Human’s digital footprints on the Qinghai-Tibet Plateau: Variations during festivals and impacts on nature reserves

  • DU Yunyan , 1, 2 ,
  • TU Wenna 1, 2 ,
  • LIANG Fuyuan 3 ,
  • YI Jiawei , 1, 2, *
  • 1. State Key Laboratory of Resources and Environmental 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. Department of Earth, Atmospheric, and Geographic Information Sciences, Western Illinois University, Macomb, IL61455, USA
*Yi Jiawei (1988-), PhD and Associate Professor, specialized in spatiotemporal data mining. E-mail:

Du Yunyan (1973-), PhD and Professor, specialized in spatio-temporal data mining. E-mail:

Received date: 2020-03-22

  Accepted date: 2020-09-10

  Online published: 2021-04-25

Supported by

Strategic Priority Research Program of the Chinese Academy of Sciences(XDA20040401)

Strategic Priority Research Program of the Chinese Academy of Sciences(XDA19040501)

National Key Research and Development Program of China(2017YFB0503605)

National Key Research and Development Program of China(2017YFC1503003)

National Natural Science Foundation of China(41901395)


Copyright reserved © 2021. Office of Journal of Geographical Sciences All articles published represent the opinions of the authors, and do not reflect the official policy of the Chinese Medical Association or the Editorial Board, unless this is clearly specified.


Mobile internet and wireless communication technologies have produced unprecedented location-aware data. Such big geospatial data can be used as a proxy measure of the ‘digital footprints’ left by us on the planet and provide a valuable opportunity to understand the dynamic and short-term human disturbance on the nature at fine scales. This study investigated the spatiotemporal variations of human’s digital footprints on the Qinghai-Tibet Plateau using smartphone-users-generated Tencent’s location request data. The results showed that human’s digital footprints cover less than 5% of Qinghai and Tibet, exhibiting either a U-shaped or an N-shaped temporal change pattern during the major festivals. Spatial changes of the digital footprints manifested a transition process from dispersion to concentration in Xining and Lhasa. Human disturbance assessment of seven large nature reserves on the plateau showed that the Qinghai Lake is the most disturbed one as shown by 14.6% of its area is stained with human digital footprints and the areal average of footprint intensity is 1.59, and the disturbance was significantly escalated during the National Day holiday. By contrast, the Qangtang and Hoh Xil are the least affected nature reserves with the two indices less than 1% and 0.1, respectively.

Cite this article

DU Yunyan , TU Wenna , LIANG Fuyuan , YI Jiawei . Human’s digital footprints on the Qinghai-Tibet Plateau: Variations during festivals and impacts on nature reserves[J]. Journal of Geographical Sciences, 2021 , 31(2) : 179 -194 . DOI: 10.1007/s11442-021-1841-0

1 Introduction

The Qinghai-Tibet Plateau (QTP) provides an immense but fragile habitat for countless unique plant and animal species (Xu et al., 2017). It is also the source of the major rivers in Asia, which provides water to nearly 40% of the world population (Xu et al., 2008). The ecosystem of the QTP is vulnerable to climate changes (Cheng et al., 2007; Fang et al., 2007; Xing et al., 2009) and human disturbance (Yu et al., 2012; Fan et al., 2015). It is a globally recognized strategy to establish natural reserves to effectively protect ecosystem and biodiversity (Watson et al., 2014; Xu et al., 2019). By 2012, China has established 155 national nature reserves on the QTP, covering a totally area of 822,400 km2 (Zhang et al., 2015). Specific regulations and laws have been legislated and enacted to alleviate human’s pressure on the natural reserves (Xu et al., 2019). The Chinese government has also implemented an ‘ecological migration’ project, which relocates the residents in protected areas of natural reserves to nearby villages or towns to better protect the environment (Mao et al., 2012). However, natural reserves are still inevitably under significant human disturbance.
Human may disturb the nature at both a short- and a long-range scale. The long-term disturbance, such as urban expansion, deforestation, mineral exploitation, excessive farming and over-grazing, etc., have been extensively studied (Yan et al., 2011; Yu et al., 2012; Fan et al., 2015; Zhao et al., 2015; Li et al., 2017). Sanderson et al. (2006) and Venter et al. (2016) quantified human pressure on nature by integrating multiple human-related thematic datasets, including human population density, land transformation, transportation network, and power infrastructure. Li et al. (2018) mapped human’s pressure on QTP and reported more intensive disturbances in the eastern QTP, southeastern QTP, and the central Tibet, and a significant increase in the northeastern QTP from 1990 to 2010. In another study, Li et al. (2018) showed declined human impacts on natural reserves in Tibet from 1990 to 2010.
Evaluating the short-term human impacts on the ecosystem of the QTP, however, remains a challenge. The QTP is a famous tourism destination in China and it attracted more than 30 million domestic and international visitors in 2018. Such a huge number of visitors pose great pressure on the ecosystem due to some destructive behaviors such as illegal crossing, littering, and wildlife interfering. Better assessment of short-term human disturbance requires knowledge of the human dynamics, which is unfortunately not available in current ambient population products (Dobson et al., 2000; Azar et al., 2013; Stevens et al., 2015).
Location-based services produce a large amount of geospatial data, also known as “digital footprints” (Weaver and Gahegan, 2010; Walden-Schreiner et al., 2018), from which short-term human dynamics within a geographic area could be inferred (Longley and Adnan, 2016; Senaratne et al., 2017). Such big data include the instant locations where smartphone or internet users make requests for location-based services, such as taxi hailing, car navigation, social media check-ins, uploading a geo-tagged microblog and/or a picture and so on. Researchers have used human’s digital footprints to estimate dynamic population (Yao et al., 2017), examine the urban diurnal rhythm (Ma et al., 2019) and human activities in protected areas (Walden-Schreiner et al., 2018).
In this study, we used Tencent’s location-request data to explore the spatiotemporal variations of human’s digital footprints in response to the major holidays and the short-term human disturbance on QTP’s nature reserves. The remainder of the paper is organized as follows. Section 2 describes the study area and the datasets we used. Section 3 elaborates the methods for measuring the spatiotemporal variations in digital footprints and assessing the human disturbance on nature reserves based on the digital footprints. Section 4 presents the analysis results, including the mean characteristics, the spatiotemporal variation patterns, and the short-term disturbance on the nature reserves in response to the major festivals in QTP. Section 5 concludes the study.

2 Materials

2.1 Study area

The QTP is located in the southwest of China with an average elevation exceeding 4,000 m. It covers an area of over 2.5 million km2, extending from the Kunlun-Qilian Mountains in the north to the southern edge of the Himalayas in the south and from the Pamir Plateau in the west to the Hengduan Mountains in the east (Figure 1). This study examined the human’s digital footprints in Qinghai and Tibet. The area of the two provinces accounted for 77% of the QTP. There are 15 prefecture-level administrative units including two capital cities, i.e., Xining and Lhasa. The study area is limited to the administrative regions in order to facilitate comparison with census data, which are collected by administrative unit. The total population of Qinghai and Tibet was about 9.35 million in 2015. All townships in the study area have been covered by 3G network1 (1According to the 2017 statistical yearbook of Qinghai and Tibet available at and and the numbers of mobile phone users per 100 people in Qinghai and Tibet are 102.9 and 87.7, respectively.2( 2 According to the 2017 annual statistics of the Ministry of Industry and Information Technology of China available at n1146904/n1648372/c6048613/content.html.)
Figure 1 Map of the Qinghai-Tibet Plateau
We examined the human disturbance on seven national nature reserves. Qinghai and Tibet have established 16 national nature reserves of different sizes. The analysis focuses on large nature reserves with an area of more than 5000 km2. Multi-part nature reserves (e.g. Yaluzangbu midstream nature reserve) with no single protected area exceeding the threshold were excluded. The total area of the seven nature reserves sums up to more than 563,000 km2, accounting for about 29% of the total area of Qinghai and Tibet. The Qangtang nature reserve, which is the largest one (~298,000 km2), was established for protecting the ungulate animals and plateau desert ecosystem. The Sanjiangyuan nature reserve is categorized as a wetland reserve that protects the origin of the Yellow River, the Yangtze River and the Lancang River. The Mount Qomolangma and Yarlung Zangbo River regions are forest reserves that protect mountain forest ecosystems and biodiversity on the plateau. Hoh Xil, Serlin Co and Qinghai Lake were established as wildlife reserves that protect endangered plateau species including Tibetan antelope, wild yak, snow leopard, black-necked crane, etc.

2.2 Tencent’s location request data

The location-request data were collected from the Tencent big data portal (available at, last access: 2 April 2019). Tencent is the most popular social media platform in China with nearly 1.1 billion monthly active users as of 2018 (https:// A record of location request is generated when a user seeks any location-based services (LBS) via Tencent’s Map Application Programming Interface. The LBS-related requests include but are not limited to car hailing, navigation, food and merchandise delivery that are generated from mobile apps including WeChat, Mobile QQ, Tencent Video, JingDong, Dianping, DiDi, Meituan Waimai, etc. (Yi et al., 2019). Such requests were aggregated into 0.01-by-0.01-degree grids, with each grid recording the hourly number of the location requests (NLR) within it. We collected and prepared the gridded NLR data from January 1, 2018 to February 28, 2019 for this study. Many recent studies are constantly trying to use location data to reveal the spatiotemporal changes of human activities. Ma et al. (2018) compared the NLR dataset with visitor numbers in a few places in China and confirmed that the NLR dataset is a good proxy measure of short-term population dynamics.

2.3 Population datasets

The study involves two population datasets. The first one is the prefecture-level 1% population survey data acquired in 2015. The second dataset is the gridded population product (GPL) developed by the LandScanTM ( The GPL products estimate the ambient population count (average over 24 hours) in every 30-by-30 arc-second grid by integrating multiple dasymetric layers, including census population, land use and land cover map, terrain map, and nighttime light satellite images.

3 Methods

3.1 Measuring human’s digital footprints

We first analyzed the spatiotemporal characteristics of the NLR. The daily time series NLR (hereafter referred to as Nd) defined as follows were constructed from the original hourly NLR dataset spanning from January 1, 2018 to February 28, 2019:
$g(i)=\left\{N d_{i}^{(1)}, N d_{i}^{(2)}, \ldots, N d_{i}^{(j)}\right\}$
where i and j denote the ith grid in the dataset and the jth day in our study time period. The grids with a total Nd less than 10 over the entire 424 days are mainly located in the wild areas and empirical observations show the very low NLR counts in these grids may be due to locating errors. As a result, we excluded such grids from this study.
We then calculated the average daily NLR (hereafter referred to as,$ \overline{N d}$), which is defined as follows:
$N d_{l}=\frac{\sum_{j=1}^{m} d_{i}^{(i)}}{m}$,
where m denotes the number of the days that are involved in calculating the$\overline{N d}$. The stability of the $\overline{N d}$ distribution is evaluated by indicator S(t), which is the mean of the absolute difference between two $\overline{N d}$ layers of the accumulative t and t+1 days.
$S(t)=\frac{1}{n}\sum\nolimits_{i=1}^{n}{\left| \frac{\sum\nolimits_{j-1}^{l+1}{Nd_{i}^{(j)}}}{t+1}-\frac{\sum\nolimits_{j=1}^{t}{Nd_{i}^{(j)}}}{t} \right|}$,
Two indexes, the occupancy rate (δ) and areal average (ε), were used to characterize the digital footprints within a specific administrative unit, nature reserve, or the entire study area. The occupancy rate describes the degree to which a region has been occupied by humans in terms of the digital footprints. It is defined as the ratio of the area with a positive Nd to the total area (Eq. (4)).
$\delta=\frac{\left.A^{\prime}\right|_{N d}>0}{A} \times 100 \%$
where A denotes the total area and ${A}'{{|}_{Nd>0}}$ denotes the summed area of the grids with a positive Nd. The second index, the areal average, describes the magnitude of human’s digital footprints in a region. It is defined as follows:
$\varepsilon=\frac{\sum_{i=1}^{n} N d_{i}}{n}$,
where n denotes the number of grids within the region.

3.2 Exploring spatiotemporal variations of digital footprints

We first analyzed the temporal variations in the aggregated digital footprints by different geographic units, including the prefecture-level administrative unit (PAU), scenic spot, and the popular route to a scenic spot, respectively. In this study, the aggregated digital footprints are the sum of the Nd of the grids within a geographic unit of interest. Variations in aggregated digital footprints are represented as a festival vector, which is composed of the Nd values of three days before, three days during, and three days after a festival. The major festivals in our study area during 2018-2019 (Table 1) include five 3-day festivals and four 7-day festivals. For the five 3-day festivals, the festival vector is constructed with the Nd values of the three days before, the three days during, and the three days right after each festival. For each of the four 7-day festivals, the vector is defined by the Nd values of the 7th, 4th, 1st day before, and the 1st, 4th, 7th day during, and the 1st, 4th, 7th day after the festival. For each festival vector, the Nd values were standardized so that different vectors are comparable.
Table 1 Major festivals celebrated on the Qinghai-Tibet Plateau from January 1, 2018 to February 28, 2019
Festivals Abbreviation Start End Duration (days)
Spring Festival 2018 SF18 2018-02-15 2018-02-21 7
Qingming Festival QM 2018-04-05 2018-04-07 3
May Day MD 2018-04-29 2018-05-01 3
Dragon Boat Festival DB 2018-06-16 2018-06-018 3
Xuedun Festival XD 2018-08-11 2018-08-17 7
Mid-Autumn Festival MA 2018-09-22 2018-09-24 3
National Day ND 2018-10-01 2018-10-07 7
New Year 2019 NY 2018-12-30 2019-01-01 3
Spring Festival 2019 SF19 2019-02-04 2019-02-10 7
We then applied the agglomerate hierarchical clustering (AHC) approach (Tan et al., 2005) to the festival vectors to discover the temporal variation patterns of the aggregated digital footprints by each of the afore-mentioned geographic units. The AHC method initially takes each vector as an individual cluster. It starts to successively merge the two closest clusters until all clusters are merged into one (Tan et al., 2005). The dissimilarity between two vectors and the proximity between two clusters that the two vectors represent were measured by the Euclidean distance and the average linkage, respectively. We focused on the large clusters to investigate the general variation patterns of the Nd during the festivals. The generalized variation pattern of a cluster is represented by the average of the vectors involved. At last, a dendrogram was used to show the hierarchy of clusters and the cluster merging order.
We also conducted analysis at the grid scale to show the spatial variations in the Nd. The Nd layers of the three days before, during, and after a festival were integrated to generate pre-, during-, and post-festival aggregated Nd layers, respectively. We then compared the three aggregated Nd layers to examine the spatial variations of the Nd on the QTP.

3.3 Assessing the impacts of digital footprints on nature reserves

Human disturbance on protected areas of the QTP was assessed in seven large nature reserves. We used the occupancy rate and the areal average of digital footprints over a year to assess the long-term disturbance, and over major festivals between 2018/01/01 and 2019/2/28 to assess the short-term disturbance. The spatial expansion of digital footprints, which is referred to as ‘festival expansion’ hereafter, was further examined in nature reserves during the festivals. The festival expansion is defined by the areas where the Nd values increased from 0 before to more than 0 during a festival. A map of footprint festival expansion shows festival-associated short-term human invasions to the nature reserves.

4 Results

4.1 Digital footprints on the QTP

Human’s digital footprints were not widely distributed across the QTP. The average daily NLR map (Figure 2a) shows that human’s digital footprints only account for around 5% of the QTP within the time period of interest. The $\overline{N d}$. values range from less than 1 to over 104. The grids with a $\overline{N d}$ more than 10 are mainly located in urbanized areas and only accounts for less than 1% of the study region. Very high $\overline{N d}$ values (greater than 103) are only found in the core areas of Xining and Lhasa, where most QTP’s population reside. The digital footprints were also found along the major transportation networks, though with relatively low $\overline{N d}$ values (less than 10). The above mean characteristics represent the long-term distribution pattern of human’s digital footprint on the QTP. The stability index, which has asymptotically decreased to 0 after September of 2018 (Figure 2b), is indicative of the stabilized spatial distribution.
Figure 2 Mean distribution of the digital footprints on the QTP during 2018-2019. (a) the average daily NLR, (b) temporal variations in the daily NLR and the stability index, (c) correlations between daily NLR and census population at a logarithmic scale
The aggregated $\overline{N d}$ values by PAU are highly correlated with the census population (CP) and the LandScan gridded population (GP) at a logarithmic scale (Figure 2c). The Spearman’s rank correlation coefficient (0.74) between log($\overline{N d}$) and log (CP) is higher than the correlation coefficient between log($\overline{N d}$) and log (GP) (0.57). However, both correlations are statistically significant at the 0.05 significance level. The correlation coefficient between log (GP) and log (CP) is 0.81 and is also statistically significant. It is noteworthy that the LandScan GP may significantly overestimate the population of Shannan and Linzhi as compared to the census population (Figure 2c). The correlations show that the aggregated $\overline{N d}$ values could be used as proxy measures of human’s impact.

4.2 Variations in digital footprints during festivals

4.2.1 Temporal variations
Temporal variations of the digital footprints within different geographic units show either a U-shaped or an N-shaped pattern. The U-shaped pattern shows a lower Nd value during the festivals than those in the pre- and post-festival days. Conversely, the N-shaped pattern shows a higher Nd value during the festival days than those in the pre-festival and post-festival days. Both two patterns were found in the vectors that were constructed for the PAUs, the scenic spots, and the popular routes (Figures 3-5). For the PAU festival vectors, the U-shaped and N-shaped patterns are found in clusters C1 and C2 (Figure 3), which account for 81% and 16% of the total vectors. For vectors of scenic spots, the corresponding clusters of the U-shaped and N-shaped patterns are C1 and C3 (Figure 4), which account for 41% and 22%, respectively. Clusters C1 and C3 show the U-shaped and N-shaped patterns of the popular route vectors, which account for 43% and 22%, respectively (Figure 5).
Figure 3 Temporal variation patterns of the Nd in different PAUs during 2018-2019. (a) the hierarchical clustering results, (b) the cluster categories of different cities for different festivals, (c) the temporal variations in the Nd in different city clusters
Figure 4 Temporal variation patterns of the Nd at different scenic areas during 2018-2019. (a) the hierarchical clustering results, (b) the cluster categories of different cities for different festivals, (c) the temporal variations in the Nd in different city clusters
Figure 5 Temporal variation patterns of the Nd along the major routes to the scenic areas during 2018-2019. (a) the hierarchical clustering results, (b) the cluster categories of different cities on different festivals, (c) the temporal variations in the Nd in different city clusters
Other clusters in Figures 3-5 show similar but slightly different generalized temporal variations. The cluster C3 of the PAU vectors shows a similar U-shaped pattern but with a peak in the middle of the festivals. The cluster C2 of the scenic-area vectors shows an N-shaped pattern but with a peak on the first day of the festivals. The cluster C3 of the popular route vectors shows a U-shaped pattern with a significant dip on the first day of the festivals.
The temporal variation patterns also varied among festivals and geographic units. The U-shaped pattern was found in every PAU and for almost all festivals except the Xuedun festival, which shows an N-shaped variation pattern (Figure 3b). Xuedun is a traditional festival that usually attracts many local Tibetans. Specific to each PAU, different festivals may show various patterns. For example, the festivals show more N-shaped than U-shaped patterns in Haidong PAU. The collective human behaviors may show different responses to a festival in different PAUs. Such different response could be attributed to the multiple facets of a PAU, such as the level of economic development, transportation accessibility, population attractiveness, ethnic structures, and tourism resources.
We further examined the generalized patterns in clusters of scenic spots and tourist routes to understand how tourism drives the temporal variations of Nd during festivals. Figure 4b shows in which scenic spot and during which festival a U-shaped or V-shaped pattern is found. The hatched grids in Figure 4b show the vectors with a mean Nd less than 10 that are mainly found in winter and spring and in Xiannuwan, Nam Co, Rongbusi scenic spots when human activities are scarce due to the extreme cold weather and therefore were excluded from this study. For example, the Menyuan scenic area is well-known for its beautiful view of the canola flower field. Every year, it attracts a large number of tourists during July and August. As shown in Figure 4b, however, the vectors of Menyuan were almost all excluded except for the Xuedun festival which was celebrated in August when visitors still had a chance to see canola flowering before the petals fall.
The temporal changes of the Nd in the scenic spots and tourist routes also varied with festivals. During the National Day, the vectors of Erlangjian, Taersi, Chaka Salt Lake, and Riyueshan all showed an N-shaped variation pattern (C3 in Figure 4). By contrast, the vectors of Nam Co and Yamzho Yumco showed a quasi-N-shaped variation pattern (C2 in Figure 4). The tourist routes R1, R4, and R5 to these scenic areas also showed a significant increase in the digital footprints during the holidays (Figure 5b). However, during the Spring Festival, most vectors of the scenic areas and tourist routes showed a U-shaped variation pattern, indicating declined tourist activities during the holidays. The religious Xuedun festival displayed an N-shaped variation pattern. During this festival, the two famous monasteries, Tashi Lhumpo and Drepung Monastery, hold the annual celebration and attract many believers and visitors across the country every year (Figure 4b). The tourism during the festivals at least partially if not completely explain the significant variations in the digital footprints over the QTP, regardless the U-shaped or the N-shaped variation pattern.
4.2.2 Spatial variation
Variation of the Nd across our study area during the festivals also manifest either a U-shaped or an N-shaped pattern at the grid scale. The grids within Xining and Lhasa were grouped into two clusters using the AHC method. The overall variations of the two major clusters displayed the U-shaped and the N-shaped pattern (Figures 6a and 6c), respectively. Those grids with a U-shaped or an N-shaped variation pattern of the Nd hereafter were referred to as U-type or N-type grids, respectively. The locations of the U-type and N-type grids are related to the distance to the city center. In the places within 18 km from the city center, there are more U-type than N-type grids. By contrast, there are more N-type than U-type grids in the areas beyond that distance (Figures 6b and 6d). Such a distribution pattern of the U-type and N-type grids was observed in almost every festival except the Xuedun festival, which shows a totally opposite pattern (dashed lines in Figures 6b and 6d). This probably indicates that the Xuedun festival, as a local religious festival, attracts local people or tourists in a way different from other festivals.
Figure 6 The U-shaped (blue line) and N-shaped (red line) variation patterns of the grids in Xining (a) and Lhasa (c). Variations in percentage of grids showing U-shaped or N-shaped variation patterns in relation to the distance to the city center of Xining (b) and Lhasa (d) during 2018-2019
The digital footprints show a transition from dispersion to concentration from pre-festival to post-festival. Figures 7a and 7c show the difference between the festival and the pre-festival Nd (Ndfestival - Ndpre-festival). The difference is negative in the urban areas but positive in the surrounding rural areas, indicating people start to leave the city before the festivals start. In contrast, the difference between the festival and the post-festival Nd (Ndpost-festival - Ndfestival) (Figures 7b and 7d) showed a positive value in the urban areas but a negative value in the surrounding rural areas, indicating people return to the city after the festivals. The dispersion and concentration process over the QTP show an opposite population migration pattern before and after the festivals.
Figure 7 Difference between the festival and the pre-festival Nd (Ndfestival - Ndpre-festival) in Xining (a) and Lhasa (c) and between the festival and the post-festival Nd (Ndpost-fstival - Ndfestival) in Xining (b) and Lhasa (d)

4.3 Digital footprints in nature reserves

Table 2 shows the two indexes δ and ε, which indicate the different degrees of the long-term human invasion in the seven nature reserves. The two indexes δ and ε range from 0.08% to 14.6% and 0.004 to 1.59, respectively. The Qinghai Lake has the highest δ and ε (14.6% and 1.59, respectively), indicating it is the natural reserve with the most significant human invasion. On the contrary, the Qangtang and Hoh Xil are the least affected nature reserves, as indicated by a less than 1% δ and a less than 0.1 ε.
Table 2 Statistics of the digital footprints and population in the nature reserves during 2018-2019
Nature reserves Area
(10,000 km2)
δ (%) Population
occupancy (%)
ε Population
per grid
Qangtang 29.8 0.08 7.88 0.004 0.095
Sanjiangyuan 15.2 3.00 37.6 0.222 1.309
Hoh Xil 4.50 0.50 2.61 0.014 0.056
Qomolangma 3.38 5.74 47.9 0.350 2.159
Serlin Co 2.03 3.12 43.0 0.017 1.242
Yarlung Zangbo 0.92 5.28 36.4 0.916 1.283
Qinghai Lake 0.50 14.6 23.5 1.590 2.382
The short-term human invasion (Table 3) during different festivals in different natural reserves is evaluated by examining the difference between the festival and the pre-festival δ (δfestival - δpre-festival) and ε (εfestival - εpre-festival). Over all the natural reserves, the changes of the indexes δ and ε range from 0 to 0.51% and 0 to 0.26, respectively. The Qinghai Lake shows the most significant changes in both δ and ε. In particular, the National Day festival shows the most significant increase in δ and ε in Qinghai Lake (1.12% and 0.81, respectively). By contrast, the Qangtang natural reserve has the lowest changes of both δ and ε, indicating there is barely any short-term invasion to this natural reserve over all festivals.
Table 3 Short-term changes of digital footprints in nature reserves during 2018-2019
Qangtang δ (%) -0.00 0.00 -0.00 0.00 0.00 0.00 -0.00 0.00 -0.00 0.00
ε -0.00 -0.00 -0.00 -0.00 -0.00 0.00 -0.00 -0.00 -0.00 0.00
Sanjiangyuan δ (%) -0.04 -0.04 -0.03 0.05 -0.03 -0.06 -0.08 0.06 0.02 0.04
ε 0.01 -0.01 0.02 -0.01 -0.02 0.02 0.00 0.03 -0.01 0.01
Hoh Xil δ (%) -0.06 -0.07 0.05 -0.03 -0.07 0.00 -0.05 -0.02 0.04 0.04
ε -0.00 -0.00 0.00 -0.00 -0.01 0.00 -0.00 0.01 0.00 0.00
Qomolangma δ (%) 0.18 0.20 -0.15 -0.27 -0.27 0.05 0.33 -0.18 0.08 0.19
ε 0.01 -0.01 -0.03 -0.05 -0.04 0.00 -0.01 -0.04 -0.01 0.02
Serlin Co δ (%) 0.02 -0.11 0.02 -0.18 0.03 0.39 -0.34 -0.33 0.07 0.16
ε 0.00 -0.00 -0.00 -0.00 -0.00 0.00 -0.00 -0.00 -0.00 0.00
Yarlung Zangbo δ (%) 0.13 0.22 -0.08 -0.19 0.16 -0.03 -0.05 -0.03 0.39 0.14
ε 0.05 0.01 -0.05 -0.09 -0.12 0.02 -0.19 0.07 0.05 0.07
Qinghai Lake δ (%) -1.00 0.59 -0.26 0.72 0.51 -0.10 -0.18 1.12 -0.13 0.51
ε -0.06 -0.06 0.07 0.49 0.31 -0.32 -0.16 0.81 -0.05 0.26

*MAC denotes the mean absolute change of index δ or ε.

Spatially, human invasion as shown by the expansion of human’s digital footprints are mainly observed along roads in all the natural reserves except Qangtang and Hoh Xil (Figure 8). Obviously, roads provide human with access to the protected areas. Moreover, the expansion of digital footprints in the Qinghai Lake is not limited along the roads and human invasion was identified near the scenic areas including Erlangjian, Heimahe, Niaodao, Xiannuwan, Jinshawan, and Haixinshan. Due to the significant human disturbance, the Niaodao scenic areas have been closed for the purpose of environmental protection since August 29, 2017. Festival expansion of digital footprints in Niaodao, however, suggests the visitation prohibition was not completely abided in 2018-2019, and stricter regulations should be implemented in festival holidays to protect the habitat from visitors’ disturbance.
Figure 8 Spatial expansion of human’s digital footprints in the nature reserves of Qangtang (a), Sanjiangyuan (b), Hoh Xil (c), Qomolangma (d), Serlin Co (e), Yarlung Zangbo Grand Canyon (f), and Qinaghai Lake (g) during 2018-2019. The small-scale maps show the spatial area of nature reserves in black color.
For comparison, we also calculated the occupancy rate and the areal average (Table 2) and the population distribution (Figure 8) in these nature reserves using the LandScan gridded population dataset. The results show that the gridded population product overestimates the human invasion in the protected areas. For instance, the occupancy rate derived from the LandScan gridded population data could be 42% more than that computed from the Nd in Qomolangma. It is very unlikely for human to occupy such a large and wild area in Qomolangma. It is worthy to note that the LandScan gridded population is an annual product, which is not appropriate to be used to identify short-term variations during festivals. By contrast, digital footprints may underestimate human impacts on the natural reserves due to multiple factors that may include but are not limited to poor signal coverage in remote areas, insufficient smartphone penetration rate, and less frequent uses of smartphones by the older and children.

5 Conclusions

This study investigated human’s digital footprints on the QTP based on Tencent’s location-request data. Such big geospatial data provides a valuable opportunity to study the short-term human impacts on the nature, which was almost an impossible task for the current existing census population and the LandScan gridded population datasets. The spatial coverage of the digital footprints only accounts for around 5% of the QTP. Digital footprints were concentrated in urban areas and along the transportation networks. By contrast, very limited footprints are observed in the remote rural areas.
The temporal variations of digital footprints in the prefectures, scenic attractions, and along the tourist roads during the festivals show either a U-shaped or an N-shaped pattern. However, the patterns varied slightly regarding a specific festival and geographic unit. Spatial changes of the digital footprints manifested a transition process from dispersion to concentration before, during, and after the festivals. Spatially, the grids near the urban centers tend to exhibit a U-shaped variation pattern and those far from the urban centers show an N-shaped variation pattern.
Human impact assessment on the seven largest nature reserves on the QTP based on the digital footprints shows that the Qinghai Lake is the natural reserve that is affected most by human activities, particularly during the National Day festival. Human’s digital footprints have occupied 14.6% of the area of the Qinghai Lake and the areal average of footprint intensity is high up to 1.59. On the contrary, the Qangtang and Hoh Xil are the least affected nature reserves with the two indexes less than 1% and 0.1, respectively.
Human impacts are mainly observed along the existing roads. However, the impacts have also been observed in some closed areas within the Qinghai natural reserve, indicating there is a need to more strictly implement regulations to better protect environment in the closed areas. Such findings confirmed that the digital footprints could be used to monitor the short-term human disturbance on nature reserves.
It is worthy to note the limitations of using the Tencent’s location-request data as a proxy of human’s digital footprints and the human impacts on the nature. Even the location-request data are generated by 1.1 billion monthly active users, they still cannot fully represent the entire population of China. Poor signal coverage, insufficient smartphone penetration rate, and infrequent uses by the elder and children are the major factors that may contribute to the underestimation of human disturbance to the nature environment. Our future studies will explore the possibility of overcoming such limitations by integrating multi-source geospatial datasets to better evaluate human’s impacts on the QTP.
Azar D, Engstrom R, Graesser J et al., 2013. Generation of fine-scale population layers using multi-resolution satellite imagery and geospatial data. Remote Sensing of Environment, 130:219-232.

Cheng G, Wu T, 2007. Responses of permafrost to climate change and their environmental significance, Qinghai-Tibet Plateau. Journal of Geophysical Research: Earth Surface, 112: F02S03.

Dobson J E, Bright E A, Coleman P R et al., 2000. LandScan: A global population database for estimating populations at risk. Photogrammetric Engineering and Remote Sensing, 66(7):849-857.

Fan J, Xu Y, Wang C S et al., 2015. The effects of human activities on the ecological environment of Tibet over the past half century. Chinese Science Bulletin, 60(32):3057-3066.

Fang H, Zhao F, Lu Y et al., 2007. Remote sensing survey of ecological and geological and environmental factors in Qinghai-Tibet Plateau. Remote Sensing for Land and Resources, 19(4):61-65.

Li S, Wang Z, Zhang Y, 2017. Crop cover reconstruction and its effects on sediment retention in the Tibetan Plateau for 1900-2000. Journal of Geographical Sciences, 27(7):786-800.

Li S, Wu J, Gong J et al., 2018. Human footprint in Tibet: Assessing the spatial layout and effectiveness of nature reserves. Science of the Total Environment, 621:18-29.

Li S, Zhang Y, Wang Z et al., 2018. Mapping human influence intensity in the Tibetan Plateau for conservation of ecological service functions. Ecosystem Services, 30:276-286.

Longley P A, Adnan M, 2016. Geo-temporal Twitter demographics. International Journal of Geographical Information Science, 30(2):369-389.

Ma T, Lu R, Zhao N et al., 2018. An estimate of rural exodus in China using location-aware data. PLoS ONE, 13(7):e0201458.


Ma T, Pei T, Song C et al., 2019. Understanding geographical patterns of a city’s diurnal rhythm from aggregate data of location-aware services. Transactions in GIS, 23(1):104-117.

Mao X F, Wei X Y, Xia J X, 2012. Evaluation of ecological migrants’ adaptation to their new living area in Three-River Headwater wetlands, China. Procedia Environmental Sciences, 13(2011):1346-1353.

Senaratne H, Mobasheri A, Ali A L et al., 2017. A review of volunteered geographic information quality assessment methods. International Journal of Geographical Information Science, 31(1):139-167.

Stevens F R, Gaughan A E, Linard C et al., 2015. Disaggregating census data for population mapping using random forests with remotely-sensed and ancillary data. PLoS ONE, 10(2):e0107042.

Tan P N, Steinbach M, Kumar V, 2005. Introduction to Data Mining. Boston, USA: Pearson Education.

Walden-Schreiner C, Leung Y F, Tateosian L, 2018. Digital footprints: Incorporating crowdsourced geographic information for protected area management. Applied Geography, 90:44-54.

Watson J E M, Dudley N, Segan D B et al., 2014. The performance and potential of protected areas. Nature, 515(7525):67-73.


Weaver S D, Gahegan M, 2010. Constructing, visualizing, and analyzing a digital footprint. Geographical Review, 97(3):324-350.


Xing Y, Jiang Q G, Li W Q et al., 2009. Landscape spatial patterns changes of the wetland in Qinghai-Tibet Plateau. Ecology and Environment Sciences, 18(3):1010-1015.

Xu W, Pimm S L, Du A et al., 2019. Transforming protected area management in China. Trends in Ecology & Evolution, 34(9):762-766.


Xu W, Xiao Y, Zhang J et al., 2017. Strengthening protected areas for biodiversity and ecosystem services in China. Proceedings of the National Academy of Sciences, 114(7):1601-1606.

Xu X, Lu C, Shi X et al., 2008. World water tower: An atmospheric perspective. Geophysical Research Letters, 35(20):1-5.

Yan J, Wu Y, Zhang Y, 2011. Adaptation strategies to pasture degradation: Gap between government and local nomads in the eastern Tibetan Plateau. Journal of Geographical Sciences, 21(6):1112.

Yao Y, Liu X, Li X et al., 2017. Mapping fine-scale population distributions at the building level by integrating multisource geospatial big data. International Journal of Geographical Information Science, 31(6):1220-1244.

Yi J, Du Y, Liang F et al., 2019. Anomalies of dwellers’ collective geotagged behaviors in response to rainstorms: A case study of eight cities in China using Smartphone location data. Natural Hazards and Earth System Sciences, 19(10):2169-2182.

Yu C, Zhang Y, Claus H et al., 2012. Ecological and environmental issues faced by a developing Tibet. Environmental Science and Technology, 46(4):1979-1980.


Zhang Y, Wu X, Qi W et al., 2015. Characteristics and protection effectiveness of nature reserves on the Tibetan Plateau, China. Resources Science, 37(7):1455-1464. (in Chinese)

Zhao G, Liu J, Kuang W et al., 2015. Disturbance impacts of land use change on biodiversity conservation priority areas across China: 1990-2010. Journal of Geographical Sciences, 25(5):515-529.