EN中文

Journal of Geographical Sciences >

2024 , Vol. 34 >Issue 2: 355 - 374

DOI: https://doi.org/10.1007/s11442-024-2208-0

Research Articles

Quantitative response of vegetation phenology to temperature and precipitation changes in Eastern Siberia

  • WEN Kege , 1, 2 ,
  • LI Cheng , 1, 2, 3, * ,
  • HE Jianfeng 1 ,
  • ZHUANG Dafang 1
Expand
  • 1. State Key Laboratory of Resources and Environment Information System, Institute of Geographic Sciences and Natural Resources Research, CAS, Beijing 100101, China
  • 2. University of Chinese Academy of Sciences, Beijing 100049, China
  • 3. Nanjing Agricultural University, Nanjing 210095, China
*Li Cheng (1991-), PhD, specialized in applications of remote sensing of resources and environment and geographic information system. E-mail:

We Kege (1987-), PhD, specialized in applications of remote sensing of resources and environment and geographic information system. E-mail:

Received date: 2023-06-13

  Accepted date: 2023-10-17

  Online published: 2024-02-06

Supported by

International Cooperation and Exchange of the National Natural Science Foundation of China(42061134019)

Major Special Project-The China High-Resolution Earth Observation System(30-Y30F06-9003-20/22)

Fold

Abstract

Significant changes to the world’s climate over the past few decades have had an impact on the development of plants. Vegetation in high latitude regions, where the ecosystems are fragile, is susceptible to climate change. It is possible to better understand vegetation’s phenological response to climate change by examining these areas. Traditional studies have mainly investigated how a single meteorological factor affects changes in vegetation phenology through linear correlation analysis, which is insufficient for quantitatively revealing the effects of various climate factor interactions on changes in vegetation phenology. We used the asymmetric Gaussian method to fit the normalized difference vegetation index (NDVI) curve and then used the dynamic threshold method to extract the phenological parameters, including the start of the season (SOS), end of the season (EOS), and length of the season (LOS), of the vegetation in this study area in the Tundra-Tagar transitional zone in eastern and western Siberia from 2000 to 2017. The monthly temperature and precipitation data used in this study were obtained from the climate research unit (CRU) meteorological dataset. The degrees to which the changes in temperature and precipitation in the various months and their interactions affected the changes in the three phenological parameters were determined using the GeoDetector, and the results were explicable. The findings demonstrate that the EOS was more susceptible to climate change than the SOS. The vegetation phenology shift was best explained by the climate in March, April, and September, and the combined effect of the temperature and precipitation had a greater impact on the change in the vegetation phenology compared with the effects of the individual climate conditions. The results quantitatively show the degree of interaction between the variations in temperature and precipitation and their effects on the changes in the different phenological parameters in the various months. Understanding how various climatic variations effect phenology changes in plants at different times may be more intuitive. This research provides as a foundation for research on how global climate change affects ecosystems and the global carbon cycle.

Key words： phenology; climate change; Siberia; asymmetric Gaussian function; GeoDetector

Cite this article

WEN Kege , LI Cheng , HE Jianfeng , ZHUANG Dafang . Quantitative response of vegetation phenology to temperature and precipitation changes in Eastern Siberia[J]. Journal of Geographical Sciences, 2024 , 34(2) : 355 -374 . DOI: 10.1007/s11442-024-2208-0

1 Introduction

Phenology is an indicator of climate. By studying the relationship between phenological change and climate change, we can understand better how climate change affects ecosystems (Lieth, 1974; Zhu, 1975; Dai, 2013; Liu, 2018). According to Zu et al. (2016), the phenology of broad-leaved forests, coniferous forests, and meadows in the springtime in northeastern China was adversely related to the springtime temperature. Additionally, there is a connection between the winter temperature and the subsequent spring phenology. Rupiya et al. (2018) reported that the mean temperature at the start of the growing season in Xinjiang was mostly responsible for the start of cotton growth. Following specific guidelines, traditional phenological observation methods have been used to record the growth of vegetation in fields (Wan et al., 1979). The site observation technique can truly and precisely capture regional phenological events. However, this approach is time- and work-intensive and can only be used to investigate a small area. In places that are inaccessible, such as the Tundra-Tagar area in eastern Siberian, there are few phenological observation stations. However, in regions where the environment is nearly entirely unaffected by anthropogenic activities, changes in the phenology may be a good indicator of how plants and animals naturally respond to climate change (Ebata et al., 2001; Zeng, 2015; Xue et al., 2016; Zhao et al., 2017).
By tracking the growth of vegetation across a vast region for a long time, remote sensing techniques can gather information about the phenology of plants (Huete et al., 2002; Varlamova et al., 2016; Jin et al., 2017). Since the 1980s, researchers have developed a number of techniques for gathering phenological data on plants using remote sensing technology. The combined normalized difference vegetation index (NDVI) curve fitting and dynamic threshold method is currently the most popular technique (Justice et al., 1985; Lloyd, 1990; Wen, 2016). Since 2000, moderate resolution imaging spectroradiometer (MODIS) data have gradually replaced advanced very high resolution radiometer (AVHRR) data as the main data source for remote sensing monitoring of phenology due to their better geographic resolution, expanded number of spectral bands, and shorter return intervals (Yu et al., 2006). Siberian vegetation phenology generated from high-resolution satellite images may contain significant errors because of the region’s heavy snowfall (Picard et al., 2005). The vegetation phenology in Siberia was generated from high-resolution satellite remote sensing images with snow cover interference. By further reducing clouds, snow, and shadows using MODIS MOD09 data, the Resource and Environmental Science and Data Center of the Chinese Academy of Sciences (RESDC) created a new set of NDVI data. Because of benchmark growth synthesis and enhanced interpolation methods, these NDVI products may more accurately depict the actual development process of the vegetation (Liu et al., 2017).
In the past, researchers have frequently used linear regression analysis to assess the impacts of many factors or to investigate the effect of a single factor on changes in the phenology of vegetation (Qiao et al., 2021; Wu et al., 2021). Linear regression analysis is a simple and rapid method for examining the direct correlations between variables. However, the complexity of these processes in terrestrial ecosystems results in spatial heterogeneity and a nonlinear distribution of the relationships between the vegetation phenology and the influencing factors. Additionally, the interactions between numerous factors that are common among geographical factors may be challenging to identify using linear regression analysis. To statistically determine the relationship between the spatial distributions of independent factors and the spatial distribution of dependent variables, the GeoDetector, which is based on the spatial statistics principle, uses the spatial heterogeneity of the features (Wang et al., 2010). Its essential principle is that two variables should have a consistent spatial distribution if they are causally related. The GeoDetector can be used to quantify the association between variations in temperature and precipitation in various months and changes in various phenological vegetation parameters. Additionally, the GeoDetector can be used to interpret and determine the degree to which distinct independent variable interactions effect dependent variables (Wang et al., 2016).
In an earlier study, we applied linear regression analysis to explore the response of vegetation phenological changes to temperature changes in the region because the Siberian region has experienced the most substantial temperature shifts (Li et al., 2023). However, despite the region’s small amount of precipitation, our research revealed that changes in precipitation still have a significant impact on the changes in the phenology of the local vegetation. Consequently, this study is a continuation of earlier studies and analyzes the responses of Siberian vegetation to the combined effect of temperature and precipitation variations.
The start of the season (SOS), end of the season (EOS), and length of the season (LOS) were extracted using the dynamic threshold approach, and the NDVI curve was fitted using the asymmetric Gaussian fitting method. The 18-year change trends of the temperature, precipitation, and three phenological indicators were investigated using linear regression. The degree of interpretation of the monthly temperature and precipitation changes from 2000 to 2017 with respect to the three phenological indicators were determined using the GeoDetector, and the extent of the effect of the interaction between the temperature and precipitation in the different months on the variations in the three phenological parameters was also investigated.

2 Data and method

2.1 Study area

As shown in Figure 1a, the study area is situated in eastern Siberia (115°-125°E) and is bordered by the Arctic Ocean to the north and the 60°N latitude line to the south. The Republic of Sakha controls the entirety of the region, which is located on Russian soil (Yakutia). As shown in Figure 1b, in this region, which extends from north to south, there are several vegetation types, including tundra, grassland, and deciduous coniferous forest. This region’s climate is typical of the continent, with significant seasonal temperature changes and considerable differences between day and night temperatures. The precipitation in the study area is sparse. In the northernmost area winter can last up to nine or ten months, while the summer is short. In the spring and fall, the temperature increases quickly. There are no significant cities or other locations suitable for development and habitation. Therefore, the specific phenological oscillations of the flora in this area may be roughly attributed to natural variability, and the effects of human activity can be ignored.
Figure 1 Study area shown in the world map and NDVI of study area (a); landcover map of the study area (b)

2.2 Data measure and process

2.2.1 Meteorology data set

The Climate Research Unit (CRU) meteorological data repository provided the meteorological information. The spatial resolution of the data is 0.5°×0.5°. The dataset spans from 1901 to the present. In this study, we employed the local NDVI data to mask and cut the global CRU data and the monthly mean temperature and precipitation data to ensure that the data were compatible with the spatiotemporal range of the NDVI data. From 2000 to 2017, a total of 216 temperature and precipitation data, or 12 observations per year, were extracted. Using the nearest neighbor method, the data were resampled to 0.01° to match the resolution of the NDVI dataset.
Eighteen years of monthly mean temperature and monthly precipitation data were extracted using linear regression analysis, the changes in the temperature and precipitation changes between 2000 and 2017 were calculated in accordance with the regression coefficients to characterize the linear trend of the monthly mean temperature and monthly precipitation.

2.2.2 NDVI data

The MODIS-NDVI data, based on MODIS MOD09Q1 reflectance data in the red and near-infrared bands, were acquired with a spatial resolution of 0.005° and a temporal resolution of 8 days. The original MODIS-NDVI data were then aggregated to 0.01°. A total of 828 images from 2000 to 2017 were utilized.

2.3 Phenological factors

2.3.1 Phenological parameters extraction

The descriptions of the three phenological factors included in this study are presented in Table 1. The three vegetation phenology parameters were derived from the fitted NDVI curve.
Table 1 Definition of vegetation growth season parameters
Vegetation growth season
parameter		 Definition
SOS When the left edge, as measured from the left minimum level, has risen to 10%.
EOS When the right edge, as measured from the right minimum level, has shrunk to 10%.
LOS Duration of the season, from the beginning to the end.
At high latitudes in the Northern Hemisphere, the NDVI curve of the vegetation growth season is asymmetrical; and the increase is more pronounced in the early stages and less pronounced in the latter stages. In this study, we used the asymmetric Gaussian fitting method, which is suitable for piecewise fitting, and used the maximum NDVI as the split point in order to fit the halves before and after each vegetation growing cycle (Jonsson et al., 2012).
To delineate the fluctuations in the vegetation growth season, a combination of numerous piecewise Gaussian functions was applied. Each combination represented a specific plant growing season. Then, using a type of smoothing technique, each Gaussian fitting curve was connected to recreate the time series.
The formula is:
$f(t)=f\left(t ; c_1, c_2, a_1, \ldots, a_5\right)=c_1+c_2 g\left(t ; a_1, \ldots, a_5\right)$
where a1,…, a5 are non-linear variables that affect how the function is shaped g(t; a1,…, a5) and c1 and c2 determine the reference plane and amplitude.
Among them:
$g\left(t ; a_1, \ldots, a_5\right)=\left\{\begin{array}{l}\exp \left[-\left(\frac{t-a_1}{a_2}\right)^{a_3}\right], t>a_1 \\ \exp \left[-\left(\frac{a_1-t}{a_4}\right)^{a_5}\right], t<a_1\end{array}\right.$
After fitting and smoothing the data, the TIMESAT software was used to extract three phenological parameters (SOS, EOS, and LOS). Given the characteristics of the shorter growing seasons and start and end of the rapid growth periods at high latitudes in the Northern Hemisphere, the threshold setting followed the following rules: it must be as close to the winter NDVI background value as possible; and it cannot be too low or the NDVI signal will be weakened by noise. Based on the results of previous studies, the dynamic criteria for the start and end of the growth season were determined to be 10% of the difference between the maximum and minimum values of the NDVI (Jonsson, 2004). The difference in the linear change in each parameter over the course of 18 years was then calculated via linear regression analysis on each parameter of the vegetation growth season. The parameters of each vegetation growth season and all of the pixels in regions with significant temperature fluctuations were subjected to trend analysis.

2.3.2 Trend analysis

Using linear regression analysis, the yearly and monthly phenology and meteorology factors for the 18-year total change was recovered. The formula is as follows:
$\left\{\begin{array}{c}k=\frac{\sum_{i=1}^n\left(x_i-\bar{x}\right)\left(y_i-\bar{y}\right)}{\sum_{i=1}^n\left(x_i-\bar{x}\right)^2}=\frac{\sum_{i=1}^n x_i y_i-n \overline{x y}}{\sum_{i=1}^n x_i^2-n \bar{x}^2} \\ b=\bar{y}-k \bar{x}\end{array}\right.$
$\bar{y}=kx+b$
$D=k\times 20$
where k is the slope, b is the intercept, n is the time period, i denotes the i years following 2000, x is the year, and y is the meteorological or phenological component. $D$ is the variation in the meteorological or phenological factor between 2000 and 2017. $\bar{x}$ is the mean value for the year, $\bar{y}$ is the mean of the meteorological or phenological factor, and y is the mean value for the year.

2.3.3 Application of GeoDetector

The GeoDetector tool can be used to evaluate spatial heterogeneity and to explore the underlying mechanism. The GeoDetector can calculate the interpretation degree of the independent variables to dependent variables. The fundamental assumption of the GeoDetector is that the distribution of X is spatially consistent with the distribution of Y when there are physically reasons and results. The value of Q indicates the proportion of Y that can be explained by X.
$Q=1-\frac{\mathop{\sum }_{h=1}^{L}\mathop{\sum }_{i=1}^{{{N}_{h}}}{{\left( {{Y}_{hi}}-{{{\bar{Y}}}_{h}} \right)}^{2}}}{\mathop{\sum }_{i=1}^{N}{{\left( {{Y}_{i}}-\bar{Y} \right)}^{2}}}=1-\frac{\mathop{\sum }_{h=1}^{L}{{N}_{h}}\sigma _{h}^{2}}{N{{\sigma }^{2}}}=1-\frac{SSW}{SST}$
the sum of all squares is:
$SST=\underset{i}{\overset{N}{\mathop \sum }}\,{{\left( {{Y}_{i}}-\bar{Y} \right)}^{2}}=N{{\sigma }^{2}}$
and the squares’ internal sum:
$SSW=\underset{h=1}{\overset{L}{\mathop \sum }}\,\underset{i}{\overset{{{N}_{h}}}{\mathop \sum }}\,{{\left( {{Y}_{hi}}-{{{\bar{Y}}}_{h}} \right)}^{2}}=\underset{h=1}{\overset{L}{\mathop \sum }}\,{{N}_{h}}\sigma _{h}^{2}$

3 Results

3.1 Distribution of the trend in climate change

With the exception of July and August, the monthly mean temperatures in the study area exhibited increasing trends from 2000 to 2017 in the majority of regions (Figure 2). The pattern of rising temperatures in the spring was quite clear.
Figure 2 Linear trend of monthly mean temperature in the study area from 2000 to 2017
More specifically, the monthly mean temperature increased by more than 2℃ in the majority of the study area in January, March, April, and October, as well as in the southern part in December. The warming exceeded 3℃, and the largest warming trends occurred in March and April. All of the regions exhibited a cooling trend, with cooling of around 1℃, except for in July, the northwest in August and the south in November.
The monthly precipitation changes (Figure 3) were remarkably diverse. The precipitation significantly decreased in the southwest in June and August, the central and northern regions in July, and the east in September. The southwest experienced a decrease in precipitation of more than 30 mm during August. In the northeast in March, the south in July, and the central-eastern region in August, the precipitation noticeably increased. The precipitation increased the most in July in the southeast, by a total of more than 30 mm.
Figure 3 Linear trend of monthly precipitation in the study area from 2000 to 2017

3.2 Pattern of phenological parameters and their change trend

Figure 4a demonstrates that the SOS was mostly concentrated between days 135 and 140, i.e., in the middle and end of May, marking the beginning of the vegetation growth season in the study area.
Figure 4 Spatial distribution of the mean SOS (a), EOS (c), and LOS (e) for the 18-year period between 2000 and 2017 and the linear regression-based difference in SOS (b), EOS (d), and LOS (f) in the same period
As shown in Figure 4b, the SOS generally advanced in the study area, by −20 to −5 days during the 18-year study period. There was only a clear trend of postponement along the coast of the Arctic Ocean in the research area’s farthest northern boundary. As shown in Figure 5, the spatial distribution of the SOS changed with the latitude. In the area of 64°-67°N at 71°N, the SOS peaked and advanced by 25 days (p = 0.0205 < 0.05).
Figure 5 SOS distribution trend along latitude
Figure 4c demonstrates that the vegetation in the study area reached the end of its growing season between mid-September and late-October. The EOS was primarily concentrated between days 265 and 275.
With a change range of −30 to 30 days for the advancement in the north and southwest and the postponement in the south, Figure 4d shows how the geographical distribution of the EOS exhibited evident indicators of a latitudinal shift. As shown in Figure 6, the difference in the EOS varied with latitude. The region north of 67°N experienced the most pronounced EOS advancement, with an advancement range of 15-30 days (p = 0.0325 < 0.05). The region south of 67°N contained the majority of the area where the EOS was delayed, and the delay ranged from 5 to 25 days. The maximum delay of 30 days occurred at 63°N (p = 0.0403 < 0.05).
Figure 6 Difference in EOS distribution trend along latitude
As can be seen from Figure 4e, the LOS was primarily concentrated between 120 and 170 days, indicating that the growth season lasted only 4-6 months. The area with the shortest LOS was located between 71°N and 72°N. Both the southernmost point of the study area and the region around 64°N had the longest growth season.
As demonstrated in Figure 4f, the shortening of the LOS, about 25 days or more, in the northern region was the greatest within 68°-72°N. The LOS in the northern part of the study area exhibited a trend for shortening, shortening by 15-25 days (p = 0.0301 < 0.05). The advancement varied from 20 to 30 days (p = 0.0379 < 0.05), with the greatest advancement occurring in the western part of the study area near 66°N and the southwestern part of the study area, up to 30 days or more.

3.3 Understanding phenological parameter changes caused by temperature and precipitation change

3.3.1 Phenological response to single climate change

The Q values in Table 2 indicate how much the regional distributions of the temperature and precipitation changes explain the spatial distribution of the phenological changes. A greater Q value denotes a stronger interpretation of the factors when interpreting them, such as the effect of the monthly temperature on the spatial changes in the SOS, EOS, and LOS temporal trends.
Table 2 Qs and the significance of phenological parameters trends interpreted by monthly mean temperature
SOS EOS LOS
Q pValue Q pValue Q pValue
T1 0.074 0.000 0.341* 0.000 0.274* 0.000
T2 0.086 0.000 0.273* 0.000 0.198 0.000
T3 0.049 0.000 0.371* 0.000 0.241* 0.000
T4 0.079 0.000 0.384* 0.000 0.275* 0.000
T5 0.016 0.606 0.332* 0.000 0.227* 0.000
T6 0.036 0.000 0.157 0.000 0.101 0.000
T7 0.052 0.000 0.128 0.000 0.061 0.000
T8 0.069 0.000 0.283* 0.000 0.231* 0.000
T9 0.105 0.000 0.375* 0.000 0.289* 0.000
T10 0.062 0.000 0.307* 0.000 0.246* 0.000
T11 0.072 0.000 0.332* 0.000 0.243* 0.000
T12 0.028 0.007 0.297* 0.000 0.203* 0.000
P1 0.020 0.173 0.157 0.000 0.123 0.000
P2 0.053 0.000 0.260* 0.000 0.238* 0.000
P3 0.046 0.000 0.101 0.000 0.075 0.000
P4 0.073 0.000 0.181 0.000 0.176 0.000
P5 0.048 0.000 0.080 0.000 0.071 0.000
P6 0.046 0.000 0.222* 0.000 0.174 0.000
P7 0.036 0.000 0.321* 0.000 0.182 0.000
P8 0.069 0.000 0.136 0.000 0.108 0.000
P9 0.036 0.000 0.100 0.000 0.106 0.000
P10 0.055 0.000 0.052 0.000 0.054 0.000
P11 0.053 0.000 0.111 0.000 0.102 0.000
P12 0.079 0.000 0.353* 0.000 0.269* 0.000
In Table 2, T1, T2, …, T12 denotes the variations in temperature between 2000 and 2017 in January, February, …, December. P1, P2, …, P12 denotes the variations in precipitation between 2000 and 2017 in January, February, …, December. * means Q > 0.2.
Table 2 illustrates that the variations in the temperature and precipitation within each month did not adequately account for the change in the SOS. A significant amount of the EOS variations was explained by the temperature and precipitation fluctuations over a number of months (Q > 0.200). The EOS change was best explained by the variations in temperature in March, April, and September. The most compelling explanation for the LOS shift was the temperature changes in January, April, and September.

3.3.2 Response of phenology to the interaction between multiple temperatures and precipitation

Tables 3-5 display the interpretation of the interactions between the monthly temperature and precipitation changes. Tables 3-5 show that the effect of the interaction between the temperature and precipitation changes in various months was more significant than for their individual effects. The high interpretation degrees for the change in the SOS were T7∩P4 (Q=0.350), P4∩P11 (Q=0.348), and T6∩P6 (Q=0.346). The high interpretation degrees for the change in the EOS were T4∩P4 (Q=0.552), T4∩P3 (Q=0.551), and P3∩P12 (Q=0.550). The high interpretation degrees for the change in the LOS were P3∩P12 (Q=0.474), T4∩P4 (Q=0.471), and P4∩P7 (Q=0.471).
Table 3 Qs of sOs trends interpreted by interaction of temperature and precipitation

Note: **Top ten in terms of interpretations.

Table 4 Qs of EOS trends interpreted by interaction of temperature and precipitation

Note: **Top ten in terms of interpretations.

Table 5 Qs of LOS trends interpreted by interaction of temperature and precipitation

Note: **Top ten in terms of interpretations.

The GeoDetector can quantitatively display the difference in the interpretation of each factor in the different months. As shown in Table 2, each phenological parameter had a good level of explanation for the temperature fluctuations in March, April, and September. This demonstrates that the major factor influencing the vegetation phenology changes was the pre-seasonal climatic change, that is, climate change has a hysteresis effect on the phenology. The influence of climate change on the vegetation phenology was somewhat gradual in spring and relatively rapid in autumn. In addition, despite the fact that vegetation did not grow in the winter, the climate change in winter accounted for a significant portion of the changes in the phenology of the vegetation.
Detecting the degree of interpretation of the various independent variable interactions on the dependent variables is the most unique function of the GeoDetector. Tables 3-5 demonstrate that the explanatory power of the effects of temperature and precipitation on the vegetation phenology is higher interpretability compared to those of the individual factors, especially the effect of temperature and precipitation in spring. Although temperature was the primary factor influencing the phenology of the vegetation in this region, temperature and precipitation together caused the vegetation to germinate in the spring and diminish in the fall. The temperature increase lead to vegetation drought stress, and thus, more precipitation was required to promote vegetation growth.

4 Discussion

In this study, first, we quantified the degree of interpretation of the phenological changes caused by the changes in different meteorological factors in different periods and their interactions in the high latitude Tundra-Taiga transition zone in the Northern Hemisphere. The differences in the explanatory power of each monthly temperature and precipitation change on the phenological parameters were revealed. Among them, the change in temperature in June and July and the change in precipitation in March, April, and May played vital roles in the change in the SOS. Notably, the change in the winter precipitation also exhibited a higher degree of interpretation for the change in the SOS. The interpretation degree for the EOS changes influenced by climate change was higher than that of the SOS. The interaction between the spring temperature and precipitation changes, as well as the changes in the late summer temperature, had a high interpretation degree for the EOS changes. The change in the LOS was the combined results of the changes in the SOS and EOS.
Significantly, the interaction between the changes in temperature and precipitation in spring had a high degree of explanation for both the SOS and EOS. This finding is consistent with the results of Fu et al. (2014). Fu highlighted that the heat requirement, which is expressed in the growing degree-days (GDDs), is a widely used method for assessing and predicting the effect of temperature on plant growth. Additionally, he found that there was a positive correlation between the GDD requirement and the precipitation in the previous season. In the high latitude Tundra-Taiga transition zone in the Northern Hemisphere, the heat requirement (GDD requirement) was mainly from the temperature in June and July. The interaction between the changes in temperature in June and July and the changes in precipitation in March, April, and May also had a high degree of explanation for the vegetation phenology. This was due to the hysteresis effect of the precipitation on the vegetation, while the effect of temperature occurred in real time. Previous studies have also revealed the promoting effect of early spring precipitation on vegetation phenology (Piao et al., 2003). The use of the GeoDetector not only confirmed the results of the above research but also quantified the explanatory degree of the interaction between the temperature and precipitation changes in the different months in spring to the vegetation phenology, and it even confirmed that the interpretation degree of this interaction was greater than the sum of the interpretation degrees of the two individual variables. However, the mechanism of the interaction with the phenology is complex. In any case, the results of the GeoDetector analysis provide direction for further research on the impact of climate change on phenology.
The vegetation types rarely changed in the study area between 2000 and 2017. To assess the overall response of all of the vegetation in the eastern Siberian Tundra-Taiga transition zone to climate change, in this study, we did not differentiate between the various types of plants and used the something to represent the overall situation of all of the vegetation. This allowed us to simply, logically, and clearly demonstrate how the phenology of the plants changed in accordance with the various climatic change features. Undoubtedly, greater knowledge of the phenological reactions of vegetation to climate change will result from analysis of how the different types of vegetation react to climate change. However, the precision of the sample data places restrictions on the findings of this study, so we did not determine how the different vegetation types adapted to climate change. As a result, we restricted the scope of our research to the general phenological responses to climate change over the entire study area.
The study area is located in the high latitude region in the Northern Hemisphere, where the environment is harsh and there is little human activity. The local climate has changed significantly in recent decades. First, the NDVI data were generated from MODIS data, and then, the growth of the vegetation was classified to determine how the plants responded to climate change in their natural context, free from artificial disturbances. Finally, we used the asymmetric Gaussian method to fit the NDVI curves and the dynamic threshold method to define the remote sensing phenology, and three vegetation remote sensing phenological parameters describing the features of the vegetation growth season variations were recovered. According to the results of this study, the SOS in the study area generally advanced, while the EOS varied temporally according to the changes in the regional temperature. The LOS mainly exhibited an increasing trend. Based on comparison of the temporal and spatial distributions of the SOS and EOS from 2000 to 2017, the SOS exhibited a more pronounced variation with latitude than the EOS (Figure 5), and the EOS was shorter in length and more concentrated than the SOS. This implies that the spatial distribution of the change trend of the EOS clearly displayed latitudinal change features (Figure 6) based on a comparison of the linear spatial and temporal change trends of the SOS and EOS from 2000 to 2017.
We infer from the GeoDetector results that the EOS is more sensitive to climate change than the SOS. The interpretation degree of the spatial distribution of the change in the EOS is higher than that of the change in the SOS regardless of the spatial distribution of the temperature change or precipitation change. This is basically consistent with the conclusion that the autumn phenology of the vegetation in the middle and high latitude regions in the Northern Hemisphere is sensitive to temperature changes (Wu et al., 2021). This result is also consistent with the fact that the local vegetation steadily recovers as the temperatures rises and more precipitation falls in spring, while it quickly declines when the temperature drops and the air becomes dryer in autumn. This is also the reason we used asymmetric Gaussian fitting. This fitting method can be applied to high latitude and plateau regions. The vegetation growth season was short and uneven. The spring growth season began slowly, while the autumn growth season ended quickly. The GeoDetector results also show that in the study area, the temperature change had a greater degree of explanation for the phenological change in the vegetation than the change in precipitation. This finding is inconsistent with our earlier research (Li et al., 2022) on how the plants in the Qilian Mountains in the mid-latitude region in the Northern Hemisphere responded to climate change. The phenological shift in the flora was more strongly impacted by the change in precipitation in the Qilian Mountains even though the fact that the two locations have comparable vegetation types and cold and dry climates.
There are still some limitations in this study even though this research represents the first quantitative interpretation of temperature and precipitation changes over the course of different months and their effects on various phenological changes in the Tundra-Tagar transition zone in the high latitude region of the Northern Hemisphere. First, the precision was restricted by the remote sensing data used for this study area, and little research has been conducted on how the various vegetation types respond to climate change. Therefore, the scope of our analysis was restricted to specific places with characteristics of both phenological variation in vegetation and climate change. In light of the findings of this study, field sampling will be conducted in regions where the phenology and climate change are readily apparent, and higher resolution meteorological data and more precise vegetation coverage data will be obtained, with the goal of concentrating on how different vegetation types are responding to climate change. Second, the study area’s climate change trend cannot be accurately captured by the low resolution of the CRU meteorological dataset. Because of the resolution of the climate data, we did not discuss the effects of small-scale special climate change regions on the vegetation phenological factors. Instead, we described how the vegetation phenological characteristics observed through remote sensing responded to the general climate change trend. After obtaining higher precision meteorological data, collecting field survey sampling data, and combining higher resolution satellite remote sensing images, the local climate and vegetation phenological parameters can be examined precisely in the future. Finally, how various vegetation types respond to climatic changes and how susceptible they are to them are still unknown. While climate change causes changes in vegetation phenology, these changes in phenology also have an impact on climate, which in turn affects climate change. Therefore, more research on the feedback mechanism between climatic and phenological factors is needed.

5 Conclusions

Based on remote sensing data, in this study, we gathered information on temperature, precipitation, and other climate change trends in the high latitude Tundra-Taiga transition zone in the Northern Hemisphere from 2000 to 2017. We also collected information on the change trends of the vegetation growth season parameters (SOS, EOS, and LOS). The impact of the spatial-temporal characteristics of climate change during the 18-year study period on the changes in the various phenological parameters of the vegetation was examined in detail. In addition, the GeoDetector was used to detect the explanatory powers of temperature and precipitation in the different months and their interactions on the changes in the various phenological parameters. The conclusions of this study are as follows.
(1) In the study area, the changes in temperature were the primary driver of the phenological change, and the changes in the pre-season temperature accounted for the majority of the changes in the SOS and EOS. Moreover, the EOS was more sensitive to climate change than the SOS.
(2) More extensive interpretation of the changes in the vegetation phenology was provided by the interaction of the monthly mean meteorological components. The combined changes in the temperature and precipitation, especially the increases in the temperature and precipitation in spring, which favored the advance of the spring phenology period, caused the changes in the vegetation phenology within the study area.
In this study, the impacts of the variations in temperature, precipitation, and their interactions on the phenological changes in vegetation were quantitatively investigated. The findings indicate that there is a strong correlation between the vegetation phenology and climate. Determining their coupling mechanism will help us to better understand how vegetation and climate change interact.

Acknowledgements

Thanks to Professor Dr. Liu Ronggao, Professor Dr. Xu Xinliang, and Associate Professor Dr. Liu Yang in the Institute of Geographic Sciences and Natural Resources Research of the Chinese Academy of Sciences, the Chinese Academy of Sciences Resource and Environmental Science Data Center (http://www.resdc.cn/), for providing data support for this research.

References

Publishing order | Descend order by publishing year | Descend order by cited within
[1]
Dai J, Wang H, Ge Q S, 2013. Changes of spring frost risks during the flowering period of woody plants in temperate monsoon area of China over the past 50 years. Acta Geographica Sinica, 68(5): 593-601. (in Chinese)

DOI

[2]
Ebata M, Tateishi R, 2001. Phenological stage monitoring in Siberia by using NOAA/AV HRR data. The 22nd Asian Conference on Remote Sensing, Singapore.

[3]
Fu Y, Piao S, Zhao H et al., 2014. Unexpected role of winter precipitation in determining heat requirement for spring vegetation green-up at northern middle and high latitudes. Global Change Biology, 20(12): 3743-3755.

DOI PMID

[4]
Huete A, Didan K, Miura T et al., 2002. Overview of the radiometric and biophysical performance of the MODIS vegetation indices. Remote Sensing of Environment, 83(1/2): 195-213.

DOI

[5]
Jin H, Jonsson AM, Bolmgren K, er al., 2017. Disentangling remotely-sensed plant phenology and snow seasonality at northern Europe using MODIS and the plant phenology index. Remote Sensing of Environment, 198: 203-212.

DOI

[6]
Jonsson P, Eklundh L, 2002. Seasonality extraction by function fitting to time-series of satellite sensor data. IEEE Transactions on Geoscience and Remote Sensing, 40(8): 1824-1832.

DOI

[7]
Jonsson P, Eklundh L, 2004. TIMESAT: A program for analyzing time-series of satellite sensor data. Computers & Geosciences, 30(8): 833-845.

DOI

[8]
Justice C O, Townshend J R G, Holben B N et al., 1985. Analysis of the phenology of global vegetation using meteorological satellite data. International Journal of Remote Sensing, 6(8): 1271-1318.

DOI

[9]
Li C, Zhuang D, He J, et al, 2023. Spatiotemporal variations in remote sensing phenology of vegetation and its responses to temperature change of boreal forest in tundra-taiga transitional zone in the Eastern Siberia. Journal of Geographical Sciences, 33(3): 464-482.

DOI

[10]
Li C, Zou Y, He J et al., 2022. Response of vegetation phenology to the interaction of temperature and precipitation changes in Qilian Mountains. Remote Sensing, 14(5): 1248.

DOI

[11]
Lieth H, 1974. Phenology and seasonality modeling. Ecological Studies, 120(6): 461.

[12]
Liu R, Shang R, Liu Y et al., 2017. Global evaluation of gap-filling approaches for seasonal NDVI with considering vegetation growth trajectory, protection of key point, noise resistance and curve stability. Remote Sensing of Environment, 189: 164-179.

DOI

[13]
Liu Y, Chen Q, Ge Q et al., 2018. Spatiotemporal differentiation of changes in wheat phenology in China under climate change from 1981 to 2010. Science China Earth Sciences, 48(7): 888-898. (in Chinese)

[14]
Liu Y, Ge Q, Dai J, 2020. Research progress in crop phenology under global climate change. Acta Geographica Sinica, 75(1): 14-24. (in Chinese)

DOI

[15]
Lloyd D, 1990. A phenological classification of terrestrial vegetation cover using shortwave vegetation index imagery. International Journal of Remote Sensing, 11(12): 2269-2279.

DOI

[16]
Piao S, Fang J, Zhou L et al., 2003. Interannual variations of monthly and seasonal normalized difference vegetation index (NDVI) in China from 1982 to 1999. Journal of Geophysical Research: Atmospheres, 108(D14): 4401.

[17]
Picard G, Quegan S, Delbart N et al., 2005. Phenology modelling in Siberia and its impact on the carbon budget. Global Change Biology, 11(12): 2164-2176.

DOI

[18]
Qiao C, Shen S, Cheng C et al., 2021. Vegetation phenology in the Qilian Mountains and its response to temperature from 1982 to 2014. Remote Sensing, 13(2):286.

DOI

[19]
Rupiya X, Yang L, 2018. Monitoring spatial-temporal change of cotton phenology in Xinjiang and its response to climate change. Remote Sensing Technology and Application, 33(5): 923-931. (in Chinese)

[20]
Varlamova E, Solovyev V, 2016. Study of NDVI vegetation index in East Siberia under global warming. XXII International Symposium Atmospheric and Ocean Optics. Atmospheric Physics. International Society for Optics and Photonics.

[21]
Wan M, Liu X, 1979. Methods of Phenological Observation in China. Beijing: Science Press, 42-66. (in Chinese)

[22]
Wang J, Li X, Christakos G et al, 2010. Geographical detectors-based health risk assessment and its application in the neural tube defects study of the Heshun region, China. International Journal of Geographical Information Science, 24(1): 107-127.

DOI

[23]
Wang J, Zhang T, Fu B. 2016. A measure of spatial stratified heterogeneity. Ecological Indicators, 67: 250-256.

DOI

[24]
Wen K, 2016. Vegetation responses to climate change in northern high latitudes (2000-2013). Beijing: University of Chinese Academy of Sciences. (in Chinese)

[25]
Wu C, Wang J, Ciais P et al., 2021. Widespread decline in winds delayed autumn foliar senescence over high latitudes. PNAS, 118(16): e2015821118.

DOI

[26]
Xue B, Guo Q, Gong Y et al., 2016. The influence of meteorology and phenology on net ecosystem exchange in an eastern Siberian boreal larch forest. Journal of Plant Ecology, 9(5): 520-530.

DOI

[27]
Yu X, Zhuang D, 2006. Monitoring forest phenophases of Northeast China based on MODIS NDVI data. Resources Science, 28(4): 111-117. (in Chinese)

[28]
Zeng L, 2015. Remote sensing for crop phenology detection: Taking corn and soybeans as examples[D]. Wuhan: Wuhan University. (in Chinese)

[29]
Zhao G, Shi P, 2017. Sources of uncertainty in exploring rangeland phenology: A case study in an alpine meadow on the central Tibetan Plateau. Journal of Mountain Science, 14(9): 1827-1838.

DOI

[30]
Zhu K, Wan M, 1975. Phenology. Beijing: Science Press, 1-4. (in Chinese)

[31]
Zu J, Yang J, 2016. Temporal variation of vegetation phenology in northeastern China. Acta Ecologica Sinica, 36(7): 2015-2023. (in Chinese)

Options
Outlines

/

〈 〉