Spatiotemporal Characteristic Prediction and Driving Factor Analysis of Vegetation Net Primary Productivity in Central China Covering the Period of 2001–2019

Abstract

Unveiling the variation mechanism of vegetation net primary productivity (NPP) and elucidating the underlying drivers of these changes is highly necessitated for terrestrial carbon cycle research and global carbon emission control. Taking Henan Province, renowned as the anciently central China and current China’s foremost grain producer, as an example, this study employed the Theil–Sen Median Trend Analysis to evaluate the spatiotemporal characteristics and trends of NPP. Correlation Analysis and Residual Analysis were used to explain the drivers of NPP dynamics. To deepen the inquiry, the Geodetector method was employed to scrutinize the multifaceted effects and interplay among diverse variables influencing NPP. The result showed demonstrated that approximately 85.72% of the area showed an increase in NPP, covering a broad geographical distribution. Notably, 89.31% of the province has witnessed a positive human-driven NPP change. It means human activities emerged as a driving force with a positive effect on vegetation NPP, consequently fostering an increasing trend of NPP. Among climatic factors, the correlation between NPP and precipitation was stronger than that between the temperature and NPP, the determined power of factors in Henan Province was population density, (0.341) > GDP (0.326) > precipitation (0.255) > elevation (0.167) > slope (0.136) > temperature (0.109), and a single factor had a lesser interaction effect than two factors. The implications of these findings extend beyond the realms of research, potentially offering valuable insights into the formulation of targeted ecosystem restoration measures tailored to the distinct context of Henan Province, and also expect to provide crucial references for carbon emission control in China and across the world.

1. Introduction

Carbon emission control is a hot issue across the world. The net primary productivity (NPP) of vegetation, a crucial parameter in carbon cycle simulation, represents the net amount of carbon absorbed by vegetation over a specific unit of space and time. This metric is computed by taking the difference between natural vegetation photosynthesis and autotrophic respiration. Vegetation carbon accumulation stands as a key component and a central variable of terrestrial carbon cycling [1,2]. Serving as a vital gauge of vegetation activity, NPP assumes a crucial role in ecological investigations, frequently utilized as a primary indicator for resource utilization and carbon fluxes within the biosphere [3]. Consequently, the dynamics of terrestrial ecosystem NPP have received heightened attention from an expanding community of researchers, particularly in the context of global changes [4,5,6,7,8].

Numerous studies underscore climate change as the foremost and direct determinant influencing vegetation activity. Remote sensing data analysis and manipulation experiments have substantiated the potential effects stemming from fluctuations in temperature, precipitation, and solar radiation on vegetation growth [9,10,11,12,13]. Meanwhile, intensive human activities have been recognized as a critical catalyst driving changes in NPP in China [14,15]. For example, a series of ecological restoration projects implemented since the late 1990s have propelled NPP growth, while inappropriate afforestation and grass planting in arid and semiarid areas have led to vegetation degradation [16]. All the aforementioned activities may introduce changes in ecosystem structures and functionalities, altering soil microbiota dynamics and soil organic matter, and modulating the exchange of energy, water, and carbon flux between the atmosphere and land surface [17]. Implicit in this context is the capacity of NPP to assimilate these climatic, ecological, geochemical, and anthropogenic influences on the planetary system. Consequently, NPP garners recognition as a pivotal variable employed to scrutinize and assess the impacts and pressure upon the natural milieu within a given ecosystem [18,19].

In this area of research, much of the literature focuses on the identification and estimation of the NPP, as well as its temporal and spatial variations; additionally, some research has attempted to identify the driving forces contributing to these variations [20,21,22,23], such as the factors of climate change and human activity [24,25,26]. However, several constraints have impeded the breadth and depth of analyses in earlier studies. For instance, the accuracy of NPP estimations is significantly influenced by the quality of data, the interpolation method, and the specific area of study [27,28,29].

Furthermore, observed trends in NPP can vary depending on the duration and scope of the data being analyzed [30].

As a major agricultural province in China, Henan Province has witnessed notable transformations in land-use patterns over recent decades. These changes have been subjects of concern in both scientific and political spheres, particularly in light of anthropic alterations [31,32,33]. Prior research on Vegetation NPP in Henan has predominantly focused on trends up until 2010, generally considering meteorological factors. However, the agricultural landscape in Henan has been undergoing significant transformations recently. Initiatives such as water and soil conservation projects, together with ecological civilization projects, have triggered noticeable alternations in the NPP level. Emerging technologies have opened up new pathways for research. Compared to traditional methods of evapotranspiration observation and estimation, remote sensing technology enables dynamic, large-scale data acquisition for non-uniform surface evapotranspiration. Therefore, employing remote sensing techniques for NPP estimation has emerged as a research hotspot [34,35,36]. The MOD17A3 data product, the global terrestrial vegetation NPP interannual developed by NASA using MODIS remote sensing parameters, the Biome-BGC model, and the light energy utilization rate model have undergone verification and application, further substantiating their utility in this research domain [6,37,38].

Ultimately, this paper aims to enrich the existing body of knowledge on vegetation NPP through an exploratory case study focusing on Henan Province. Utilizing an expanded dataset of MOD17A3 data and advanced preprocessing methods, we analyze changes in vegetation NPP spanning the years of 2001–2019.

Specifically, the study aims to identify phase-related shifts in the terrestrial ecosystem NPP across varying temporal, including annual and seasonal, and spatial, from national to regional scales, conditions. Theil–Sen median trend analysis, correlation analysis, residual analysis, and other methods are used in this study, which are popular and advanced in the study field. To keep up with the wave of research, Geodetector is used as a tool to explore the relationships between factors. We seek to reveal the relative contributions of climatic or anthropogenic factors that trigger shifts in NPP trends. The findings are expected to provide critical guidance for the development of ecosystem restoration strategies in Henan Province.

2. Materials and Methods

2.1. Study Area

Henan Province is situated in the North China Plain, with geographical coordinates ranging from 31°23′ N to 36°22′ N and 110°21′ E to 116°39′ E. It encompasses an area of 165,700 km² (Figure 1). The province features diverse and dramatic topological variations, including higher elevations in the west and lower regions in the east. Notably, the Taihang, Funiu, Tongbai, and Dabie Mountains border the province’s northern, western, and southern boundaries. The central and eastern regions are located within the Huang-Huai-hai alluvial plain, while the northwest sector is part of the Nanyang Basin. Collectively, the province’s plains, basins, mountains, and hills account for 55.7%, 26.6%, and 17.7% of the total area, respectively. From a climatic perspective, Henan straddles both the north subtropical and warm temperate zones, characterized by humid and semi-humid conditions under a monsoonal climate. The average annual temperature ranges between 12.1 and 15.7 °C, and annual precipitation varies from 1380 to 532 mm. The climatic and topographic diversity has given rise to various vegetative formations, fostering rich and diverse plant resources, which also significantly benefit agriculture development.

2.2. Datasets

The annual net primary productivity (NPP) data were derived from NASA’s MOD17A3 dataset, available from the Moderate Resolution Imaging Spectroradiometer (MODIS) (available at: https://ladsweb.modaps.eosdis.nasa.gov/, accessed on 12 September 2021). This dataset offers interannual records of global land vegetation NPP, synthesized from MODIS remote sensing parameters, the Biome-BGC model, and a light energy utilization model. It has been both verified and widely applied. For this study, NPP data from 2001–2019 were downloaded and processed, with outlier data points removed to ensure accuracy.

Meteorological data were sourced from the National Meteorological Data Service Center (available at: http://data.cma.cn/, accessed on 9 October 2021). A total of 104 surface meteorological stations were selected for this study. Climate factors, specifically the annual average temperature (°C) and annual precipitation (mm), were extracted for each station for the years 2001 to 2019. The Kriging method was employed for spatial interpolation of these meteorological data, with elevation considered an additional variable.

Digital Elevation Model (DEM) data were obtained from the Geospatial Data Cloud (available at: https://www.gscloud.cn/, accessed on 9 October 2021). From this dataset, topographic features such as elevation and slope were extracted for the study area.

Data on human activities, considered to be influential factors for NPP change, were sourced from the Henan Province Bureau of Statistics (available at: https://tjj.henan.gov.cn/tjfw/tjcbw/tjnj/, accessed on 9 October 2021). Specifically, Gross Domestic Product (GDP) and population density (PD) data from 2001 to 2019 were incorporated into the study.

2.3. Theil–Sen Median Trend Analysis to Explore the Trend of Data Series

The Theil–Sen median trend method, in conjunction with the Mann–Kendall (M–K) test, was employed to analyze the spatiotemporal variations in NPP. The Theil–Sen median [39,40,41] is calculated as follows:

$$\rho =median\frac{x_j-x_i}{j-i}, 1<i<j<n.$$

(1)

In this formula, $x_j$ and $x_i$ represent the time series of NPP, and $n$ denotes the length of the time series. A negative value of $\rho$ indicates an increasing trend in the data, and vice versa.

The M–K test is robust in that it does not mandate the sample data adhere to a particular distribution; thus, it can test non-normally distributed data and exclude outlier values. This test has found widespread application in long-term time series analyses across various domains, including hydrology, climatology, and vegetation science. The M–K test statistic is calculated according to

$$S={\sum }_{i=1}^{n-1}{\sum }_{j=i+1}^{n}sgn(x_j-x_i),$$

(2)

with

$$sgn\left(x_j-x_i\right)=\left\{\begin{array}{c}1, (x_j-x_i>0)\\ 0,{ (x}_j-x_i=0)\\ -1, (x_j-x_i<0) \end{array}\right..$$

(3)

The statistic S approximates a normal distribution when subjected to the subsequent Z-transformation:

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

(4)

where n is the sample size. If $n>8$, S follows an approximately normal distribution. The average and variance are computed as

$$E\left(S\right)=0,$$

(5)

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

(6)

The normalized Z follows a standard normal distribution. In a two-sided test, a significant trend change is inferred if $\left|Z\right|\ge Z_{1-\propto /2}$. All analyses were conducted using MATLAB 2020a software, specifically focusing on the Theil–Sen median trend and the M–K test.

2.4. Correlation Analysis to Recognize Mutual Interference of Various Influencing Factors

In order to disentangle the mutual interference of various influencing factors, this study employed partial correlation analysis, computing the correlation coefficient between NPP and climate variables, namely temperature and precipitation.

The calculation of these coefficients was grounded in the simple correlation coefficient framework [42]. The linear correlation analysis was executed according to the following formula:

$$R_{xy}=\frac{{\sum }_{i=1}^n(x_i-x )(y_i-y)}{\sqrt{{\sum }_{i=1}^n{(x_i-x )}^2{\sum }_{i=1}^n{(y_i-y)}^2}},$$

(7)

where $x_i$ denotes the NPP value for year $i$; $y_i$ represents either the temperature or precipitation for the same year; $\overline{x}$ and $\overline{y}$ are the means respective variable, and $R_{xy}$ serves as the correlation coefficient.

For the computation of the partial correlation coefficient, we utilized the following equation:

$$R_{xy,z}=\frac{R_{xy}-R_{xz}·R_{yz}}{\sqrt{\left(1-R_{xz}^2\right)\left(1-R_{yz}^2\right)}},$$

(8)

where $x, y, z$ represent NPP, temperature, and precipitation, respectively; $R_{xy}$ indicates the partial correlation coefficient between NPP and temperature conditioned on the same level of precipitation, while $R_{xz}$ and $R_{yz}$ represent the linear correlation coefficient between NPP and temperature, and between temperature and precipitation, respectively.

2.5. Residual Analysis to Detect Driving Factors for NPP

In this study, the residual analysis method is employed to quantify the relative contributions of human activities and climate change to variations in vegetation NPP.

Initially, a binary linear regression model is formulated on a pixel-by-pixel basis, utilizing NPP time-series data as the dependent variable and the interpolated time-series data of temperature and precipitation as independent variables. Parameters within this model are subsequently calculated. Following this, the predictive values of vegetation NPP are derived based on existing temperature and precipitation data with the regression model. The linear trend rate of these annual NPP predictive values serves as a gauge for assessing the climate impact on NPP. To further scrutinize the role of human activities in affecting NPP, the residuals between observed and predicted NPP values are calculated. The linear trend rate of these annual NPP residuals helps to pinpoint the extent of human influence on NPP. The Miami model [43,44] is employed for predictive value calculations, defined by the following formula:

$${NPP}_{CC}=min\left\{\left[1+\frac{3000}{EXP\left(1.315-0.119T\right)}\right],\left(3000\left[1-EXP\left(-0.000664P\right)\right]\right)\right\},$$

(9)

$${NPP}_{HA}={NPP}_{obs}-{NPP}_{CC},$$

(10)

where ${NPP}_{CC}$ represents the predicted value based on the regression model, ${NPP}_{obs}$ is the observed value; $T$ denotes average temperature, $P$ signifies annual precipitation, and ${NPP}_{HA }$ is the residual value.

Subsequently, using a univariate linear regression method, the linear trends of ${NPP}_{CC}$ and ${NPP}_{HA}$ are determined as follows:

$$slope=\frac{n\times {\sum }_{i=1}^n\left(i\times M_i\right)-{\sum }_{i=1}^{n}i{\sum }_{i=1}^n{M}_i}{n\times {\sum }_{i=1}^n{i}^2-{\left({\sum }_{i=1}^{n}i\right)}^2},$$

(11)

where $slope$ refers to the linear trend rate of ${NPP}_{HA}$ or ${NPP}_{CC}$; n indicates the total number of years under consideration; $M_i$ represents the value of ${NPP}_{HA}$ or ${NPP}_{CC}$ for a given year $i$.

The primary drivers and their relative contributions to NPP changes are identified in

Table 1. The Criteria for the Division of Driver and Contribution.

$\mathit{s}\mathit{l}\mathit{o}\mathit{p}\mathit{e}$ $\left({\mathit{N}\mathit{P}\mathit{P}}_{\mathit{o}\mathit{b}\mathit{s}}\right)$	Drivers	Criteria for Division of Drivers		Contributions (%)
		$\mathit{s}\mathit{l}\mathit{o}\mathit{p}\mathit{e} \left({\mathit{N}\mathit{P}\mathit{P}}_{\mathit{C}\mathit{C}}\right)$	$\mathit{s}\mathit{l}\mathit{o}\mathit{p}\mathit{e} \left({\mathit{N}\mathit{P}\mathit{P}}_{\mathit{H}\mathit{A}}\right)$	Climate Change	Human Activity
>0	CC&HA	>0	>0	$\frac{slope\left({NPP}_{CC}\right)}{slope\left({NPP}_{obs}\right)}$	$\frac{slope\left({NPP}_{HA}\right)}{slope\left({NPP}_{obs}\right)}$
	CC	>0	<0	100	0
	HA	<0	>0	0	100
<0	CC&HA	<0	<0	$\frac{slope\left({NPP}_{CC}\right)}{slope\left({NPP}_{obs}\right)}$	$\frac{slope\left({NPP}_{HA}\right)}{slope\left({NPP}_{obs}\right)}$
	CC	<0	>0	100	0
	HA	>0	<0	0	100

Note: $slope \left({NPP}_{obs}\right)$, $slope \left({NPP}_{obs}\right)$, corresponded to their respective trend rates; in this context, CC refers to influences driven by climate, HA signifies those driven by human activities, and CC and HA represents a combination of both factors.

.

2.6. Geodetector to Identify Spatial Stratified Heterogeneity

Spatial Stratified Heterogeneity (SSH) describes the phenomenon where units within individual strata exhibit greater similarity to each other than to those in different strata. Geodetector serves as a statistical instrument to identify this spatial stratified heterogeneity and uncovering their underlying causes [45,46,47]. Q-statistic is employed to quantify the extent of spatial stratified heterogeneity, identify explanatory variables, and analyze the interplay among them.

$$Q=1-\frac{{\sum }_{h=1}^L{N}_h{\sigma }_h^2}{N{\sigma }^2}=1-\frac{SSW}{SST},$$

(12)

$$SSW={\sum }_{h=1}^L{N}_h{\sigma }_h^2, SST=N{\sigma }^2.$$

(13)

In this formula, h(1, …, L) designates the strata of either the dependent variable Y or the explanatory factor X; $N_h$ and N denote the account of units in layer h and the overall research zone, respectively, while ${\sigma }_h^2$ and ${\sigma }^2$ denote the variance within the layer and across the whole area. The terms SSW and SST stand for the “Within Sum of Squares” and “Total Sum of Squares”, respectively. The range of q is [0, 1], with larger values signifying a more pronounced influence on the dependent variable Y, which, in this study, is the NPP.

The interaction among the factors can be categorized as outlined in

Table 2. The Criteria for Factor Interaction.

Criteria	Interaction Type
$q(X_1\cap X_2)>\left(q\right(X_1)+q(X_2\left)\right)$	Nonlinear, enhance
$q(X_1\cap X_2)=\left(q\right(X_1)+q(X_2\left)\right)$	Independent
$q(X_1\cap X_2)>MAX\left(q\right(X_1)+q(X_2\left)\right)$	Bi-linear, enhance
$MIN\left(q\right(X_1)+q(X_2\left)\right)<q(X_1\cap X_2)<MAX\left(q\right(X_1)+q(X_2\left)\right)$	Unilateral, weaken
$q(X_1\cap X_2)<MIN\left(q\right(X_1)+q(X_2\left)\right)$	Nonlinear, weaken

.

3. Results

3.1. Temporal Variation Characteristics of NPP

The interannual variation patterns of vegetation NPP and the interannual change rate in Henan Province from 2001 to 2019 are displayed in Figure 2. The NPP values fluctuated within a range of 200–500 gC·m⁻²·a⁻¹, with an average value of 409.69 gC·m⁻²·a⁻¹. The peak NPP was recorded in 2015, reaching 473.13 gC·m⁻²·a⁻¹, while the lowest was in 2001, at 302.27 gC·m⁻²·a⁻¹. Overall, NPP has been on an upward trajectory, increasing at a rate of 3.091 gC·m⁻²·a⁻¹. This increasing trend accelerated post 2014, when the average NPP reached 438 gC·m⁻²·a⁻¹, an 8.8% increase compared to the average NPP of 398 gC·m⁻²·a⁻¹ from 2011–2013. According to the five-year averages presented in

Table 3. Average NPP.

Years	2001–2005	2006–2010	2011–2015	2016–2019
Average NPP (gC·m−2·a−1)	381.902	421.158	411.812	427.630

, the most recent period (2016–2019) exhibited the highest average NPP at 427.63 gC·m⁻²·a⁻¹, whereas the earliest five years (2001–2015) had the lowest at 381.902 gC·m⁻²·a⁻¹. When comparing the annual NPP values for the years 2001–2019 to the overall annual average, the years 2008, 2015, and 2018 stood out with notably higher values, while 2001, 2002, and 2013 had the lowest. These changes in NPP appear to correlate with the introduction of the ecological conservation project in Henan Province. Specifically, in 2014, the government issued a Forestry and Ecological Restoration Project, which likely contributed to the observed uptick in NPP values.

3.2. Spatial Variation Characteristics of NPP

Figure 3 illustrates the spatial distribution of the annual average NPP across Henan Province for the years 2001–2019. The data reveal distinct regional variations: the southern regions, including Sanmenxia, Nanyang, Xinyang, and Zhumadian, generally exhibit higher NPP values, often exceeding 400 gC·m⁻²·a⁻¹. These areas are characterized by flat terrains blanketed by forests, boasting high vegetation cover and photosynthetic efficiency. In contrast, northern regions like Anyang, Puyang, Hebi, and Jiyuan, which are situated in the southern foothills and eastern slopes of Taihang Mountains, have lower NPP values, typically ranging from 200 to 400 gC·m⁻²·a⁻¹. These regions are marked by rugged, stony terrains and a drier climate, resulting in sparse vegetation. Urban areas, large reservoirs, and riversides also exhibit lower NPP values, generally below 300 gC·m⁻²·a⁻¹, due to the impacts of industrialization, urbanization, and soil degradation.

The M–K test validated the NPP variation trend analyses during the study period. Results, categorized into eight levels as detailed in

Table 4. Results of SEN Trend analysis in Henan.

Classification Standard		Variation Trend	(%)
$\rho <0$	$\left|Z\right|\le 1.675$	Non-Significant degradation	12.59
	$1.675<\left|Z\right|<1.96$	Slight degradation	0.45
	$1.96<\left|Z\right|<2.576$	Degradation	0.65
	$\left|Z\right|>2.576$	Significant degradation	0.58
$\rho >0$	$\left|Z\right|\le 1.675$	Non-Significant increase	49.45
	$1.675<\left|Z\right|<1.96$	Slight increase	9.08
	$1.96<\left|Z\right|<2.576$	Increase	13.17
	$\left|Z\right|>2.576$	Significant increase	14.02

, revealed significant heterogeneity in NPP across Henan Province, as visualized in Figure 4. The sen-trend values for NPP ranged from −22.33 to 52.83. Overall, the province displayed a net increase in NPP; regions with an increasing trend (SEN trend $\rho >0$) accounted for 85.72% of the area. Within this, areas showing a non-significant increase in NPP made up 49.45%, while those with a significant increase constituted 36.3%. Conversely, regions experiencing a decline in NPP made up 14.27% of the total area. Notably, urban agglomeration, such as Zhengzhou, Xinxiang, Jiaozuo, as well as ecological conservation areas along the Yellow River showed decreased NPP. This decline is primarily ascribed to the extensive destruction of vegetation and accelerated soil erosion, both consequences of rapid urbanization and industrialization in these regions.

3.3. Correlation Analysis of NPP and Climate Factors

NPP fluctuations are primarily determined by climatic variables such as regional hydrothermal conditions, temperature, and precipitation. To disentangle the interference of these parameters, individualized partial correlation analyses were conducted, the results of which are illustrated in Figure 5a,b.

The calculated partial correlation coefficient for the relationship between NPP and temperature oscillated within a range of −0.857 to 0.884, yielding a mean value of −0.015 and a standard deviation of 0.232. These results revealed latitudinal differences in the correlation between NPP and temperature. Specifically, areas with a positive correlation, constituting 69.56% of the data set, were geographically clustered in the northern sectors of Henan Province, encompassing Jiyuan, Jiaozuo, Shangqiu, etc. In contrast, negative correlations, comprising 30.44% of the dataset, predominantly manifested in the southern cities of Nanyang and Xinyang.

In the context of precipitation, the partial correlation coefficients spanned a range from −0.849 to 0.904, with an associated mean and standard deviation of 0.228 and 0.279, respectively. The preponderance of positive correlation, accounting for a significant 79.80% of the province, was ubiquitously distributed, with the conspicuous exception of the southwestern region. Within this particular region, a concentration of negative correlations was observed. The maximal absolute values in partial correlation coefficients, indicative of heightened sensitivity of NPP to precipitation fluctuations, were localized in the southeastern precincts of the province. Comparative analysis of the correlation coefficients substantiates the notion that precipitation exerts a more robust influence on NPP than temperature.

Utilizing the residual method, an NPP-climate model was established, incorporating data on NPP, precipitation, and temperature from 2001 to 2019 in Henan Province. Calculations of predicted NPP residual discrepancies, actual NPP, and predicted NPP were computed to assess the contributory impact of climatic variables and human activities on variations in vegetation NPP. These analytical outcomes are graphically delineated in Figure 6.

Based on the results, regional specificity is exhibited by the dominant factors influencing NPP alterations, as cataloged in

Table 5. The results of Residual analysis in Henan.

Drivers	HA Negative	CC Negative	HA & CC Negative	HA & CC Positive	CC Positive	HA Positive
Area (%)	5.64	0.46	4.59	56.66	0.51	32.14

. In an estimated 56.66% of the province, climate change (CC) and human activities (HA) exert a positive influence on NPP change, as the most widely distributed. Areas singularly influenced by CC, either in a positive or negative manner, constitute a scant 1% of the overall province. Approximately 32.14% of the province’s area manifests NPP augmentation driven predominantly by HA. This is notably the case in the southeastern region, including Nanyang, Zhumadian, and Zhoukou, where arable land is the predominant landscape and is thus significantly impacted by human activities. Conversely, 4.59% of the study area indicates an NPP decline, attributed to the confluence of both HA and CC. This trend is principally localized in the eastern precincts of the Xinyang city. In summation, the NPP changes over the past 19 years (2001–2019) predominantly stem from a synergy between HA and CC across the province.

The impacts of climatic fluctuations, specifically precipitation and temperature, as well as human activities on NPP change, were evaluated discretely in this section. In the vast majority of regions, encompassing approximately 83.49% of the studied geographical expanse, climate variables manifested a positive influence on NPP. These territories are predominantly located in the western regions of Henan, such as Sanmenxia, Luoyang, and Nanyang. Largely comprised of forested and grassland areas, these regions are relatively insusceptible to human-induced alterations. Conversely, a negative influence of climate change on NPP was observable in a smaller proportion of areas, accounting for roughly 16.51%, with significant concentrations in urban centers such as Zhengzhou and Kaifeng. With respect to human activities, a positive influence on NPP was discerned across an expansive 94.29% of the study area. In stark contrast, 5.71% of the studied territories exhibited a decline in NPP, attributed to human activities, displaying a rather sporadic spatial distribution. Cumulatively, the data underscore the beneficial role of human interventions in elevating NPP levels, overshadowing the contributory effects of climatic variables in most regions.

3.4. Contribution to the NPP Change of Environment and Human Activities Factors

To identify specific drivers of NPP changes more accurately, this study extends beyond traditional climatic variables such as precipitation and temperature to incorporate factors such as elevation and slope for capturing natural environment changes, as well as gross domestic product (GDP) and population density (PD) to account for human activities. The Geodetector method is utilized to assess the contributions of these four environmental factors and two socio-economic factors. While most factors exhibit a general trend of increase over time, anomalies are observed in precipitation, GDP, and population density, which peaked in 2009. Importantly, these factors display substantial temporal heterogeneity.

Among the examined variables, precipitation, GDP, and PD emerge as particularly influential. Socio-economic factors, characterized by a GDP q value of 0.326 and a PD q value of 0.341, appear to exert a stronger impact on NPP than environmental factors. Precipitation, with a q value of 0.225, surpasses other environmental factors in its influence. The results are summarized in

Table 6. Spatial divergence effect of each factor on average NPP detected by Geodetector (2001–2019).

	P	TEM	ELEVATION	SLOPE	GDP	PD
q value	0.255	0.109	0.167	0.136	0.326	0.341

.

During the interaction detection phase, various types of relationships between the six driving factors are identified, including nonlinear enhancement and two-factor enhancement. This suggests an interplay among variables. Although the interactions exhibit some consistency over the years, notable variations are also present. Heat maps generated from data for the years 2001, 2010, and 2019 serve to visualize these intricate relationships, as shown in Figure 7 and Figure 8.

In 2001, the interaction between precipitation and other factors was notably strong, as was the case with elevation. Specifically, the interactive explanatory force between rainfall and elevation registered 0.321, while interactions between precipitation and slope and precipitation and population density recorded values of 0.271 and 0.263, respectively. In 2010, the interaction intensity involving precipitation surpassed that of other factors. The most notable relationships were observed between precipitation and elevation, as well as precipitation and population density, with respective values of 0.565 and 0.558. In contrast, the year 2019 exhibited an escalation in the interaction between precipitation and other factors, as well as between elevation and other factors. The interaction between precipitation and population density reached a peak value of 0.388, closely followed by the value of 0.384 for the interaction between precipitation and elevation. The interaction between elevation and population density ranked third with a value of 0.368. Notably, the intensity of interaction between natural and anthropogenic factors in 2010 was more robust than in the years 2001 and 2019.

Over the entire study period from 2001 to 2019 (Figure 9), the interaction between human activity variables and other factors consistently maintained a high q value, all exceeding 0.320 and ranging from 0.326 to 0.490. The interaction between precipitation and elevation consistently remained the most influential, registering the highest value at 0.518. The q values of the two-factor that can be clearly noticed are larger than those of the single factor, indicating that the interaction of the NPP receiving the influence of the two factors is greater than that of the single factor.

4. Discussion

Climatic factors play an important role in shaping the dynamics of NPP [48,49,50,51]. In this study, factors such as precipitation, temperature, elevation, and slope were employed to account for climatic fluctuations given their integral association with plant growth. Remarkably, approximately 83.49% of the province manifests a positive contribution rate with climate-induced NPP. Within the context, temperature and precipitation emerge as predominant drivers, influencing not only plant photosynthesis and respiration rate but also root system development and nutrient utilization efficiency.

During the years of 2001–2019, Henan Province has experienced a marginal temperature increase. The partial correlation coefficients between temperature and NPP range from −0.4 to 0.51. Owing to its relatively high latitude positioning, northern Henan benefits from moderate temperature elevations, which in turn foster vegetation photosynthesis and prolong the growing season. This results in a subsequent enhancement of vegetation NPP. Conversely, the uptick in temperature amplifies evapotranspiration rates in both plant and soil water, thereby engendering a negative correlation between vegetation NPP and temperatures in southern of Henan, such as Nanyang and Xinyang.

Precipitation constitutes the principal water source of vegetation and is indispensable for optimal growth conditions [52,53,54,55,56]. The majority of the study area exhibited a positive correlation between NPP and precipitation, with partial correlation coefficients ranging from −0.849 to 0.904. However, an excessive surge in precipitation levels can incite soil erosion and compromise vegetation integrity. Additionally, an increase in rainy days weakens the photosynthesis of vegetation, culminating in a negative correlation demonstrated in a few regions.

With economic growth and social development, increased population and accelerated urban expansion and industrialization have generated a suite of human activities [57,58]. These activities intensify natural resource exploitation and soil erosion, resulting in a decline in NPP levels. Conversely, strategic interventions, such as the improvement of agricultural management practices and the implementation of eco-civilization projects, contribute positively to NPP. Results from the residual method analysis indicate that approximately 89.31% of the Henan Province witnessed a positive human-driven NPP change, affirming the beneficial role of human activities on vegetation NPP. In summary, government-led sustainability initiatives have played a crucial role in the positive trend, substantially boosting vegetation’s NPP in Henan Province.

The results of this study established that climatic variables, particularly precipitation, exert a tangible impact on the spatial configuration and overall distribution of NPP in Henan Province. Precipitation emerged as the dominant variable influencing vegetation growth and its year-to-year changes, showing a positive correlation with NPP. This finding aligns with the results of the partial correlation analysis. In the realm of economic indicators, GDP serves as a key metric for assessing a country’s or region’s economic status and development; population density, representing the number of individuals a species within a specific area, serves as a valuable metric for both demographic analysis and ecosystem assessment. In this paper, GDP and population density are identified as indicators of anthropogenic influences. These factors demonstrate greater determined power than climate factors, and significantly impact vegetation NPP.

While this study provides an in-depth analysis of NPP’s spatial and temporal characteristics in Henan Province and its interplay with climatic factors and human factors, limitations exist in several areas: this study relies on MOD17A3H NPP data with a spatial resolution of 500 m, spanning from 2001 to 2019. This limited spatial resolution and relatively short time series constrain the depth of our analysis. Methodologically, the residual method has inherent limitations. For instance, key climatic variables like solar radiation and wind speed were not incorporated into the multiple regression models. The Geodetector methods also confined the analysis to a set list of climatic and human factors, neglecting important variables such as the pace of the urbanization process, ecological initiatives, and agricultural technology progression. These omissions introduce a measure of uncertainty into the findings, highlighting the necessity for future field studies to validate these results. Future research should aim to assess the impact of large-scale human activities and climatic changes on NPP and its driving mechanism.

5. Conclusions

Unveiling the variation mechanism of vegetation net primary productivity (NPP) and elucidating the underlying drivers of these changes is highly necessitated for terrestrial carbon cycle research and global carbon emission control. This study employed the methods of Theil–Sen trend analysis, partial correlation coefficients, residual analysis, and Geodetector statistical tool to assess the spatiotemporal variations and trends of Net Primary Productivity (NPP) in Henan Province to understand the impacts of climate change and human activities on NPP. Results show the following:

(1)

From 2001 to 2019, NPP values in Henan Province fluctuated from 300 to 500 gC·m⁻²·a⁻¹, averaging at 409.73 gC·m⁻²·a⁻¹. NPP demonstrated a generally upward trend, with a notable degree of spatial heterogeneity. The Sen trend ranged from −1.84 to 7.78, with an average growth rate of 2.97. Approximately 85.72% of the area showed an increase in NPP, covering a broad geographical distribution. Nearly half the area (49.45%) remained stable, while 14.02% and 13.17% showed significant and moderate improvements, respectively.

(2)

The partial correlation coefficient between NPP and temperature ranged from −0.857 to 0.884, with a mean value of −0.015. About 50.62% of the total area, mainly concentrated in the northeastern cities of Jiyuan, Jiaozuo, and Shangqiu, showed a positive correlation. The partial correlation coefficient between NPP and precipitation ranged from −0.849 to 0.904, averaging at 0.228. Approximately 79.80% of the total area, mainly in southeastern cities like Anyang, Kaifeng, and Zhoukou, displayed a positive correlation. The partial correlation between NPP and temperature varied latitudinally, while and correlation between NPP and precipitation was stronger and mainly distributed in the southeast.

(3)

There is significant heterogeneity in the impacts of climate change and human activities on vegetation NPP change in Henan Province. Human activities exert a more pronounced influence on NPP compared to environmental factors, generally contributing to an upward trajectory in NPP levels. The driving factors of vegetation NPP in Henan were detected to show that 89.31% of the province has witnessed a positive human-driven NPP change, and the determined power of factors of vegetation NPP spatial differentiation in Henan follows the order of population density > GDP > precipitation > elevation > slope > temperature. The changes in NPP were not driven by a single factor, but by the combination of every factor. The q values of two-factor interaction were all greater than those of a single factor, and the interaction was expressed as two-factor enhanced and non-linear enhanced type.

This study underscores the pivotal roles of precipitation and temperature in shaping NPP variations and emphasize the importance of a comprehensive approach that considers multiple interacting factors. These insights enrich our understanding of the nexus between NPP and its determinants, laying a solid foundation for future ecological initiatives and policy decisions in the region. Nevertheless, the limitations exist that the chosen time frame may be insufficient for large-scale analysis, and the study only accounts for a limited range of human activities. Larger time frame and much more human activities are expected to be incorporated into analysis in the future.