The Warming and Wetting Ecological Environment Changes over the Qinghai-Tibetan Plateau and the Driving Effect of the Asian Summer Monsoon

Known as the"Roof of the World", the Qinghai-Tibet Plateau (TP) has an average elevation of more than 4, 000 m with high altitude, complex terrain, and unique weather and climate. Therefore, the TP is also compared with the South Pole and the North Pole and known as "the third pole" of the world (Qiu ^[1]). Being an important part of water cycle, precipitation over the TP not only affects the distribution of water resources in East Asia and South Asia, but also has a significant impact on the water cycle (Gao et al. ^[2]), permafrost evolution (Cheng and Jin ^[3]), desertification process (Zheng et al. ^[4]), and ecological environment of the Plateau itself (Shen et al. ^[5]). Under the background of global warming, especially in the past 20 years, the temperature of the TP has been increasing significantly, and its increase is much higher than the average temperature increase in China (CMA ^[6]). At the same time, the precipitation over the TP has also changed to some extent. The spatial distribution of the temperature rise is relatively uniform, while the precipitation and vegetation changes over the TP show obvious regional differences (Shen et al. ^[7]; Yu et al. ^[8]).

Some studies have shown that the vegetation productivity on the TP has increased significantly over the past 30 years (Li ^[9]; Ding et al. ^[10]). The greening period of plateau vegetation generally tends to become earlier (Piao et al. ^[11]). The area of vegetation improvement is larger than that of vegetation degradation (Han et al. ^[12]). The vegetation coverage has decreased significantly in the southern part of the Himalayas and the southern part of Qinghai Lake, followed by the central and southern parts of the Three-River Headwaters. The areas with no obvious vegetation change are mainly distributed in the northern Tibetan Plateau and the Qaidam Basin (Yu et al. ^[8]; Zhuo et al. ^[13]).

Temperature is one of the most important factors affecting vegetation growth. However, it is a great difficulty to explain the responses and feedback of the TP's alpine ecosystem to anomalous variation of precipitation. Some studies have also shown that there is a significant correlation between precipitation and vegetation growth (Shen et al. ^[14]; Liu et al. ^[15]; Zhang et al. ^[16]). The change of precipitation is the main factor regulating the change of ecological environment (Shen et al. ^[17]; Li et al. ^[18]; Wang et al. ^[19]). With the background of the overall warming of the TP, the acquisition of accurate precipitation data with high spatial and temporal resolution is an important basis not only for study on the regional differences of vegetation change on the TP, but also for reliable data support for exploring the response of vegetation to climate change.

At present, the most important and direct way to obtain precipitation data is the observation from ground rain gauges. Most studies on precipitation over the TP are based on rain gauge data (Yang et al. ^[20]; Zhang et al. ^[21]). While compared with satellite data, data from ground stations cannot offer a wider observation range and a smaller space interval, so it cannot uniformly represent the precipitation on the ground. The Tropical Rainfall Measuring Mission (TRMM) has provided many widely used precipitation products with high spatial and temporal resolution (Kummerow et al. ^[22]; Huffman et al. ^[23]; Huffman et al. ^[24]). Many studies have been carried out to validate TRMM-based products, especially TRMM-3B42 and TRMM-3B43 products, against gauge data at different geographic regions and different times in order to assess their ability for describing precipitation pattern at un-gauged areas. However, due to the complexity of the terrain and the limitation of technical means, the accuracy of satellite data in the high-latitude areas with complex terrain often needs to be further improved (Fang et al. ^[25]; Yong et al. ^[26]; Chen et al. ^[27]; Ballari et al. ^[28]; As-Syakur et al. ^[29]; Tarek et al. ^[30]).

Known as the"water tower"of Asia, the TP plays an important role in the water vapor transport of the Asian monsoon, and a pivotal role in regulating and changing climate change. Aiming at uncovering how temperature and precipitation facilitate the change of vegetation environment in the main body of the TP, this paper firstly combines both the observation precipitation data from ground stations and the products of satellite effectively through the variational method. Based on the new products with high precision and high spatial and temporal resolution, distributions and changes of precipitation in the TP can be better described. Then, it can be used to explain the reason why there are significant regional differences in such environmental changes of vegetation. Moreover, the research is intended to investigate, among the driving factors of warming and wetting in the Plateau, how monsoon influences the heterogeneous regional variation characteristics of ecological environment through the change of water vapor transport structure.

(1) Satellite-based TRMM precipitation

The TRMM satellite is a remote sensing satellite jointly developed by the National Aeronautics and Space Administration (NASA) and the Japan Aerospace Exploration Agency (JAXA). The precipitation radar (PR) sensor carried by TRMM, which is the world's first satellite-borne rain radar, can obtain quasi-global precipitation data with high temporal and spatial resolution (Huffman et al. ^[23]; Huffman et al. ^[24]). The TRMM 3B43 used in this study is a monthly precipitation product with a time resolution of one month and a time span from January 1998 to February 2019. The space coverage is global 50° S-50° N and the spatial resolution is 0.25° × 0.25°.

(2) MODIS Leaf Area Index (LAI) products

The Leaf Area Index (LAI), defined as half the total leaf area per unit ground surface area (Chen and Black ^[31]), can reflect the characteristics of regional plant growth and has been an important indicator in ecological and environmental changes. The Moderate Resolution Imaging Spectroradiometer (MODIS) Leaf Area Index (LAI) products (MOD15A2H) are the products with the longest time series and the best completeness so far. They are inversed based on a canopy radiation transfer model and a lookup table using MODIS daily surface reflectance products (MOD09), solar zenith angle, sensor observation angle, and land cover type (MOD12) (Knyazikhin et al. ^[32].) In this study, MOD15A2H with 8-day revisiting cycle and 500 m spatial resolution from 2000 to 2019 in Collection 6 is downloaded via the Earth Observing System Data and Information System (EOSDIS) of National Aeronautics and Space Administration (NASA) (https://earthdata.nasa.gov/). Monthly mean LAI is calculated with the products during each month from May to September, and they are further averaged to get the average values during vegetation growing season.

(3) Ground-based meteorological data

The variational processing method integrating remote sensing and ground observation is adopted in this paper to conduct variational correction of the TRMM satellite products. For this correction, we use ground rain gauge data from 2, 425 stations of the National Meteorological Information Center, China Meteorological Administration, with missing data and abnormal values eliminated.

The t2m data is from the"Dataset of monthly surface meteorological elements in China"released by the National Meteorological Information Center, China Meteorological Administration. The data of 2, 479 stations in the dataset from 1951 to 2013 underwent strict quality control in the development of ground basic meteorological data and in the production of datasets.

Aiming at overcoming the limitations of the observational data of ground stations and the precipitation products of TRMM satellite, this paper adopts the variational method to integrate the two kinds of data. Hence, the comprehensive variational correction processing technology, integrating the satellite remote sensing data and ground observation data, is used to modify TRMM satellite products (Xu ^[33]; Weng and Xu ^[34]). According to the variational principle, the functional equation depends on multiple independent variables.

In this equation, U(x, y) must satisfy the following Euler equation:

The precipitation field of the satellite data is set as R (x, y), and the corresponding rainfall gauge data field at the finite point is R(I, J). Then on the station coordinate (I, J), the difference field of the above two fields is the error field:

In fact, owing to the limited number of observation sites, it is necessary to construct a more generalized correction factor field function C_r (x, y) to meet the following conditions:

That is, $\sum\limits_{i} \sum\limits_{j}\left(C_{r}-C_{r}^{\sim}\right)^{2} $ reaches a minimum. A new factor field C_r (x, y) through variational correction is obtained, and the factor field after the variational revision is finally achieved:

This research program has achieved significant results in the objective correction of satellite remote sensing reanalysis on several aspects including satellite remote sensing retrieval of urban heat islands, the atmospheric element field in dust storm processes, and Threading Building Blocks (TBB) data under clear sky conditions (Cheng et al. ^[35]; Cheng et al. ^[36]; Bai et al. ^[37]).

To evaluate the performance of TRMM 3B43 products and the verified products, grid data are compared with ground rain gauge data using the nearest grid-to-point matching approach. Several statistical indicators are used to quantify the consistency between satellite precipitation and station data.

(1) Root mean square error (RMSE):

where n is the total number of samples; x_i and yi represent gauge observations and the corresponding values of the satellite, respectively. RMSE can be used to measure the deviation between the satellite precipitation and the observed data from the stations. The smaller the absolute value is, the closer the satellite precipitation is to the observed data.

(2) Correlation coefficient (r):

The correlation coefficient is an exponential representing the degree of linear correlation between two variables. As for its absolute value, the closer to 1, the stronger the linear correlation between two groups of numbers will be. r greater than 0 represents positive correlation, while r less than 0 represents negative correlation.

(3) Excluding method

Excluding method is also adopted to further examine the ability of TRMM and verified TRMM precipitation products. To ensure sample independence during the testing, part of the station data are excluded before the revision. Since they are excluded in the correction, they can be compared with the original as well as the revised TRMM products independently.

The entire-layer water vapor flux (Q) is calculated using the following Equation:

Q can be divided into two directions:

where g is acceleration due to gravity, u/v is the zonal/meridional wind, q is specific humidity, Ps is the surface atmospheric pressure, Pt is the top atmospheric pressure and is set to be 300hPa, and qu and qv are the zonal and meridional water vapor fluxes, respectively.

As a result of the high altitude, complex terrain, harsh climate, and sparse population, the TP has a severe shortage of weather stations compared with central and eastern China. The spatial distribution of meteorological observation stations on the TP is also uneven. Most of the meteorological stations are located in the east and south of the Plateau with a minority of the stations in the central and northwest parts of the Plateau. There is a large area of observation"blank area"in the central and northwestern TP (Fig. 1a). It spans nearly 10° from south to north and 15° from east to west, accounting for nearly 1/3 of the plateau area. The minority stations in the "blank area" fail to represent the precipitation characteristics of the overall central and western part of the Plateau. As a result, it is difficult to analyze the precipitation in the plateau region because of the complexity of the terrain and the limitations of precipitation observation.

Figure 1.  (a) The distribution of regional stations on the Tibetan Plateau (black dots) and interpolation supplementary stations (red dots); (b) spatial distribution of the Root-mean-square error (units: mm d^-1) over the TP; (c) locations of 6 sites that have been eliminated; (d) RMSE between ground rain gauge data and TRMM precipitation products before (blue columns) and after variational revision (red columns) and optimized variational revision (green columns); (e) locations of 7 sites in Nepal, where the red dots are the eliminated sites; (f) RMSE between ground rain gauge data and TRMM precipitation products before (blue columns) and after variational revision (red columns) and optimized variational revision (green columns).

Compared with the precipitation data of the stations, the satellite data has wide observation range with a smaller space interval, which can uniformly represent the precipitation on the ground. However, due to the complexity of the terrain and the limitation of technical means, the accuracy of satellite data in the high latitude areas with complex terrain often needs to be further improved. Fig. 1b shows the annual RMSE between ground rain gauges and TRMM products during 1998-2018. The spatial distribution of RMSE in the four seasons is consistent, showing the feature of gradual decrease from southeast to northwest. There are three areas with large errors, namely the Yunnan-Guizhou Plateau area south of the Hengduan Mountains on the southeast side of the Qinghai-Tibet Plateau, the southeast Tibet area near the bend of the YarlungZangbo River, and the area near the Himalayas. From the perspective of seasonal distribution, the RMSE shows the largest value in summer, which is about 1~2mm d^-1 in most stations but could reach more than 5mm d^-1 in individual sites in the Himalayan Mountains, Qaidam Basin and other places. In spring and autumn, the RMSE of most stations is about 0.1~1mm d^-1. It shows the smallest error in winter, with 0.5 mm d^-1 and below in most sites. Only a few stations are characterized with RMSE reaching 1mm d^-1.

The sparsely distributed observation stations in the central and western regions of China have resulted in an observation"blank area", where the correction cannot be conducted effectively. To solve this problem, we try to interpolate the observation data of the stations on the grid based on the Cressmen style analysis (every three degrees, shown in Fig. 1a). That is to say, the interpolation results at the grid points (red dots) are considered as supplementary rainfall data so that the site values are distributed in this area evenly. These data are then used for variational correction with TRMM precipitation products.

In order to further verify the revised results, the elimination method is adopted to check the accuracy of TRMM precipitation products after correction. Six stations (Fig. 1c) with RMSE greater than 3 are excluded when the variational correction is adopted. Therefore, these 6 sites are independent samples relative to the revised results. The RMSE between TRMM products before and after correction and the true values of the six stations are shown in Fig. 1d. The RMSE of most stations after correction is slightly lower than that before correction, which means the stations do not directly participate in the process of correction, and their errors can still be reduced to a certain extent. From the perspective of correlation, the correlation coefficient is only 0.22 between original TRMM precipitation data and station precipitation of the six stations in Fig. 1c during the summer time (June, July and August) from 1998 to 2018. However, after correction, the correlation coefficient between the two increases to 0.74, indicating a stronger linear correlation. The correlation coefficient of the optimized variational revision reaches 0.82, also showing the higher quality of the optimized revised TRMM precipitation products.

Among the above stations, site sta 4 (Fig. 1c) is located in the"blank area"of the TP, with only a few sites surrounding it. After variational correction and optimized variational correction, the RMSE of the site in summer decreases from 4.62 to 1.63 and 1.33 respectively. In other words, with the use of the interpolation variational correction method, TRMM satellite precipitation variational products can objectively supplement the"blank area"of precipitation observation stations on the Qinghai-Tibet Plateau, and therefore effectively improve the accuracy of TRMM precipitation products.

The same method is used to correct TRMM precipitation products in Nepal, south of the Tibetan Plateau (Fig. 1e and 1f). After the exclusion of three different stations, the performance of TRMM precipitation products at these three stations before and after variational correction is compared and analyzed. Since the interpolation optimization effect concentrates mainly in the central and western plateau, it will have little impact on the Nepal site. As a result, the RMSE in summer from 1998 to 2018 after correction is slightly lower than that before correction. However, the RMSE of TRMM precipitation products after optimized variational correction is the smallest and the revised TRMM precipitation products are closer to the real values.

In summary, it can be concluded that the corrected TRMM precipitation products are closer to the real values; therefore, the variational method has a certain universal significance in revising satellite data. We have combined the station precipitation data with TRMM precipitation products by the optimized variational method to make the corrected TRMM precipitation products with uniform distribution, which are closer to the real values. The following research is based on the corrected TRMM precipitation products.

The Leaf Area Index (LAI), defined as half the total leaf area per unit ground surface area (Chen and Black ^[31]), can reflect characteristics of regional plant growth and has been an important indicator in ecological and environmental changes. Climate warming has improved the vegetation coverage of the TP (Fig. 2d), promoted the productivity of the Plateau vegetation, and advanced spring vegetation phenology (Ding et al. ^[10]). Affected by altitude and latitude, the vegetation on the TP grows mainly from May to September. Therefore, the growing season from May to September was took as the key period of warming and humidification to explore the impact of temperature and precipitation changes on vegetation environmental changes.

Figure 2.  The spatial distribution of the trend of (a) precipitation (stations, units: mm d^-1year^-1), (b) air temperature (units: ℃ 10year^-1), (c) precipitation (verified TRMM, units: mm 10year^-1) and (d) LAI (units: 10year^-1) over the Tibetan Plateau (25°-40°N, 70°-105°E) in growing season from 2000 to 2018.

The temperature of the TP has shown a uniform and significant warming trend in the past 20 years. The change rates of temperature and precipitation in the vegetation growing seasons from 2000 to 2018, obtained from the limited ground-based weather stations, indicate that the characteristics of precipitation change are significant regional differences compared with the uniformly increasing temperature changes over the whole Plateau (Fig. 2a and 2b). Based on the variational correction of TRMM satellite with precipitation station data, the distribution of precipitation variability in the high spatial resolution is obtained (Fig. 2c). It can be found that the spatial distribution of precipitation variability is significantly different from the north to the south, which can be divided into two regions: the northern region A and the southern region B. Precipitation in the southern part of the plateau is mainly decreasing, where LAI is decreasing in approximately 13.9% of this area. Precipitation increases in northeast and northwest TP, but decreases in central TP. As a result, only 3% of pixels these area shows a downward trend in LAI.

With the lagging effects of precipitation on vegetation change over the TP considered (Cong et al. ^[38]; Shen et al. ^[39]; Shen et al. ^[40]), the spatial distribution of monthly precipitation and LAI trend is analyzed and compared. Fig. 3 describes the interannual variation of precipitation and LAI in different months from 2000 to 2018. The distribution characteristics of the change graphs of the latter lagging one month behind the former are very consistent. That is to say, there is a one-month lagging effect on the distribution of precipitation and LAI changes. Among them, the precipitation in June and July corresponds to the LAI in July and August, which respectively account for the largest proportion of downtrend and uptrend in the Plateau region, namely, the months with the most significant effects of vegetation degradation and warming and wetting. The correlation between the trends of precipitation and LAI in the TP shows that the change of vegetation has a one-month lagging effect on precipitation.

Figure 3.  Distribution of monthly mean precipitation change rate over the Tibetan Plateau (25°- 40°N, 70°- 105°E) from April to August (a1-e1) during 2000-2018 (units: mm 10year^-1); monthly mean LAI change rate from May to September (a2-e2) during 2000-2018.

In order to discuss the effect and contribution of temperature and precipitation on vegetation, the changes of air temperature in the northern region (A) and southern region (B) of the Plateau are synchronized with the interannual variation of LAI (Fig. 4a and 4c). In general, both temperature and LAI in region A and B show a significant upward trend. The interannual correlation coefficients between monthly LAI and temperature in the two regions are 0.535 and 0.201, respectively, reaching a confidence level of 95% (Fig. 5a and 5c). Considering the response time of plateau ecological change to precipitation, the variations of precipitation and LAI in region A have correlation with one-month lag (Fig. 4b). The 1-month lagging correlation coefficient of LAI with precipitation in the current month is 0.217, reaching the confidence of 95% (Fig. 5b). However, the inter-annual variations of precipitation and LAI in region B show a slightly reverse trend (Fig. 4d), and the lag correlation coefficient does not reach the reliability standard (Fig. 5d), demonstrating the complex response of ecological change to precipitation in the Plateau.

Figure 4.  Normalized variation curves of LAI (green) and temperature (black) in regions A (a) and B (c) from May to September, 2000-2018; the standardized variation curves of precipitation (blue) from April to August and LAI (green) from May to September in regions A (b) and B (d) from 2000 to 2018. The r on the top of each subgraph represents correlation coefficient between LAI and temperature or precipitation; r1 and r2 on the bottom represent correlation coefficient between each variable and time, which indicates the significance of the trends.

Figure 5.  The correlation between standardized LAI and temperature in Regions A (a) and B (c) from May to September, 2000-2018; the correlation between standardized precipitation from April to August and LAI from May to September in regions A (b) and B (d) from 2000 to 2018.

LAI correlation coefficient distribution of each pixel on the plateau is analyzed (Fig. 6). The temperature response of LAI is positive in approximately 88.3% of the TP area, with 51.5% being significant (with more than 90% confidence), especially in the central, eastern, and northeastern parts (areas without LAI are excluded). In the northern part of the Plateau (region A), the correlation of vegetation to temperature response is higher, and the positive correlation between LAI and temperature response is in more than 93.2% of this area. From the perspective of the impact of precipitation, LAI is also positively correlated with precipitation before rainy season on the TP (Fig. 6b). In region A, about 69.7% of the pixels shows positive correlation between LAI and precipitation, with 27.0% being significant. The positive correlation is mainly concentrated in the northeast of the Qinghai-Tibet Plateau, including Qinghai Lake and its surrounding areas. Compared with Fig. 2c, it can be seen that the precipitation in this region presents a significant upward trend, so the LAI in this region increases and presents a significant trend of warming and wetting. In region B, 52.5% of the area shows positive correlation, 11.3% being significant. Only about 8.1% of the pixels exhibits significantly negative correlations. Generally speaking, although the response is not very sensitive, the LAI in most regions is positively correlated with the change of precipitation.

Figure 6.  Correlation distribution of standardized LAI with (a) temperature in growing season (May to September) and (b) precipitation in preseason (April to August) of 2000-2018 (Dots indicate that the correlation coefficient reaches the confidence level of 90%); (c) the frequency distributions of correlation coefficient corresponding to (a); (d) the same as (c) but corresponding to (b).

Previous studies have shown that the starting date of vegetation growing season is more sensitive to the inter-annual variations of precipitation before rainy season in more arid areas than in wetter areas (Shen et al. ^[7]). As the result, vegetation change is not sensitive to the response of precipitation in the southern area of the TP (region B) which is a humid region with abundant precipitation. While in the northern TP (region A), vegetation is more sensitive to precipitation before rainy reason, which plays a vital role in vegetation growth.

These results indicate that the overall trend of vegetation growth change in the southern and northern parts of the TP is improving under the background of climate warming, but it is also modulated by the regional difference of precipitation change trend. The monthly LAI variation patterns in the growing season were consistent with that of monthly precipitation variation before the rainy reason of the growing season. Moreover, the increasing precipitation further contributed to the growing of vegetation in northern part of the TP (region A). Under the fact that the increasing of temperature over the TP is two times higher than the global average, the strong heterogeneous distribution of precipitation over the TP also has a significant effect on vegetation status in this region.

Asian monsoon and westerly wind are closely related to precipitation and water vapor transport over the TP (Xu et al. ^[41]; Xu et al. ^[42]; Xu et al. ^[43]). Due to the lagging effect of precipitation on vegetation, this paper focuses on the changes of precipitation and water vapor transport before the rainy season of the growing season, from April to August. The correlation between precipitation before the rainy reason and whole-layer water vapor transport in regions A and B of the TP (Fig. 7a/b) can describe the synergistic influence of water vapor transport from Asian monsoon and westerly wind on precipitation. The South Asian monsoon water vapor mainly come from the Indian Ocean, the Arabian Sea, and the Bay of Bengal, and is closely related to the trans-equatorial warm and humid air currents of the southern hemisphere ocean water vapor sources in the southern Indian Ocean. The precipitation in region A is mainly influenced by the south Asian monsoon and westerly winds (Fig. 7a). The water vapor in region B is mainly from the Bay of Bengal from the south of the Plateau, influenced by the East Asian monsoon system. The water vapor source comes from the South China Sea, the Bay of Bengal, and the Arabian Sea and is controlled by the subtropical high. Is there an interannual variation of water vapor transport structure over the Tibetan Plateau in South and East Asia during 1998-2018? The correlation distribution of water vapor flux and precipitation at the southern boundary of region A and region B before the rainy season in 1998-2018 are shown as Fig. 7c and 7d. It can be found that the precipitation changes in the north and south regions are closely related to the main water vapor flow on their southern boundaries respectively. Therefore, the southern boundary of the two regions is selected as the key component of water vapor transport in this region. The water vapor flux qv at the southern boundary of the two regions shows an opposite trend (Fig. 7e and 7f). Influenced by the increasing trend of inter-annual variation of wind field in the South Asian monsoon at 500hPa (Fig. 7g), qv on the southern boundary of region A is high at both ends and slightly low in the middle of water vapor transport with uptrend (Fig. 7e). On the other hand, the water vapor transport in southern region B at 700hPa shows a weakening trend, which shows uptrend at both ends, with downtrend in the middle (Fig. 7f). That is to say, the upper-level water vapor transport affecting precipitation in region A has been strengthened in the past two decades, while the weakening of water vapor transport at low level leads to the decline of precipitation in region B.

Figure 7.  (a) The correlation coefficient distribution of precipitation in TP region A and global water vapor flux averaged in preseason of 2000-2018 (The shadow represents the value of the correlation coefficient); (b) the same as (a) but in region B; (c) the correlation distribution between water vapor flux qv at the southern boundary of region A and precipitation; (d) the same as (c) but of region B; (e) the trend of water vapor flux qv averaged in preseason at the southern boundary of region A at 500hPa from 2000 to 2018; (f) the same as (e) but of region B at 700hPa; (g) the trend of wind field on 500hPa averaged in preseason from 2000 to 2018 (Vector: wind field variability; color: v_wind variable).

In order to further trace the source and transport path of water vapor in upper and lower levels and analyze their changes, Fig. 7g and 7h shows the changes of wind field at 500hPa and 700hPa. Originating from the trans-equatorial flow in the southern hemisphere, water vapor influenced by the South Asian monsoon on 500 hPa enters the TP from the west boundary and reaches the northern and central parts of the Plateau. It shows an increasing trend, which provides an important guarantee for the atmospheric circulation water vapor transport to promote the warming and wetting effect of the vegetation environment on the TP. On the other hand, water vapor from the Bay of Bengal, which climbs into the southern edge of the eastern TP on 700 hPa, shows a weakening characteristic. It constitutes the background of the circulation field change of the precipitation reduction, and inhibits the environmental warming and wetting process in the southern plateau to some extent.

Above all, it can be found that the warming and wetting of the TP promotes the changes of the vegetation environment, but there are regional differences in the northern and southern parts of the TP. Apart from temperature rise, which is one of the main causes, these regional differences are significantly correlated with the regional characteristics of spatial and temporal distribution of precipitation over the TP. Central South Asia and East Asia monsoon, the driving factors of TP warming and wetting, also affect the regional heterogeneity of ecological environment through the significant changes of interannual and interdecadal water vapor transport structure.

With the optimizing interpolation variational correction method, the TRMM satellite precipitation variational products can be used not only as the objective supplements of the"blank area" of the TP precipitation observation stations, but also to improve the accuracy of TRMM precipitation products. The TRMM precipitation products can be applied to describing the high-resolution precipitation distribution characteristics, and to integrating the actual ground observation data. After correction, the product can describe the spatial and temporal distribution characteristics of precipitation over the TP in a more objective and refined manner.

There is a resemblance pattern between the variation of vegetation growth and precipitation in the southern and northern regions of the TP in recent two decades. Moreover, a one-month lagging correlation effect is identified between precipitation and LAI variability distribution.

The South Asian monsoon originated from the trans-equatorial airflow in the southern hemisphere is strengthening. Its intensification and extension to the northern and central parts of the TP provide the key driving background conditions for the atmospheric circulation to promote the warming and wetting water vapor transport of the vegetation environment. The warming and wetting of the TP promotes the changes of the vegetation environment, but there are regional differences in the changes of the vegetation environment in the northern and southern parts of the TP. In addition to the main cause of temperature rise, these regional differences are correlated with the regional characteristics of spatial and temporal distribution of precipitation over the TP. The South Asian and East Asian monsoons, as the major driving factors of the warming and wetness of the TP, also influence the regional variation of the ecological environment of the TP through the inter-annual and inter-decadal changes of water vapor transport structure.

Data Availability Statement: ERA-Interim (ECMWF, https://www.ecmwf.int) reanalysis monthly data are part of the European Center for Medium-range Weather Forecasts. The TRMM science data (https://disc.gsfc.nasa.gov/datasets/TRMM_3B43_7/summary?keywords=TRMM) are provided by the NASA and the National Space Development Agency of Japan. The t2m and ground rain gauge data of China are released by the National Meteorological Information Center (http://cdc.cma.gov.cn), China Meteorological Administration. All the original datasets and code needed to reproduce the results shown in this paper are available upon request. MOD15A2H with 8-day revisiting cycle and 500 m spatial resolution from 2000 to 2019 in Collection 6 is downloaded via the Earth Observing System Data and Information System (EOSDIS) of National Aeronautics and Space Administration (NASA) (https://earthdata.nasa.gov/).