, regional, and global scales. This study evaluates the GLS datasets with respect to their spatial coverage, temporal consistency, geodetic accuracy, radiometric calibration consistency, image completeness, extent of cloud contamination, and residual gaps. In general, the three latest GLS datasets are of a better quality than the GLS-1990 and GLS-1975 datasets, with most of the imagery (85%) having cloud cover of less than 10%, the acquisition years clustered much more tightly around their target years, better co-registration relative to GLS-2000, and better radiometric absolute calibration. Probably, the most significant impediment to scientific use of the datasets is the variability of image phenology (i.e., acquisition day of year). This paper provides end-users with an assessment of the quality of the GLS datasets for specific applications, and where possible, suggestions for mitigating their deficiencies. (C) 2013 Elsevier Inc. All rights reserved.

DOI

12
Lambin E.F., Geist H., 2006. Land-use and Land-cover Change. Berlin Heidelberg: Springer.

13
Lambin E F, Turner B L, Geist H Jet al., 2001. The causes of land-use and land-cover change: Moving beyond the myths. Global Environmental Change, 11(4): 261-269.Common understanding of the causes of land-use and land-cover change is dominated by simplifications which, in turn, underlie many environment-development policies. This article tracks some of the major myths on driving forces of land-cover change and proposes alternative pathways of change that are better supported by case study evidence. Cases reviewed support the conclusion that neither population nor poverty alone constitute the sole and major underlying causes of land-cover change worldwide. Rather, peoples responses to economic opportunities, as mediated by institutional factors, drive land-cover changes. Opportunities and constraints for new land uses are created by local as well as national markets and policies. Global forces become the main determinants of land-use change, as they amplify or attenuate local factors.

DOI

14
Letourneau A, Verburg P H, Stehfest E, 2012. A land-use systems approach to represent land-use dynamics at continental and global scales. Environmental Modelling & Software, 33: 61-79.Most of the current global land cover datasets and global scale land-use models use a classification of land cover based on the dominant land cover type within a distinct region or pixel. Such a classification disregards the diversity and intensity of human influence on land systems. In this paper we propose a novel way of classification and modeling land-use using a classification based on land-use systems (LUSs) that represent specific combinations of human-environment interactions. A cluster analysis was used to identify and map these LUSs. The analysis accounted for population density, accessibility to market places, land-use/cover types and livestock densities. A conceptual framework was developed to model dynamics in LUSs accounting for both land cover and land management changes. LUSs changes were simulated based on changes in both local socio-economic and biophysical conditions and regional-scale changes in demand for agricultural products. The new land-use systems change model was used in the context of the integrated assessment model IMAGE.

DOI

15
Li S, Verburg P H, Lv Set al., 2012. Spatial analysis of the driving factors of grassland degradation under conditions of climate change and intensive use in Inner Mongolia, China.Regional Environmental Change, 12(3): 461-474.Abstract<br/>In recent years, steppe degradation in North China has become a serious environmental problem. Most research on steppe degradation is conducted at the level of communities or at the scale of small regions. To better understand the spatio-temporal variation and driving factors of grassland degradation, monitoring and analysis at broad regional scales are needed. This paper systematically describes the state and characteristics of steppe degradation at the Xilinhot plateau, makes an in-depth empirical analysis of the natural and man-made causes leading to degradation, and analyzes what driving factors have influenced degradation in this typical steppe region over the last 20 years. Ten biophysical and socio-economic variables, including altitude, slope, precipitation, temperature, soil conditions, distance to river, distance to highway, population density, sheep unit density, and fencing policy, were evaluated on their impact on observed patterns of degradation. The results indicate that all of these factors had a significant influence on the process of steppe degradation. During the first 10 years, from 1991 to 2000, steppe degradation increased, but after 2000, the degradation trend has, to some extent, reversed. The analysis indicates that the measures taken by the government, such as fencing vulnerable areas, played an important role in this change. The results advance the understanding of grassland degradation and contribute to constructing an empirical and theoretical base for grassland management and planning.<br/>

DOI

16
Li Z, Tang H, Yang Pet al., 2012. Spatio-temporal responses of cropland phenophases to climate change in Northeast China. Journal of Geographical Sciences, 22(1): 29-45.Abstract<br/><p class="a-plus-plus">We investigated the responses of cropland phenophases to changes of agricultural thermal conditions in Northeast China using the SPOT-VGT Normalized Difference Vegetation Index (NDVI) ten-day-composed time-series data, observed crop phenophases and the climate data collected from 1990 to 2010. First, the phenological parameters, such as the dates of onset-of-growth, peak-of-growth and end-of-growth as well as the length of the growing season, were extracted from the smoothed NVDI time-series dataset and showed an obvious correlation with the observed crop phenophases, including the stages of seedling, heading, maturity and the length of the growth period. Secondly, the spatio-temporal trends of the major thermal conditions (the first date of ⩾ 10°C, the first frost date, the length of the temperature-allowing growth period and the accumulated temperature (AT) of ⩾ 10°C) in Northeast China were illustrated and analyzed over the past 20 years. Thirdly, we focused on the responses of cropland phenophases to the thermal conditions changes. The results showed that the onset-of-growth date had an obvious positive correlation with the first date of ⩾ 10°C (P &lt; 0.01), especially in the northern part of the Songnen Plain, the eastern part of the Sanjiang Plain and the middle and eastern parts of Jilin Province. For the extracted length of growing season and the observed growth period, notable correlations were found in almost same regions (P &lt; 0.05). However, there was no obvious correlation between the end-of-growth date and the first frost date in the study area. Opposite correlations were observed between the length of the growing season and the AT of ⩾ 10°C. In the northern part of the Songnen Plain, the eastern part of the Sanjiang Plain and the middle part of Jilin and Liaoning Provinces, the positive correlation coefficients were higher than the critical value of 0.05, whereas the negative correlation coefficients reached a level of 0.55 (P &lt; 0.05) in the middle and southern parts of Heilongjiang Province and some parts of the Sanjiang Plain. This finding indicated that the crop growth periods were shortened because of the elevated temperature; in contrast, the extended growth period usually meant a crop transformation from early- or middle-maturing varieties into middle or late ones.</p><br/>

DOI

17
Liu J, Liu M, Tian Het al., 2005. Spatial and temporal patterns of China’s cropland during 1990-2000: An analysis based on landsat tm data. Remote Sensing of Environment, 98(4): 442-456.

18
Liu J, Liu M, Zhuang Det al., 2003. Study on spatial pattern of land-use change in China during 1995-2000. Science in China Series D:Earth Sciences, 46(4): 373-384.It is more and more acknowledged that land-use/cover dynamic change has become a key subject urgently to be dealt with in the study of global environmental change. Supported by the Landsat TM digital images, spatial patterns and temporal variation of land-use change during 1995 -2000 are studied in the paper. According to the land-use dynamic degree model, supported by the 1km GRID data of land-use change and the comprehensive characters of physical, economic and social features, a dynamic regionalization of land-use change is designed to disclose the spatial pattern of land-use change processes. Generally speaking, in the traditional agricultural zones, e.g., Huang-Huai-Hai Plains, Yangtze River Delta and Sichuan Basin, the built-up and residential areas occupy a great proportion of arable land, and in the interlock area of farming and pasturing of northern China and the oases agricultural zones, the reclamation I of arable land is conspicuously driven by changes of production conditions, economic benefits and climatic conditions. The implementation of "returning arable land into woodland or grassland" policies has won initial success in some areas, but it is too early to say that the trend of deforestation has been effectively reversed across China. In this paper, the division of dynamic regionalization of land-use change is designed, for the sake of revealing the temporal and spatial features of land-use change and laying the foundation for the study of regional scale land-use changes. Moreover, an integrated study, including studies of spatial pattern and temporal process of land-use change, is carried out in this paper, which is an interesting try on the comparative studies of spatial pattern on change process and the change process of spatial pattern of land-use change.

DOI

19
Liu Y, Wang D, Gao Jet al., 2005. Land use/cover changes, the environment and water resources in Northeast China. Environmental Management, 36(5): 691-701.<a name="Abs1"></a>Land use/cover in Northeast China went through extensive changes during the 1990s. This report explores the interaction between these changes and the environment, and the implication of these changes for rational allocation of water resources. Two maps of land use/cover produced from 1990 and 2000 Landsat TM satellite images were overlaid in ArcInfo to reveal changes in land cover. Results indicate that farmland and grassland decreased by 386,195 and 140,075 ha, respectively, while water, built-up areas, and woodland increased by 238,596, 194,231, and 192,682 ha, respectively. These changes bore a mutual relationship with the environmental change. On the one hand, climate warming made some of these changes (e.g., conversion of woodland and grassland to farmland) possible. On the other hand, the changed surface cover modified the local climate. These changes, in turn, caused severe environmental degradation and increased flooding. The change between dry field and rice paddy, in particular, raised severe implications for the proper allocation of limited water resources in the Northeast. Efforts are needed to coordinate their rational allocation to reap maximum and sustainable return over the entire area, not just in some localities. Results obtained in this study should be of interest to the international audience of <i>Environmental Management</i> in that they highlight the interactive nature of human activities and the environment and the off-site impact of these activities on the environment.

DOI

20
Pontius Jr R G, Cornell J D, Hall C A S, 2001. Modeling the spatial pattern of land-use change with geomod2: Application and validation for Costa Rica. Agriculture, Ecosystems & Environment, 85(1-3): 191-203.The objective of this paper is to simulate the location of land-use change, specifically forest disturbance, in Costa Rica over several decades. This paper presents a GIS-based model, GEOMOD2, which quantifies factors associated with land-use, and simulates the spatial pattern of land-use forward and backward in time. GEOMOD2 reads rasterized maps of land-use and other biogeophysical attributes to determine empirically the attributes of land that tend to use. Then GEOMOD2 uses the patterns of those biogeophysical attributes to simulate the spatial pattern of land-use change. GEOMOD2 can select locations for land-use change according to any of three decision rules based on (1) nearest neighbors, (2) stratification by political sub-region, and/or (3) the pattern of biogeophysical attributes. GEOMOD2 simulates the progressive loss of closed-canopy forest in Costa Rica for 1940, 1961 and 1983, which are the years for which maps of land-use are available. Also, GEOMOD2 extrapolates the pattern of land-use to the year 2010. When GEOMOD2 extrapolates land-use change over several decades, it is able to classify correctly between 74 and 88% of the grid , for two categories: forest versus non-forest. Over various simulation runs, Kappa ranges from 0.31 to 0.53. The model's ability to predict the location of disturbance is best when the model is driven by the location of biogeophysical characteristics, most importantly lifezones.

DOI

21
Pontius Jr R G, Schneider L C, 2001. Land-cover change model validation by an ROC method for the Ipswich watershed, Massachusetts, USA. Agriculture, Ecosystems & Environment, 85(1-3): 239-248.Scientists need a better and larger set of tools to validate land-use change models, because it is essential to know a model's prediction accuracy. This paper describes how to use the relative operating characteristic (ROC) as a quantitative measurement to validate a land-cover change model. Typically, a crucial component of a spatially explicit simulation model of land-cover change is a map of suitability for land-cover change, for example a map of probability of deforestation. The model usually selects locations for new land-cover change at locations that have relatively high suitability. The ROC can compare a map of actual change to maps of modeled suitability for land-cover change. ROC is a summary statistic derived from several two-by-two contingency tables, where each contingency table corresponds to a different simulated scenario of future land-cover change. The categories in each contingency table are actual change and actual non-change versus simulated change and simulated non-change. This paper applies the theoretical concepts to a model of deforestation in the Ipswich watershed, USA.

DOI

22
Pontius R G, Huffaker D, Denman K, 2004. Useful techniques of validation for spatially explicit land-change models. Ecological Modelling, 179(4): 445-461.For illustration, these techniques are applied to assess an LUCC model called Geomod, which predicts land change in the 22 towns of the Ipswich River Watershed in northeastern Massachusetts, USA. For this application, the Null Resolution is approximately 102km. At resolutions finer than 102km, the Null model performs better than Geomod, which performs better than the Random model. At resolutions coarser than 102km, both Geomod and the Random models perform better than the Null model, but Geomod and the Random models are nearly indistinguishable beyond the 102km resolution.

DOI

23
Rounsevell M, Annetts J E, Audsley Eet al., 2003. Modelling the spatial distribution of agricultural land use at the regional scale. Agriculture, Ecosystems & Environment, 95(2): 465-479.Agriculture is the most important land use in Europe in geographic terms and because of this it plays a central role in the quality of the wider environment. Whilst the spatial patterns of agricultural land use have changed considerably in recent times, further changes are likely as a result of the influences of policy reform, socio-economics and climate change. Understanding, therefore, how agricultural land use might respond to global environmental change drivers is a research question of considerable importance. The first step, however, in projecting potential future changes in agricultural land use is to be able to understand and represent in models both the socio-economic and physical processes that control current land use distributions.Thus, this paper presents an approach to modelling the spatial distribution of agricultural land use at the regional scale. The approach is based on the simulation of farm-scale decision making processes (based on optimisation) and the response of crops to their physical environment. Regional scale applications of the model are undertaken through the use of spatially-variable, geographic data sets (soils, climate and topography) combined with economic data. Examples of the application of the model are given for two regions of England: the north-west and east Anglia. These regions were selected to give examples of contrasting land use systems within the context of northern European agriculture. The model results are compared statistically with observed distributions of agricultural land use for the same regions in a quasi-validation exercise. The comparison suggests that the model is very good at representing land use that is aggregated at the regional level, and at representing general spatial trends in land use patterns. Some differences were observed, however, in land use densities between the modelled and observed data.The results suggest that the basic hypothesis of the model: that farmers are risk averse, profit maximisers, is a reasonable assumption for the regions studied. However, further study of decision making processes would be likely to improve our ability to model agricultural land use distributions. This includes, for example, the role of farmer attitudes to risk, differing views on future prices and profitability, and the effect of time lags in the decision process.

DOI

24
Rounsevell M D A, Reginster I, Araújo M Bet al., 2006. A coherent set of future land use change scenarios for europe. Agriculture, Ecosystems & Environment, 114(1): 57-68.

25
Schaldach R U D, Alcamo J, Koch Jet al., 2011. An integrated approach to modelling land-use change on continental and global scales.Environmental Modelling & Software, 26(8): 1041-1051.Land-use and land-cover change are important drivers of global environmental change, affecting the state of biodiversity, the global carbon cycle, and other aspects of the earth system. In this article we describe the development of the land-use model LandSHIFT, which aims to simulate land-use and land-cover change on the continental and global scale. The model is based on a “land-use systems” approach, which describes the interplay between anthropogenic and environmental system components as drivers of land-use change. LandSHIFT’s modular structure facilitates the integration of different components that cover key parts of land-use systems. The model prototype combines a module for the simulation of land-use change dynamics with a module for calculating crop yields and net primary productivity of grassland. LandSHIFT is driven by country-level model inputs including time-series of socio-economic variables as well as agricultural production data. This information is regionalized to land-use grid maps with a cell size of 5 arc-minutes. Here, the model clearly differentiates between the land-use activities settlement, crop cultivation and grazing. By using standardized input–output formats, LandSHIFT can be combined with other models for conducting complex simulation studies.

DOI

26
Serneels S, Lambin E F, 2001. Proximate causes of land-use change in Narok District, Kenya: A spatial statistical model. Agriculture, Ecosystems & Environment, 85(1-3): 65-81.This study attempts to identify how much understanding of the driving forces of land-use changes can be gained through a spatial, statistical analysis. Hereto, spatial, statistical models of the proximate causes of different processes of land-use change in the Mara Ecosystem (Kenya) were developed, taking into account the spatial variability of the land-use change processes. The descriptive spatial models developed here suggest some important factors driving the land-use changes that can be related to some well-established theoretical frameworks. The explanatory variables of the spatial model of mechanised agriculture suggest a von Thunen-like model, where conversion to agriculture is controlled by the distance to the market, as a proxy for transportation costs, and agro-climatic potential. Expansion of smallholder agriculture and settlements is also controlled by land rent, defined, in this case, by proximity to permanent water, land suitability, location near a tourism market, and vicinity to villages to gain access to social services (e.g. health clinics, schools, local markets). This difference in perception of land rent reflects the widely different social and economic activities and objectives of smallholders versus the large entrepreneurs involved in mechanised farming. Spatial heterogeneity as well as the variability in time of land-use change processes affect our ability to use regression models for wide ranging extrapolations. The models allow evaluating the impact of changes in driving forces that are well represented by proximate causes of land-use change.

DOI

27
Sleeter B M, Sohl T L, Loveland T Ret al., 2013. Land-cover change in the conterminous United States from 1973 to 2000. Global Environmental Change, 23(4): 733-748.Land-cover change in the conterminous United States was quantified by interpreting change from satellite imagery for a sample stratified by 84 ecoregions. Gross and net changes between 11 land-cover classes were estimated for 5 dates of Landsat imagery (1973, 1980, 1986, 1992, and 2000). An estimated 673,00002km 2 (8.6%) of the United States’ land area experienced a change in land cover at least one time during the study period. Forest cover experienced the largest net decline of any class with 97,00002km 2 lost between 1973 and 2000. The large decline in forest cover was prominent in the two regions with the highest percent of overall change, the Marine West Coast Forests (24.5% of the region experienced a change in at least one time period) and the Eastern Temperate Forests (11.4% of the region with at least one change). Agriculture declined by approximately 90,00002km 2 with the largest annual net loss of 12,00002km 2 02yr 611 occurring between 1986 and 1992. Developed area increased by 33% and with the rate of conversion to developed accelerating rate over time. The time interval with the highest annual rate of change of 47,00002km 2 02yr 611 (0.6% per year) was 1986–1992. This national synthesis documents a spatially and temporally dynamic era of land change between 1973 and 2000. These results quantify land change based on a nationally consistent monitoring protocol and contribute fundamental estimates critical to developing understanding of the causes and consequences of land change in the conterminous United States.

DOI

28
Turner II B L, Janetos A C, Verburg P Het al., 2013. Land system architecture: using land systems to adapt and mitigate global environmental change. Global Environmental Change, 23(2): 395-397.Land systems (mosaics of land use and cover) are human environment systems, the changes in which drive and respond to local to global environmental changes, climate to macro-economy (Foley et al., 2005). Changes in land systems have been the principal proximate cause in the loss of habitats and biota globally, long contributed to atmospheric greenhouse gases, and hypothesized to have triggered climate changes in the early Holocene (Ruddiman, 2003). Land use, foremost agriculture, is the largest source of biologically active nitrogen to the atmosphere, critical to sources and sinks of carbon, and a major component in the hydrologic cycle (e.g., Bouwman et al., 2011). Changes in land systems also affect regional climate (Feddema et al., 2005; Pielke, 2005), ecosystem functions, and the array of ecosystem services they provide. Land systems, therefore, are a central feature of how humankind manages its relationship with nature-intended or not, or whether this relationship proceeds sustainably or not.

DOI

29
Václavík T., Lautenbach S., Kuemmerle T.et al., 2013. Mapping global land system archetypes.Global Environmental Change, 23(6): 1637-1647.Land use is a key driver of global environmental change. Unless major shifts in consumptive behaviours occur, land-based production will have to increase drastically to meet future demands for food and other commodities. One approach to better understand the drivers and impacts of agricultural intensification is the identification of global, archetypical patterns of land systems. Current approaches focus on broad-scale representations of dominant land cover with limited consideration of land-use intensity. In this study, we derived a new global representation of land systems based on more than 30 high-resolution datasets on land-use intensity, environmental conditions and socioeconomic indicators. Using a self-organizing map algorithm, we identified and mapped twelve archetypes of land systems for the year 2005. Our analysis reveals similarities in land systems across the globe but the diverse pattern at sub-national scales implies that there are no 'one-size-fits-all' solutions to sustainable land management. Our results help to identify generic patterns of land pressures and environmental threats and provide means to target regionalized strategies to cope with the challenges of global change. Mapping global archetypes of land systems represents a first step towards better understanding the global patterns of human-environment interactions and the environmental and social outcomes of land system dynamics.

DOI

30
van Asselen S., Verburg P.H., 2013. Land cover change or land-use intensification: Simulating land system change with a global-scale land change model. Global Change Biology, 19(12): 3648-3667.Land-use change is both a cause and consequence of many biophysical and socioeconomic changes. The CLUMondo model provides an innovative approach for global land-use change modeling to support integrated assessments. Demands for goods and services are, in the model, supplied by a variety of land systems that are characterized by their land cover mosaic, the agricultural management intensity, and livestock. Land system changes are simulated by the model, driven by regional demand for goods and influenced by local factors that either constrain or promote land system conversion. A characteristic of the new model is the endogenous simulation of intensification of agricultural management versus expansion of arable land, and urban versus rural settlements expansion based on land availability in the neighborhood of the location. Model results for the OECD Environmental Outlook scenario show that allocation of increased agricultural production by either management intensification or area expansion varies both among and within world regions, providing useful insight into the land sparing versus land sharing debate. The land system approach allows the inclusion of different types of demand for goods and services from the land system as a driving factor of land system change. Simulation results are compared to observed changes over the 1970-2000 period and projections of other global and regional land change models.

DOI

31
van Vliet J, Bregt A K, Hagen-Zanker A, 2011. Revisiting kappa to account for change in the accuracy assessment of land-use change models.Ecological Modelling, 222(8): 1367-1375.Land-use change models are typically calibrated to reproduce known historic changes. Calibration results can then be assessed by comparing two datasets: the simulated land-use map and the actual land-use map at the same time. A common method for this is the Kappa statistic, which expresses the agreement between two categorical datasets corrected for the expected agreement. This expected agreement is based on a stochastic model of random allocation given the distribution of class sizes. However, when a model starts from an initial land-use map and makes changes to it, that stochastic model does not pose a meaningful reference level. This paper introduces K-Simulation, a statistic that is identical in form to the Kappa statistic but instead applies a more appropriate stochastic model of random allocation of class transitions relative to the initial map. The new method is illustrated on a simple example and then the results of the Kappa statistic and K-Simulation are compared using the results of a land-use model. It is found that only K-Simulation truly tests models in their capacity to explain land-use changes over time, and unlike Kappa it does not inflate results for simulations where little change takes place over time. (C) 2011 Elsevier BM. All rights reserved.

DOI

32
Veldkamp A, Fresco L O, 1996. Clue: A conceptual model to study the conversion of land use and its effects. Ecological Modelling, 85(2/3): 253-270.A dynamic model to simulate Conversion of Land Use and its Effects (CLUE) is presented. For an imaginary region, CLUE simulates land use conversion and change in space and time as a result of interacting biophysical and human drivers. Within CLUE regional land use changes only if biophysical and human demands cannot be met by existing land use. After a regional assessment of land use needs, the final land use decisions are made on a local grid level. Important biophysical drivers are local biophysical suitability and their fluctuations, land use history, spatial distribution of infrastructure and land use, and the occurrence of pests and diseases. Important human land use drivers in CLUE are population size and density, regional and international technology level, level of affluence, target markets for products, economical conditions, attitudes and values, and the applied land use strategy. Initial CLUE simulations suggest that the integrated land use approach of CLUE can make a more realistic contribution to predictions of future land cover than currently used biophysical equilibrium approaches.

DOI

33
Verburg P H, Erb K, Mertz Oet al., 2013. Land system science: Between global challenges and local realities. Current Opinion in Environmental Sustainability, 5(5): 433-437.

DOI

34
Verburg P H, Neumann K, Nol L, 2011. Challenges in using land use and land cover data for global change studies. Global Change Biology, 17(2): 974-989.Land use and land cover data play a central role in climate change assessments. These data originate from different sources and inventory techniques. Each source of land use/cover data has its own domain of applicability and quality standards. Often data are selected without explicitly considering the suitability of the data for the specific application, the bias originating from data inventory and aggregation, and the effects of the uncertainty in the data on the results of the assessment. Uncertainties due to data selection and handling can be in the same order of magnitude as uncertainties related to the representation of the processes under investigation. While acknowledging the differences in data sources and the causes of inconsistencies, several methods have been developed to optimally extract information from the data and document the uncertainties. These methods include data integration, improved validation techniques and harmonization of classification systems. Based on the data needs of global change studies and the data availability, recommendations are formulated aimed at optimal use of current data and focused efforts for additional data collection. These include: improved documentation using classification systems for land use/cover data; careful selection of data given the specific application and the use of appropriate scaling and aggregation methods. In addition, the data availability may be improved by the combination of different data sources to optimize information content while collection of additional data must focus on validation of available data sets and improved coverage of regions and land cover types with a high level of uncertainty. Specific attention in data collection should be given to the representation of land management (systems) and mosaic landscapes.

DOI

35
Verburg P H, Overmars K P, Huigen M Get al., 2006. Analysis of the effects of land use change on protected areas in the philippines. Applied Geography, 26(2): 153-173.<h2 class="secHeading" id="section_abstract">Abstract</h2><p id="">Deforestation and forest degradation are the most important land use change processes in the Philippines. These processes are an important threat to the highly rated biodiversity of the country. Only a small fraction of the natural forest that once covered the country remains. In spite of different policies that aim to reduce logging recent commercial deforestation, illegal logging and agricultural expansion pose an important threat to the remaining forest areas.</p><p id="">In this paper we discuss the role of (land use) modeling approaches for assessing the threats and trade-offs of protecting the designated nature areas. At the national level different scenarios of land use change and implementation of the protected area policy are evaluated and discussed based on a spatially explicit land use allocation model. For one of the main national parks, the Northern Sierra Madre Nature Park, a detailed analysis is presented based on in-depth knowledge of the region. The two modeling approaches discussed in this paper aim at different scales and provide complementary types of information to support the planning and management of nature conservation strategies. The combination of land use change analysis at different scales respects the hierarchical organization of the land use system and addresses different levels of protected area management. The results indicate that land use change models are useful tools to inform protected area management as long as the selection of the model approach is based on the research and policy questions at the appropriate scale.</p>

DOI

36
Verburg P H, Schot P P, Dijst M Jet al., 2004. Land use change modelling: Current practice and research priorities. Geojournal, 61(4): 309-324.Land use change models are tools to support the analysis of the causes and consequences of land use dynamics. Scenario analysis with land use models can support land use planning and policy. Numerous land use models are available, developed from different disciplinary backgrounds. This paper reviews current models to identify priority issues for future land use change modelling research. This discussion is based on six concepts important to land use modelling: (1) Level of analysis; (2) Cross-scale dynamics; (3) Driving forces; (4) Spatial interaction and neighbourhood effects; (5) Temporal dynamics; and (6) Level of integration. For each of these concepts an overview is given of the variety of methods used to implement these concepts in operational models. It is concluded that a lot of progress has been made in building land use change models. However, in order to incorporate more aspects important to land use modelling it is needed to develop a new generation of land use models that better address the multi-scale characteristics of the land use system, implement new techniques to quantify neighbourhood effects, explicitly deal with temporal dynamics and achieve a higher level of integration between disciplinary approaches and between models studying urban and rural land use changes. If these requirements are fulfilled models will better support the analysis of land use dynamics and land use policy formulation.

DOI

37
Verburg P H, Schulp C, Witte Net al., 2006. Downscaling of land use change scenarios to assess the dynamics of European landscapes. Agriculture, Ecosystems & Environment, 114(1): 39-56.Europe's rural areas are expected to witness massive and rapid changes in land use due to changes in demography, global trade, technology and enlargement of the European Union. Changes in demand for agricultural products and agrarian structure are likely to have a large impact on landscape quality and the value of natural areas. A spatially explicit, dynamic, land use change model has been used to translate European level scenarios into a high resolution assessment of changes in land use for the 25 countries of the European Union. Scenarios differ in worldview, ranging from enhanced global cooperation towards strong regionalisation on one hand and strong to weak government intervention on the other. Global economic and integrated assessment models were used to calculate changes in demand for agricultural area at country level while a spatially explicit land use change model was used to downscale these demands to land use patterns at 1 km2 resolution. The land use model explicitly accounts for the variation in driving factors among countries and the path dependence in land use change trajectories. Results indicate the large impact abandonment of agricultural land and urbanization has on European landscapes and the different scenarios indicate that spatial policies can make an important contribution to preserve landscapes containing high natural and/or historic values. Furthermore, the dynamic simulations indicate that the trajectory of land use change has an important impact on resulting landscape patterns as a result of the path-dependence in land use change processes. The results are intended to support discussions on the future of the rural area and identify hotspots of landscape change that need specific consideration.

DOI

38
Verburg P H, Soepboer W, Veldkamp Aet al., 2002. Modeling the spatial dynamics of regional land use: The CLUE-S model.Environmental Management, 30(3): 391-405.lt;a name="Abs1"></a><div class="AbstractPara"> <div class="">Land-use change models are important tools for integrated environmental management. Through scenario analysis they can help to identify near-future critical locations in the face of environmental change. A dynamic, spatially explicit, land-use change model is presented for the regional scale: CLUE-S. The model is specifically developed for the analysis of land use in small regions (e.g., a watershed or province) at a fine spatial resolution. The model structure is based on systems theory to allow the integrated analysis of land-use change in relation to socio-economic and biophysical driving factors. The model explicitly addresses the hierarchical organization of land use systems, spatial connectivity between locations and stability. Stability is incorporated by a set of variables that define the relative elasticity of the actual land-use type to conversion. The user can specify these settings based on expert knowledge or survey data. Two applications of the model in the Philippines and Malaysia are used to illustrate the functioning of the model and its validation.

DOI PMID

39
White R, Engelen G, 2000. High-resolution integrated modelling of the spatial dynamics of urban and regional systems. Computers, Environment and Urban Systems, 24(5): 383-400.An emerging branch of geocomputing involves the modelling of spatial processes. A variety of techniques are being used, the most important being traditional regionalized system dynamics approaches, multi-agent systems, and cellular automata (CA). The techniques are frequently combined to model processes operating at different spatial scales. Urban and regional models based on CA give good representations of the spatial dynamics of land use. In a current application, a cellular model of The Netherlands at 500 m resolution is driven by a macro-scale dynamical spatial interaction model defined on 40 economic regions; this model is in turn driven by national planning projections and policy goals. Given the national totals, the macro-scale model generates regional demands for population and a number of economic activities. These demands are translated into demands for cell space, which the CA then attempts to locate. In turn, information on conditions at the cellular level, such as the quantity and quality of land available to various activities and actual densities at the cellular scale, are returned to the regional model to modify parameter values there. Linking the two models operating at the two scales improves the performance of both. The results of high-resolution modelling of spatial dynamics raise several methodological issues. One of the most pressing concerns evaluation of the results. Another issue concerns predictability. To the extent that these models capture the evolving nature of real cities and regions, they cannot be strictly predictive.

DOI

40
Wu W, Shibasaki R, Yang Pet al., 2007. Global-scale modelling of future changes in sown areas of major crops.Ecological Modelling, 208(2-4): 378-390.Land use and its dynamics have attracted much attention from researchers due to their ecological and socio-economic implications. Many studies have used a modelling approach to evaluate land use changes and their effects. Most of these models were designed for the analysis of past, present and future cropland changes at different scales and few have been designed for the study of dynamic changes in sown areas of crops within croplands. This paper presents an integrated modelling approach to simulate dynamically the changes in sown areas for the world's major crops at a global scale. This approach was based on three core models. A crop choice decision model, the Multinomial Logit model, was used to track and simulate the crop choice decisions made by individual farmers. A crop yield model, the GIS-based Environmental Policy Integrated Climate (EPIC) model, was utilized to estimate yields of different crop types under a given biophysical and management environment, while a crop price model, the International Food Policy and Agricultural Simulation (IFPSIM) model, was employed to assess the price of crops on the international market. Through data exchange, the crop choice decision model was linked with the crop yield and crop price models to allow the study of the dynamic feedback loop between changes in agricultural land use and biophysical and socio-economic driving factors. Sensitivity analysis and empirical validation for the model were conducted after the construction of the model. The model validation indicated the reliability of the model for addressing the complexity of current agricultural land use changes and its capacity for investigating long-term scenarios in the future. Finally, the model was used to simulate future scenarios over a time frame of 30 years with five-year increments, beginning from the year 2000. The simulation results provided insights into potential global cropping patterns, variation in rates and trajectory of changes in sown areas for major crops over the test period. These results can improve understanding of projected land use changes and explain their causes, locations and consequences, and provide support for land use planning and policy making.

DOI

41
Wu W, Shibasaki R, Yang Pet al., 2008. Validation and comparison of 1 km global land cover products in China. International Journal of Remote Sensing, 29(13): 3769-3785.

42
Wu W, Verburg P H, Tang H, 2014. Climate change and the food production system: Impacts and adaptation in China.Regional Environmental Change, 14(1): 1-5.There are increasingly more choices from a complex of data resources, classification algorithms, and methods of training sample selections. To increase the repeatability of digital classifications of remotely sensed data with consistently high accuracy, it is essential to use optimal classification options or factors. In this paper, two temporal sets of Landsat thematic mapper (TM) data, three classifiers and three approaches of training sample selections were tested for mapping deforestation. The use of these different factors can have significant effects on classification accuracy. The mixed effects of the three factors can also magnify the variations of classification accuracy. The use of bi-temporal data, a spatial pectral classifier, and hybrid training samples results in steadily higher classification accuracy than the combination of uni-temporal data, a spectral classifier, and image training samples. For the purpose of characterizing managed forest lands, even a small increase in overall accuracy of image classification is important because it may represent a large decrease in the variations of the producer's and user's accuracy, which in turn can reduce the uncertainties of area measurements for forest coverage.

DOI

43
Xia T, Wu W, Zhou Qet al., 2014. Spatio-temporal changes in the rice planting area and their relationship to climate change in Northeast China: A model-based analysis. Journal of Integrative Agriculture, 13(7): 1575-1585.

44
Yao Y, Ye L, Tang Het al., 2015. Cropland soil organic matter content change in Northeast China, 1985-2005.Open Geosciences, 7(1), 234-243.Soil organic matter (SOM) content is one of the most important indicators of soil quality and hence the productive capacity of soils. Northeast China (NEC) is the most important region in grain production in China. In this study,we assessed the spatiotemporal change of cropland SOM content in NEC using sampling data of 2005 and survey data of 1985. We also analysed the driving forces behind the SOM content change. Our results showed that SOM content decreased in 39% of all the cropland in NEC, while increase in SOM content was only detected on 16% of the cropland. SOM remained unchanged in nearly half (i.e. 45%) of the cropland. Our results also revealed that cropping intensity and fertilizer application were the two most important factors driving SOM change. Overall, results from this research provided novel details of the spatiotemporal patterns of cropland SOM content change in NEC which was not revealed in earlier assessments. The datasets presented here can be used not only as baselines for the calibration of process-based carbon budget models, but also to identify regional soil quality hotspots and to guide spatial-explicit soil management practices.

DOI

45
Ye L, Tang H, Yang Get al., 2015. Adopting higher-yielding varieties to ensure Chinese food security under climate change in 2050.Procedia Environmental Sciences, 29: 281.Challenges of ensuring food security under climate change require urgent and substantial increase in the focus of research, innovation, transformation of knowledge, and rapid adoption of available technologies. Here we simulate the effects of the adoption of higher-yielding varieties of rice, wheat and maize crops into the food production systems on China's food security index (FSI, or relative food surplus per capita) in 2050, using the CERES crop models, climate change and a range of socio- economic and agronomic scenarios which were developed following two contrasting development pathways in line with the IPCC A2 and B2 emission scenarios, respectively. The obtained results predict a slightly positive effect of climate change on the FSI, but the magnitude of this positive effect cannot compensate the negative effects of population growth, urbanization rate and the rising affluence on the future trends of the FSI. The outcomes of the adoption of higher-yielding varieties show that a systematic adoption of higher-yielding varieties can raise the average FSI values by a margin of 16 and 27 units under the A2 and B2 scenarios, respectively, during the 2030-2050 period, compared to the average predicted FSI values of -2 and 8 percentage points under A2 and B2 during the same period. This suggests that systematic adoption of higher-yield varieties is an effective measure for Chinese agriculture not only to ensure food security but also to build adaptive capacity to climate change in 2050.

DOI

46
Ye L, Tang H, Zhu Jet al., 2008. Spatial patterns and effects of soil organic carbon on grain productivity assessment in China.Soil Use and Management, 24(1): 80-91.Abstract Top of page Abstract Introduction Materials and methods Results and discussion Soil management options Conclusions References In this paper, we present an assessment of the content and effects of cropland soil organic carbon (SOC) on grain productivity at the national scale in China using a Web-based Land Evaluation System. Homogeneous 5km 5km grid data sets of climate, crop, soil and management parameters were created and grain production in 2005 was simulated. Attempts were made to incorporate SOC into the land evaluation procedure and to quantify the potential effects of SOC deficiency on grain productivity. Results were statistically analysed and the modelling approach was validated. National cropland SOC maps were generated. At the national scale, the cropland SOC content averaged 1.20, 0.58, 0.41, 0.31 and 0.26% for the five 20-cm sections consecutively from the surface downwards. At the regional scale it tended to decline slightly from northeast (1.63%) to southwest (1.11%). On average, 64% of grain yield was lost due to SOC deficiency for the humid provinces and 7% for the arid and sub-arid ones. Soil management options are suggested based on the simulation results.

DOI

47
Ye L, Van Ranst E, 2009. Production scenarios and the effect of soil degradation on long-term food security in China.Global Environmental Change, 19(4), 464-481.Food security in China underlies the foundation of the livelihood and welfare for over one-fifth of the world's population. Soil degradation has an immense negative impact on the productive capacity of soils. We simulated the effect of soil degradation, which occurs in combination with increases in population size, urbanization rate, cropping intensity and decrease in cropland area, on long-term food security in China using a web-based land evaluation system. Our results predict that food crops may experience a 9% loss in productivity by 2030 if the soil continues to be degraded at the current rate (business-as-usual scenario, BAU). Productivity losses will increase to the unbearable level of 30% by 2050 should the soil be degraded at twice the present rate (double-degradation scenario, 2× SD). China's capacity for producing food from agricultural crops will be either adversely affected by the loss of cropland area (130, 113 and 107 million ha in 2005, 2030 and 2050, respectively) or favorably affected by agricultural intensification (in terms of the multi-cropping index at 120, 133 and 147% in 2005, 2030 and 2050, respectively). The loss of cropland is predicted to cause a 13–18% decrease in China's food production capacity by 2030–2050 relative to its 2005 level of 482 Mt, while agricultural intensification is predicted to cause an 11–23% increase. In total, China will be able to achieve a production level of 424 and 412 Mt by 2030 and 2050, respectively, under BAU, while this production will be only 386 and 339 Mt under 2× SD, respectively. In per capita terms, the relationship between food supply and demand will turn from an 18% surplus in 2005 to 3–5%, 14–18% and 22–32% deficits by 2030–2050 under the zero-degradation (0× SD), BAU and 2× SD scenarios, respectively. Our results show that the present-day production capacity will not sustain the long-term needs of a growing population under the current management level. Technical countermeasures and policy interventions need to be enacted today in order to avoid food insecurity tomorrow.

DOI

48
Ye L, Xiong W, Li Zet al., 2013. Climate change impact on China food security in 2050.Agronomy for Sustainable Development, 33(2): 363-374.Abstract<br/><p class="a-plus-plus">Climate change is now affecting global agriculture and food production worldwide. Nonetheless the direct link between climate change and food security at the national scale is poorly understood. Here we simulated the effect of climate change on food security in China using the CERES crop models and the IPCC SRES A2 and B2 scenarios including CO2 fertilization effect. Models took into account population size, urbanization rate, cropland area, cropping intensity and technology development. Our results predict that food crop yield will increase +3–11 % under A2 scenario and +4 % under B2 scenario during 2030–2050, despite disparities among individual crops. As a consequence China will be able to achieve a production of 572 and 615 MT in 2030, then 635 and 646 MT in 2050 under A2 and B2 scenarios, respectively. In 2030 the food security index (FSI) will drop from +24 % in 2009 to −4.5 % and +10.2 % under A2 and B2 scenarios, respectively. In 2050, however, the FSI is predicted to increase to +7.1 % and +20.0 % under A2 and B2 scenarios, respectively, but this increase will be achieved only with the projected decrease of Chinese population. We conclude that 1) the proposed food security index is a simple yet powerful tool for food security analysis; (2) yield growth rate is a much better indicator of food security than yield per se; and (3) climate change only has a moderate positive effect on food security as compared to other factors such as cropland area, population growth, socio-economic pathway and technology development. Relevant policy options and research topics are suggested accordingly.</p><br/>

DOI

49
Yu Q, Wu W, Verburg P Het al., 2013. A survey-based exploration of land-system dynamics in an agricultural region of Northeast China. Agricultural Systems, 121: 106-116.Understanding the complexity of agricultural systems requires insight into the human–environment interactions. In this paper we used survey data to analyze land system change and its relation to farmer’s attitudes in a typical agricultural region of Northeast China, focusing on land tenure, crop choice and intensification. Our survey shows that land transfer was fairly common across the study area: average farmland acreage per household almost doubled from 1.3ha by early 1980s to 2.6ha by early 2010s, especially due to urban migration of farmers. The survey indicates an increase in land transfers over time with a sharp decrease of the average period of land transfer contracts. Crop choice displays a trend of decreasing diversity as several cereal crops are no longer grown in the study region and the majority of bean cultivation has been replaced by maize and tobacco. Land transfers can explain part of these changes, butnot necessarily the full change to a dominance of a smaller number of crops at the region level. Irrigation intensity is related to the locations of rivers, while agricultural inputs, along with land transfer and crop allocation, show a spatial pattern which is related to road accessibility. Survey results show that two family characteristics ( education level and the initially allocated land rights ) and two socioeconomic factors ( infrastructure and crop prices ) are important in making land transfer decisions, while external factors such as market , policy , local cropping system , and agricultural disasters have substantially influenced crop choice decisions. The survey approach is very valuable to analyze land system changes from a stakeholder’s perspective, especially in the absence of statistical data atfarm level.

DOI

50
Yu Q, Wu W, Yang Pet al., 2012. Proposing an interdisciplinary and cross-scale framework for global change and food security researches. Agriculture, Ecosystems & Environment, 156: 57-71.Food security is greatly affected by the consequences of global change, especially its impact on agriculture. Currently, global change and food system interaction is a hot issue across the scientific community. Scientists have tried to explain this interaction from different perspectives, and the issues related to this interaction can be classified as (1) crop yield and productivity in response to global change; (2) crop distribution and allocation in relation with global change; (3) general impacts on food security. However, most of the existing studies lack consistency and continuity. As food systems exist at the intersection of the coupled and natural system, the interdisciplinary context of global change and food security requires an integrated and collaborative framework for better describing their importance and complexity. To do so, we decompose global change/food security studies into different levels in accordance with the previous mentioned issues, field, regional, and global, and categorize them into the life sciences, earth and environmental sciences, and social and sustainability sciences, respectively (yet not necessarily one to one correspondence). At the field level, long-term observations and controlled experiments in situ are important for exploring the mechanism of how global change will affect crop , and for considering possible adaptation methods that may maximize crop productivity. At the regional level, priority should be given to monitoring and simulating crop production (animal production and fishery are not included here) within large areas (a region or a continent). At the global level, food security studies should be based on scenario assessments to prioritize adaptations under the changed environment, using integrated socioeconomic iogeophysical measures.

DOI

51
Zhang S, Zhang X, Huffman Tet al., 2011. Influence of topography and land management on soil nutrients variability in Northeast China.Nutrient Cycling in Agroecosystems, 89(3): 427-438.It is well recognized that soil nutrient content varies across the landscape, but the nature and degree of that variability with respect to landscape position is still poorly understood and documented. Slope steepness and aspect, climate and land management are known to affect soil nutrient distribution in a field, but the relative and cumulative strengths of these effects are less well investigated. Four hundred and thirty-five topsoil samples collected from a typical Mollisol under intensive crop management in Northeast China were used to analyze the influence of landscape position, climate and land management on the spatial variability of soil organic matter (SOM), total nitrogen (TN) and total phosphorus (TP). Both geo-statistics and traditional statistics were used to analyze the data, and significant spatial variability was found for SOM (22.5&#8211;86.6&nbsp;g&nbsp;kg<sup>&#8722;1</sup>), TN (0.98&#8211;4.26&nbsp;g&nbsp;kg<sup>&#8722;1</sup>) and TP (0.26&#8211;1.80&nbsp;g&nbsp;kg<sup>&#8722;1</sup>). The distribution of all 3 nutrients was found to be influenced by human activity and by landscape. When both slope degree and slope aspect were considered, the results differed from when only aspect or steepness was considered independently. In a northern aspect, SOM and TN were significantly higher on slopes of 0&#8211;2% than on steeper slopes, in a south-eastern aspect they were significantly higher on slopes of 0&#8211;2, 2&#8211;3 and 3&#8211;4% than on slopes &gt;4% and in a south-western aspect those nutrients on slopes of 2&#8211;4% were significantly higher than on slopes of &gt;5%. Cross-slope tillage effectively increased SOM, TN and TP by 33.8, 23.3 and 22.4%, respectively compared to down-slope tillage, indicating the potential for adoption of a nutrient-retaining management practice in the Mollisol region of northeast China.

DOI

Options
Outlines

/

〈 〉