Change in Alpine Grassland NPP in Response to Climate Variation and Human Activities in the Yellow River Source Zone from 2000 to 2020

Abstract

Identifying the relative contributions of climate change and human activities to alpine grassland dynamics is critical for understanding grassland degradation mechanisms. In this study, first, the actual NPP (NPP_a) was obtained by MOD17A3. Second, we used the Zhou Guangsheng model to simulate the potential met net primary productivity (NPP_p). Finally, the NPP generated by anthropogenic activities (NPP_h) was estimated by calculating the difference between NPP_p and NPP_a. Then, the relative contributions of climate change and human activities to NPP changes in grasslands were quantitatively assessed by analyzing trends in NPP_p and NPP_a. Thereby, the drivers of NPP change in the Yellow River source grassland were identified. The results showed that the temperature and precipitation in the study area showed a warm-humid climate trend from 2000 to 2020. The NPP_p and NPP_a increased at a rate of 1.07 g C/m² and 1.51 g C/m² per year, respectively, while the NPP_h decreased at a rate of 0.46 g C/m² per year. It can be seen that human activities had a positive effect on the change of NPP in the Yellow River source grassland from the change rate. The relative contribution analysis showed that 55.90% of grassland NPP increased due to climate change, 40.16% of grassland NPP increased due to human activities, and the grassland degradation was not significant. The research results can provide a theoretical basis and technical support for the next step of the Yellow River source grassland ecological protection project.

1. Introduction

Alpine grassland is the most important ecosystem in the Yellow River source zone, accounting for about 80% of the region. As a unique grassland ecosystem in Northwest China, alpine grassland is the most abundant and complete biological gene bank and the largest plateau germplasm bank in the Qinghai-Tibet Plateau and is also an important animal husbandry production base and ecological security barrier in China [1,2,3]. However, because of the high altitude and the complex area of cold terrain, the study area is densely covered with rivers, lakes, marshes, and snow-capped mountains, highly sensitive to environmental changes [4], and is one of the extremely fragile ecological environments in China. In recent decades, the vegetation coverage, water conservation function, runoff, and soil freezing of the Yellow River source have all changed significantly, which directly reduced the stability of the grassland ecosystem. The destruction of the grassland ecosystem will further cause environmental degradation problems, including soil erosion, land desertification, grassland degradation, glaciers melting, rat infestations, etc. The unique sensitive environmental characteristics of the Yellow River source are difficult to restore once they degrade, threatening the water resources and ecology of the middle and lower reaches of the Yellow River, as well as security and regional sustainability [5,6]. Therefore, the condition of alpine grassland in the Yellow River source reflects the ecological health of the area. The study of the temporal and spatial variation characteristics of grassland under the influences of multiple factors such as climate change and human activities in the Yellow River source is helpful for protecting the ecological health of the area. The mechanism of grassland productivity change is more complicated due to the mutual interference and interaction between climate change and human activities. Therefore, quantitative evaluation of the impact of climate change and human activities on grassland productivity change has gradually become a hot spot in the study of grassland productivity change.

Climate is the main factor that affects the temporal and spatial distribution of grasslands [7]. The reason is that the growth environment and growth of grasslands are determined by precipitation and air temperature, and the changes in precipitation and air temperature play a vital role in the change in plant productivity of grasslands all over the world [8]. However, with the increase in population, the influence of human activities such as overgrazing, on grassland becomes more and more obvious [9,10,11,12]. So far, there have been a lot of studies on grassland NPP as an indicator to measure the impact of climate change and human activities on grassland ecosystems. The methods used are mainly divided into two categories: residual trend method and model-based anthropogenic NPP allocation method. The residual trend method was used to estimate the contribution of each factor to the NPP change for each pixel by constructing a function of grassland NPP change with different climatic factors and human activities [13]. Whereas the model-based anthropogenic NPP allocation method used climate-driven models to simulate grassland NPP_p, and used remote sensing models to simulate the NPP_a, with an artificially generated NPP defined as the difference between the NPP_p and NPP_a [14]. However, the residual trend method to quantitatively evaluate the impact of climate change and human activities on grassland productivity is affected by data sources, noise and accuracy, but it still has obvious shortcomings [15]. For example, Luo et al. used the residual trend method to analyze the correlation between net primary productivity and climate change and human activities in Qinghai-Tibet Plateau from 2001 to 2015 [13]; Wei et al. simulated climate and NPP_a with the Zhou Guangsheng model and CAS A model, respectively, and discussed grassland dynamics and its driving factors in Qinghai Province from 2001 to 2016 [14]. Based on the water use efficiency of vegetation determined by the ratio of carbon dioxide flux equation (equivalent to NPP) to water vapor flux equation (equivalent to evapotranspiration) on the vegetation surface, Zhou Guangsheng and Zhang Xinshi deduced a regional evapotranspiration model linking energy balance equations and water balance equations according to two well-recognized water balance equations and heat balance equations. Meanwhile, the net primary productivity model of natural vegetation is based on plant physiological and ecological characteristics and obtained a model suitable for calculating potential net primary productivity according to the precipitation in China and the net radiation data obtained, which can better reflect the vegetation characteristics in China [16,17,18].

Therefore, based on the research of Feng and Zhao [19,20], this paper takes the alpine grassland at the Yellow River source as the research object, and used the Zhou Guangsheng model and MOD17A3 to simulate the NPP_p and NPP_a of the alpine grassland from 2000 to 2020. Moreover, we estimated the NPP generated by human activities through the difference between the NPP_p and NPP_a and compared the trend of the NPP_p and NPP_a changes with time. Finally, we clarified the contribution of climate change and human activities to the change of NPP in the alpine grassland. The research results provide a theoretical basis for the sustainable utilization of alpine grassland resources and the rational layout of animal husbandry at the source of the Yellow River.

2. Materials and Methods

2.1. Study Area

The Yellow River source zone is located in the southeast of Qinghai Province, between 33°0′ N–36°2′ N, 96°2′ E–102°2′ E (Figure 1). The length of the source part of the Yellow River is 1273 km, accounting for approximately 23.28% of the total length of the Yellow River [21]. The Yellow River source zone belongs to Monsoon-Influenced Sub-Arctic Climate and Tundra Climate in the Köppen climate classification. The study area has more rainfalls from July to September each year under the influence of the southwest monsoon, and from October to December, it is gradually dry and cold under the influence of the cold wave high on the Qinghai-Tibet Plateau [22]. The topography of the study area is undulating, and it is generally high in the west and low in the east. The altitude is between 2630–6009 m, and the average altitude is above 4000 m. According to the GlobeLand30 dataset, the main land use type in the study area in 2020 was grassland (Figure 2), accounting for 89.75% of the study area. The grassland types mainly include alpine meadow, alpine grassland, and swamp meadow (Figure 3).

2.2. Data and Processing

The remote sensing image data used in this study were GlobeLand30 and MOD17A3HGF.006. The GlobeLand30 dataset was derived from the 30 m global land cover data released by the Ministry of Natural Resources of China (http://www.globallandcover.com (accessed on 18 January 2022)). The remote sensing images were mainly multispectral images, including Landsat TM, ETM+, OLI, and China Environmental Disaster Mitigation Satellite (HJ-1) images with a spatial resolution of 30 m, andGaofen-1 (GF-1) image with 16 m spatial resolution. The data format was GeoTiff [23,24]. The classification system of the GlobeLand30 dataset includes 10 land types: cultivated land, woodland, grassland, shrub land, wetland, water body, tundra, artificial surface, bare land, glaciers, and permanent snow cover. The land types in the study area were reclassified into six categories: cultivated land, forest land, grassland, water body, artificial surface, and unused land, and then the grassland was spatially analyzed. MOD17A3HGF.006 is a MODIS image with a spatial resolution of 500 m. The NPP is obtained by calculating the net photosynthesis (PSN) every eight days. Net photosynthesis is the difference between the total primary productivity and the maintenance respiration. This research relied on the Google Earth Engine (GEE) platform to calculate the annual NPP data from 2000 to 2020 by calculating the annual total as the final coverage of the study area. The meteorological data from 2000 to 2020 used in this study are derive from the National Earth System Science Data Center, National Science and Technology Infrastructure of China (http://www.geodata.cn (accessed on 28 November 2021)), with a spatial resolution of 1 km [25].

2.3. Method

2.3.1. Meteorological Data Processing Methods

The Mann-Kendall method is a non-parametric statistical test method that has been widely used to study the time series of climate and hydrological elements such as temperature, precipitation, and runoff, as well as the abrupt changes and trends of raster-scale spatiotemporal data [26,27]:

$${\mathrm{Z}}_{\mathrm{s}}=\left\{\begin{array}{c}\frac{\mathrm{S}-1}{\sqrt{\mathrm{Var}\left(\mathrm{S}\right)}},\mathrm{i}\mathrm{f} \mathrm{S}>0\\ 0,\mathrm{i}\mathrm{f} \mathrm{S}=0\\ \frac{\mathrm{S}+1}{\sqrt{\mathrm{Var}\left(\mathrm{S}\right)}},\mathrm{i}\mathrm{f} \mathrm{S}<0\end{array}\right.,$$

(1)

$$\mathrm{S}={\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{n}-1}{\displaystyle \sum }_{\mathrm{j}=\mathrm{i}+1}^{\mathrm{n}}\mathrm{sgn}\left({\mathrm{X}}_{\mathrm{j}}-{\mathrm{X}}_{\mathrm{I}}\right),$$

(2)

$$\mathrm{Var}\left(\mathrm{S}\right)=\frac{\mathrm{n}\left(\mathrm{n}-1\right)\left(2\mathrm{n}+5\right)}{18},$$

(3)

$$\mathrm{sgn}\left({\mathrm{X}}_{\mathrm{j}}-{\mathrm{X}}_{\mathrm{i}}\right)=\left\{\begin{array}{c}1,{\mathrm{X}}_{\mathrm{j}}-{\mathrm{X}}_{\mathrm{i}}>0\\ 0,{\mathrm{X}}_{\mathrm{j}}-{\mathrm{X}}_{\mathrm{i}}=0\\ -1,{\mathrm{X}}_{\mathrm{j}}-{\mathrm{X}}_{\mathrm{i}}<0\end{array}\right.,$$

(4)

where is the length of the time series, i is 2000, 2001, …, 2020. Under a given significance level α, when |Zs| > u_1−α/2, the time series data of the study are at the α level. Significant changes are generally taken as α = 0.05. When |Zs| > 1.96, the time series has a significance α < 0.05, and |Zs| < 1.96 means significance α > 0.05.

2.3.2. Grassland NPP Treatment Methods

In this study, the NPP_p of grassland was calculated by introducing the Zhou Guangsheng driving model. The NPP_h is derived from the difference between the NPP_p and NPP_a. So as to quantitatively evaluate the relative contribution of climate change and human activities to the change of grassland NPP. The specific algorithm of the Zhou Guangsheng model is as follows [18,20]:

$${\mathrm{NPP}}_{\mathrm{p}}={\mathrm{RDI}}^2\times \frac{\mathrm{P}\times \left(1+\mathrm{RDI}+{\mathrm{RDI}}^2\right)}{\left(1+\mathrm{RDI}\right)\left(1+{\mathrm{RDI}}^2\right)}\times \mathrm{Exp}\left(-\sqrt{9.87+6.25\mathrm{RDI}}\right),$$

(5)

$$\mathrm{RDI}={\left(0.629+0.237\mathrm{PER}-0.00313{\mathrm{PER}}^2\right)}^2,$$

(6)

$$\mathrm{PER}=\frac{\mathrm{BT}\times 58.93}{\mathrm{P}},$$

(7)

$$\mathrm{BT}=\frac{\sum \mathrm{T}}{12},$$

(8)

where NPP_p is the potential net primary productivity and RDI is the radiation index of dryness. P is the annual precipitation (mm), PER is the potential evapotranspiration rate, BT is the annual average temperature (°C), and T is the monthly mean temperature (°C).

The specific algorithm of NPP affected by human activities is as follows:

$${\mathrm{NPP}}_{\mathrm{h}}={\mathrm{NPP}}_{\mathrm{p}}-{\mathrm{NPP}}_{\mathrm{a}},$$

(9)

where NPP_h is the NPP generated by anthropogenic activities, NPP_p is the potential net primary productivity, and NPP_a is the actual NPP.

After obtaining the annual NPP_p, NPP_a, and NPP_h of grassland in the study area from 2000 to 2020, the temporal and spatial variation trend of grassland NPP in the study area from 2000 to 2020 was analyzed by means of linear regression. The specific algorithm is as follows [14]:

$$\mathrm{k}=\frac{\mathrm{n}\times {{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}\mathrm{i}\times {\mathrm{NPP}}_{\mathrm{i}}-{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}\mathrm{i}\times {{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{NPP}}_{\mathrm{i}}}{\mathrm{n}\times {{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{i}}^2-{\left({{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}\mathrm{i}\right)}^2},$$

(10)

where k is the change trend of NPP, k > 0 means NPP increases, k < 0 means NPP decreases, n = 21, i = 1, 2, 3, …, 21, NPP_i refers to the annual average NPP in year i.

According to the changing trends of NPP_p, NPP_a, and NPP_h, four causes for NPP changes are hypothesized, as shown in

Table 1. Causes of changes in NPP.

Type	Causes of Changes in NPP
kNPPa ≥ 0 and kNPPp ≥ kNPPh	Increased by climate variation
kNPPa ≥ 0 and kNPPp < kNPPh	Increased by human activities
kNPPa < 0 and kNPPp ≥ kNPPh	Decreased by climate variation
kNPPa < 0 and kNPPp < kNPPh	Decreased by human activities

Note: kNPP_a, kNPP_p, and kNPP_h indicate the slope value of the NPP_a, NPP_p, and NPP_h, respectively.

[28,29].

2.3.3. Correlation Analysis

To further analyze the link between climate change and NPP change, this study measures the correlation between climate change and NPP change based on the correlation coefficient, which is calculated as follows [19]:

$$\mathrm{r}=\frac{{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}\left[\left({\mathrm{x}}_{\mathrm{i}}-\stackrel{-}{\mathrm{x}}\right)\left({\mathrm{y}}_{\mathrm{i}}-\stackrel{-}{\mathrm{y}}\right)\right]}{\sqrt{{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{\left({\mathrm{x}}_{\mathrm{i}}-\stackrel{-}{\mathrm{x}}\right)}^2{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{\left({\mathrm{y}}_{\mathrm{i}}-\stackrel{-}{\mathrm{y}}\right)}^2}},$$

(11)

where r is the correlation coefficient; n = 21, representing the time span from 2000 to 2020; x_i is the NPP value of the i-th year, y_i is the climate value (temperature or precipitation) of the i-th year, and $\stackrel{-}{\mathrm{x}}$ is the average annual NPP value, $\stackrel{-}{\mathrm{y}}$ is the multi-year climatic average. The correlation level is divided into significant negative correlation (r < 0, p < 0.05), a non-significant negative correlation (r < 0, p > 0.05), a non-significant positive correlation (r > 0, p > 0.05), and a significant positive correlation (r > 0, p < 0.05).

3. Results

3.1. Climate Change in the Study Area from 2000 to 2020

From 2000 to 2020, the temperature in the study area showed an upward trend (as shown in Figure 4). The annual average temperature was −3.99 °C, the average annual minimum temperature was −4.71 °C (2000), and the maximum temperature was −3.50 °C (2020). The overall fluctuation of the annual average temperature showed an upward trend (z > 0), but it is not a clear trend (p > 0.01). Through linear fitting, the average temperature increase rate in 21 years was 0.28 °C/10 a. From a spatial point of view, the temperature in the northern and eastern regions was higher, and the temperature in the central and western regions was lower. The temperature in most regions showed a trend of increasing temperature. The annual precipitation was 534.36 mm, the annual minimum precipitation was 453.47 mm (2001), and the annual maximum precipitation was 639.22 mm (2018). The overall fluctuation of the annual precipitation shows an upward trend (z > 0), but there is no obvious trend (p > 0.01). Through linear fitting, the average precipitation increase rate over 21 years was 45.22 mm/10 a. From a spatial point of view, precipitation decreased from east to west, with minimum precipitation in the northwest, and maximum precipitation in the southeast. Overall, the study area showed a warm-humid climate trend from 2000 to 2020.

3.2. Variation in NPP in the Study Area Grassland from 2000 to 2020

From 2000 to 2020, the NPP_p and NPP_a of the Yellow River source grassland showed a steady upward trend (as shown in Figure 5), and NPP_h showed a downward trend, but these changes were not significant (R² were 0.57, 0.57, and 0.17, respectively). The NPP_p was the lowest in 2002, at 56.6 g C/m², highest in 2018 at 94.87 g C/m², and the multi-year average was 74.51 g C/m². The NPP_a was lowest in 2004 at 182.55 g C/m², the highest in 2018 at 234.07 g C/ m², and the multi-year average was 205.40 g C/m². Moreover, the NPP_h was lowest in 2013, at −161.57 g C/m², highest in 2004, at −110.98 g C/m², and the multi-year average was −131.09 g C/m². Overall, the NPP_a in the study area was less than NPP_p from 2000 to 2020, and the NPP was mostly supplemented by human activities and showed an increasing trend. Thus, the impact of artificial restoration measures on the grassland at the Yellow River source was positive.

There were three dominant types of grassland in the study area: alpine meadow (AM), alpine steppe (AS), and swamp meadow (SM), covering up to 86.88% of the study area (as shown in Figure 6). For alpine meadow, the multi-year average NPP_p was 77.48 g C/m², NPP_a was 214.35 g C/m², and NPP_h was −136.81 g C/m². For alpine grassland, the multi-year average NPP_p was 56.09 g C/m², NPP_a was 150.29 g C/m², and NPP_h was −94.99 g C/m². For swamp meadow, the multi-year average of NPP_p was 45.63 g C/m², NPP_a was 106.39 g C/m², and NPP_h was −59.36 g C/m². The NPP_p and NPP_a of alpine meadows are higher than those of alpine steppe and swamp meadows, with the greatest contributions from human activities.

The annual average NPP_p of the Yellow River source grassland from 2000 to 2020 decreased from the southeast to the northwest (as shown in Figure 7). The distribution trend was consistent with the spatial distribution of the annual average precipitation, indicating that NPP_p is closely related to precipitation. From 2000 to 2020, the NPP_p of the grassland in the study area increased significantly, with an average increase of 1.07 g C/m² a, and the increase rate in the eastern part was significantly higher than that in the western area. The spatial distribution of the annual average NPP_a from 2000 to 2020 was the same as that of NPP_p, with the distribution decreasing from southeast to northwest. From 2000 to 2020, the NPP_a of most grasslands in the study area showed an increasing trend, with an average increase of 1.51 g C/m² a; 4.06% of the grasslands were degraded. The grasslands were mainly located in Qumalai, Dari, Gander, and Jiuzhi. Compared with the increasing trend in NPP_p, the increase of NPP_a was larger. The grassland restoration projects carried out in the study area over many years have thus achieved good results. The NPP_a increased from west to east, especially in the four counties in the north of the study area, due to human activities. From 2000 to 2020, the NPP_h of the grasslands in the study area increased and then halved. A total of 63.87% of the grasslands supplemented by human activities showed an increasing trend in primary productivity. Where the primary productivity showed a decreasing trend, and this indicates that the grassland utilization rate is decreasing.

3.3. Correlation between NPP and Climatic Factors in the Grasslands in the Study Area from 2000 to 2020

Through correlation analysis between the NPP in the study area grassland and the precipitation and temperature from 2000 to 2020 (as shown in Figure 8), the correlation coefficient between NPP_p and temperature was between −0.68 and 0.88, and 4.08% of the NPP_p was negatively correlated with temperature. The NPP_p was highly positively correlated with precipitation, and the correlation coefficient between NPP_p and precipitation in most grasslands was above 0.8, indicating that precipitation was the main factor affecting grassland NPP_p in the study area. The correlation between NPP_a and temperature coefficients ranged from −0.60 to 0.89. Among the grasslands, 6.07% had a negative correlation between NPP_a and temperature, the average correlation coefficient between NPP_a and precipitation was 0.20, and 14.78% of grasslands had a negative correlation between NPP_a and precipitation. There was a negative correlation between NPP_a and precipitation in 14.78% of grasslands, and the correlation coefficient between NPP_a and precipitation was the highest in the region with the largest increase of NPP_a in the northern part of the study area. The average correlation coefficient between NPP_h and the air temperature was −0.07, 65.19% of grasslands have a negative correlation between NPP_h and temperature, and the average correlation coefficient between NPP_h and precipitation was 0.34. The grassland NPP_h in the area with the largest increase in NPP_a in the north was highly negatively correlated with the precipitation, and the grassland NPP_h in the area with the largest increase in NPP_h in the east was highly positively correlated with the precipitation.

3.4. Relative Contributions of Climate Change and Human Activities to the NPP Changes

The relative contribution of climate change and human activities to grassland NPP changes was quantitatively evaluated by overlaying the variation trends of NPP_p, NPP_a, and NPP_h. As shown in Figure 9, 96.06% of the grassland NPP in the study area showed an increasing trend, of which 55.90% was increased due to the influence of climate change, mainly in the east, an area which had the largest increase in precipitation. The grassland NPP was increased due to the influence of human activities, mainly in the northern and western regions of the study area. The grassland degraded areas were mainly distributed in scattered areas in Qumalai, Dari, Gande, and Jiuzhi. The contribution of human activities in the Yellow River source to grassland restoration was increasing year by year from 2000 to 2020, due to the long-term implementation of returning grazing land to grassland and ecological restoration projects in the Three River Sources.

From the perspective of the different grassland types, 60.51% of the grassland NPP in alpine meadows increased due to climate change, and 35.05% of the grassland NPP increased due to human activities (as shown in

Table 2. Contribution to change in NPP.

Contribution to Change in NPP	AM	AS	SM
Increased by human activities	35.05%	73.64%	51.12%
Increased by climate variation	60.51%	25.91%	47.62%
Decreased by human activities	2.49%	0.32%	0.63%
Decreased by climate variation	1.95%	0.13%	0.63%

and Figure 10). Moreover, 73.64% of the grassland in alpine steppe increased due to human activities, and only 25.91% of the grassland NPP increased due to human activities. In addition, the proportion of the swamp meadow NPP increased by climate change and human activities was close to 5:5. The reason for the difference in NPP changes in different grassland types is that alpine meadows, as the main type of grassland at the source of the Yellow River, are affected by climate change and human activities, and the proportion of the increase is close to the whole Yellow River source, while alpine steppe is mainly distributed in the northern region of the source of the Yellow River, especially in Xinghai and Tongde, which are the most densely populated areas at the source of the Yellow River. Swamp meadows are the least distributed grassland type at the source of the Yellow River. The eastern swamp meadows are mostly affected by climate change, and the western ones are mostly affected by human activities.

4. Discussion

The temperature in the study area shows a spatial distribution of high temperatures in the northeast and low in the southwest (Figure 4). The temperature increase area is much larger than the temperature decrease area, and the temperature increase rate also shows a trend of high in the east and low in the west. The spatial distribution of precipitation is higher in the southeast and lower in the northwest, and the increase in precipitation also shows a trend of higher in the east and lower in the west. Jin et al. found that the spatial distribution of temperature in t the Three-River-Source region from 1961 to 2019 shows a warming gradient along the northwest-southeast direction, and the increase in warming temperature and precipitation from west to east is becoming more and more obvious [30]. The annual average temperature in the source region of the Yellow River from 2000 to 2020 is −4.71 °C and the annual temperature is −3.99 °C. According to the linear fitting results, the average temperature increase rate over 21 years is 0.28 °C/10 a. Especially between 2012 and 2020, the temperature increase rate is more obvious. The temperature increase rate at the Three Rivers source was close to 0.40 °C/10 a during the same time period, much higher than the global average temperature increase rate [30]. The precipitation in the Yellow River source from 2000 to 2020 was between 453.47 and 639.22 mm, and the annual precipitation is 534.36 mm. According to the linear fitting results, the average increase rate of precipitation in 21 years is 45.22 mm/10 a, especially between 2016–2020, the precipitation increase trend is more evident. During the same time period, the increase in precipitation in the Three Rivers source was close to 50 mm/10 a, which was higher than the global average, areas at the same latitude, and the whole region of China [31,32]. To sum up, the study area presents a warm-humid climate trend.

The NPP_p and NPP_a of grassland in the study area tended to increase significantly, showing a spatial distribution of high in the east and low in the west. Related studies have also shown similar conclusions, such as Zhang et al. used Carnegie-Ames-Stanford to estimate the changing trend of vegetation NPP in TRSR from 1982 to 2012, and the results showed that the average NPP increased in 31 years [33]. Zhang et al. found that the Yellow River source grassland showed a recovery trend from 2001 to 2016, and NPP increased significantly [34]. Tang et al. used machine learning to simulate the temporal and spatial trend of aboveground biomass (AGB) in the Yellow River source grassland from 2001 to 2020, and the results showed that 69.51% of the grassland AGB showed an upward trend [35]. This increase in NPP_a indicates that the grassland is in a state of recovery as a whole, and the decline of NPP_h indicates that the grassland is more and more affected by human activities. The variation in NPP_p and NPP_a of different grassland types was the same as that for the whole region. As the main grassland type, alpine meadows had the highest NPP levels. The grassland NPP_p in the study area had a very high positive correlation with precipitation, but little correlation with air temperature, and 4.08% of the area had a negative correlation with air temperature. The NPP_a had no obvious correlation with precipitation and air temperature. A total of 14.78% of the area had a negative correlation with precipitation, and 6.07% of the area had a negative correlation with temperature. The NPP_h had a high correlation with precipitation, but the correlation with temperature was not obvious. A total of 65.19% of the area had a negative correlation with temperature. The area of grassland NPP increase affected by climate change (55.03%) is larger than the area affected by human activities (41.05%). From the perspective of spatial distribution, the areas with increasing grassland NPP increase affected by climate change are distributed in the eastern part of the study area, which is affected by human activities. The areas with increasing grassland NPP are concentrated in the northern and western regions of the study area. On the whole, in the study area, grassland degradation caused by human activities has been significantly reduced, and there is a shift towards the restoration of grasslands, such as the construction of ecological barriers on the Qinghai-Tibet Plateau since 2000 [36].

5. Conclusions

This study quantified the relative contributions of climate change and human activities to the changes in NPP and identified the factors driving the change of NPP in the Yellow River source grassland. The results show that temperature and precipitation increased by 0.28°C/10 a and 45.22 mm/10 a, respectively, which showed a warm-humid climate trend. The NPP_p showed a significant increasing trend, and the spatial distribution characteristics were consistent with the average precipitation. Comparing different grassland types, it was seen that alpine meadow—as the main grassland types in the study area—have higher NPP_p and NPP_a than alpine steppe and swamp meadow. In addition, the relative contribution analysis shows that climate change is still the main driving factor for the increase of NPP in the grassland of the Yellow River source, with 55.90% of the grassland NPP increased by climate change. As the main grassland type of the Yellow River source, the increase of NPP in alpine meadow is mainly affected by climate change. With the official establishment of the Three-River-Source National Park, the protection of the grasslands at the Yellow River source has gained extensive attention. The next step will be the implementation of the China Water Tower conservation project, the third phase of conservation and restoration of the Three-River-Source region, and the conservation of mountains, water, forests, fields, lakes, and grasses in the Yellow River Valley of Qinghai Province. With the implementation of projects such as the Thousand Mile Protection Zone Project in the upper reaches of the Yellow River, the NPP of the Yellow River source grassland will continue to increase, and we will consider the influences of such other human activities and add the human population in future work.