Forecasting and Evaluation of Ecosystem Services Supply-Demand Under SSP-RCP Scenarios in the Henan Segment of the Yellow River Basin, China

Abstract

Equilibrating the supply and demand for ecosystem services (ESs) is essential for sustainable development. Nonetheless, elements like policy modifications, land utilization, and climate change are profoundly transforming the dynamics of ecosystem service supply and demand (ESSD). As a result, there is an imperative necessity to methodically evaluate and predict these alterations by including both social and environmental elements. This study utilized the Henan region of the Yellow River Basin (HYRB) as a case study to forecast alterations in the supply and demand for three ESs—water production (WY), carbon storage (CS), and food production (FP)—under three scenarios for 2030 and 2050, grounded in the SSP-RCP framework. We further evaluated the supply–demand equilibrium at both grid and county degrees. The results indicate the following key findings: (1) From 2020 to 2050, there are significant spatial differences in the supply and demand of these services. While the supply of CS and FP exceeds demand, the supply of WY falls short. (2) The supply–demand ratios for WY and CS are projected to decline under all scenarios, whereas FP is expected to continue growing. Surplus areas for WY and CS are aggregated in the northwest, southwest, and central areas, while FP surpluses are found in the eastern and northern plains. Deficits for all three services are primarily located in urban areas. (3) The dominant spatial patterns of supply–demand matching also vary. WY and CS exhibit high–low agglomeration patterns, particularly in the northwest and southwest mountain regions, while FP shows low–low agglomeration, mainly in the southwest and northwest mountain areas. These findings enhance comprehension of the dynamics of ESSD, serving as a foundation for environmental preservation and sustainable advancement in the Yellow River Basin, China.

1. Introduction

Ecosystem services (ESs) denote the circumstances and activities supplied by natural ecosystems and their species that underpin human life and well-being [1,2]. This is a critical area of interdisciplinary research encompassing ecology, geography, and management [3,4,5,6]. Humans benefit from ESs, but this consumption can create an imbalance between supply and demand [7], leading to ecosystem degradation and threatening the sustainability of these services [8]. The interplay between the ESSD illustrates the dynamic transfer from natural ecosystems to human societies. This procedure is crucial for ecological stability and sustained socio-economic progress [9]. As industrialization and urbanization rapidly advance, China faces numerous environmental challenges, including ecosystem degradation, pollution, and imbalanced land development. These challenges disturb the equilibrium of ESSD, endangering ecological security and sustainable development [10,11]. Therefore, modeling future ESSD is essential for optimizing natural resource allocation, improving ecosystem management, and ensuring regional ecological security and sustainable development.

Initial investigations into ESs predominantly concentrated on the supply aspect, encompassing the measurement of supply, mechanisms of impact, and the examination of trade-offs and synergistic interactions [12,13,14]. As ES research deepened, scholars have increasingly concentrated on the correlation between ESs and human well-being, leading to a gradual shift in emphasis toward demand. Human well-being can be realized through access to abundant and reliable ESs. Without understanding ecosystem needs, it is impossible to develop ecological restoration and governance strategies that meet ecosystem needs. The heightened emphasis on the effects of ESs on human well-being has coincided with an expanding corpus of study regarding the supply and demand of these services [15,16,17]. However, modern study into the ESSD predominantly focuses on historical assessments [18], with relatively few studies forecasting and analyzing future trends in ESSD.

As studies have advanced, numerous researchers have investigated the implications of forthcoming climate shifts and alterations in land use on ecosystems across different scenarios [19,20,21]. The majority of research has concentrated on the supply of ESs under different scenarios. Additionally, many scholars have predicted future demands for water resources, food security, and energy supply and demand under varying conditions [22,23,24,25,26,27]. However, comprehensive assessments of ESSD under scenarios that combine climate change and socioeconomic changes remain relatively scarce. The initiation of the sixth phase of the Coupled Model Intercomparison Project (CMIP6) has propelled research on integrated climate–socioeconomic scenarios, enabling more accurate predictions of ESSD [28,29]. CMIP6 comprises two elements: Representative concentration pathways (RCPs) and shared socioeconomic pathways (SSPs) [30]. SSPs outline potential future social development and challenges, while RCPs represent future atmospheric greenhouse gas concentration levels. The integration of SSPs and RCPs facilitates a more profound investigation of the potential effects of forthcoming environmental alterations on ESs, including food production, water yield, and carbon storage [31,32].

HYRB spans the transitional area between the river’s middle and lower sections. It covers eight cities: Puyang, Jiaozuo, Luoyang, Jiyuan, Xinxiang, Sanmenxia, Kaifeng, and Zhengzhou, encompassing an area of 36,000 km². The region is densely populated, a significant agricultural area in China, is the core of the Central Plains Urban Agglomeration [33,34], and also holds significant ecological functions within the Yellow River Basin. For the past few years, rapid urbanization and industrial growth have increased energy demand, placed pressure on water resources, and reduced the availability of cultivable land. These factors have exacerbated conflicts over water, energy, and grain resources, limiting the progress of national strategies concerning the HYRB [35]. In this study, we modeled and quantified the ESSD, including WY, CS, and FP, in the HYRB region under the SSP-RCP scenario framework. The study’s aims were as follows: (1) Using the InVEST model, we calculated the supply and demand of multiple ESs in HYRB 2020. (2) The PLUS model was utilized to project land-use distributions for different scenarios for 2030 and 2050 within the study area. Subsequently, the InVEST model was coupled with relevant statistical models, including geographically weighted regression (GWR), to predict and analyze ESSD in various scenarios. (3) Derived from the supply–demand ratio of ESs and the bivariate local spatial autocorrelation metric, we examined the spatial and temporal matching of ESSD across different scales. This study serves as a critical reference for implementing ecological conservation and high-quality development strategies in the HYRB, delivering empirical evidence and theoretical frameworks for basin-scale ecosystem management. Furthermore, it advances methodological innovations in quantifying spatiotemporal ESSD dynamics and their alignment under SSP-RCP scenarios, thereby establishing a robust scientific basis for optimizing ecological security pathways and sustainable development policies in comparable river systems globally.

2. Materials and Methods

2.1. Study Area

The HYRB section is situated in the northern–central region of Henan Province, spanning latitudes 33°41′–36°6′N and longitudes 110°21′–116°6′E (Figure 1). The river traverses eight cities: Puyang, Jiaozuo, Luoyang, Jiyuan, Xinxiang, Sanmenxia, Kaifeng, and Zhengzhou. The whole length is 711 km, encompassing an area of approximately 36,200 km². This area lies in the transition between China’s second and third topographic layers, marked by considerable height differences, with elevated terrain in the west and lower elevations in the east. The research area exhibits a characteristic continental monsoon climate, with four distinct seasons. As of the end of 2020, the population of HYRB was around 40.83 million, or 41.11% of Henan province’s total population. The urbanization rate was 59.93%, exceeding the provincial average of 55.43%. The GDP of HYRB attained CNY 2.68 trillion, accounting for 48.72% of the province’s GDP. For the past few years, accelerated urbanization and industrialization have resulted in heightened energy demand, significant imbalances in water supply and demand, and a decrease in cultivated land. These issues have caused serious conflicts over water, energy, and food resources, obstructing the execution of the national policy for ecological preservation and superior development in the HYRB.

2.2. Data Sources and Pre-Processing

This study primarily employs natural environmental and socio-economic information. The natural environmental data include land use, DEM, carbon density, soil depth, NDVI, evapotranspiration, and meteorological and climatic data. The socio-economic information primarily consists of population density, GDP, water demand, and crop yield. Detailed data sources and descriptions are provided in

Table 1. Data and descriptions used in this study.

Data		Spatial Resolution	Year	Source
Natural environment data	Land use/cover	30 m	2000 2005 2010 2020	Resource and Environmental Science and Data center, Chinese Academy of Sciences (https://www.resdc.cn, accessed on 15 March 2024)
	DEM	30 m	-	Geospatial Data Cloud (https://www.gscloud.cn, accessed on 15 March 2024)
	Urban built-up area	1 km	2020	National Tibetan Plateau Data Center (https://www.tpdc.ac.cn, accessed on 15 March 2024)
	Carbon Density	-	-	National Ecological Science Data Center (https://www.nesdc.org.cn, accessed on 15 March 2024)
	Precipitation	1 km	2020	Space-time Tri-polar Environmental Big Data Platform (https://loess.geodata.cn, accessed on 15 March 2024)
	Potential evapotranspiration	1 km	2020	National Earth System Science Data Center (https://www.geodata.cn accessed on 15 May 2024)
	Soil depth	90 m	-	National Earth System Science Data Center (https://soil.geodata.cn, accessed on 15 May 2024)
	NDVI	1 km	2020	Resource and Environmental Science Data Platform (https://www.resdc.cn, accessed on 15 May 2024)
	Future precipitation	1 km	2030 2050	https://www.worldclim.org, accessed on 15 May 2024
	Future potential evapotranspiration	1 km	2030 2050	National Tibetan Plateau/Third Pole Environment Data Center (https://www.tpdc.ac.cn/home, accessed on 15 May 2024)
	Future land use simulation	1 km	2030 2050	https://figshare.com, accessed on 15 May 2024)
Socio-economic data	Carbon emissions	-	2020	China Emissions Accounts and Datasets (CEADs) (https://www.ceads.net.cn, accessed on 15 October 2024)
	Population density	100 m	2020	WorldPop Dataset (https://www.worldpop.org, accessed on 15 October 2024)
	Water demand crop production	-	2020	“Henan Statistical Yearbook”, “County Statistical Yearbook”, “Henan Province Water Resources Bulletin” and Water Resources Bulletin of Each City
	Future population density	1 km	2030 2050	National Tibetan Plateau/Third Pole Environment Data Center (https://www.tpdc.ac.cn/home, accessed on 15 October 2024)
	Future GDP	1 km	2030 2050	https://zenodo.org, accessed on 15 October 2024

. All data were preprocessed using ArcGIS 10.8, which involved projection transformation, clipping, extraction by mask, and resampling. The datasets were standardized to a uniform spatial resolution of 100 m × 100 m. Specifically, down-scaling for coarse resolution data, which was achieved by applying bilinear interpolation to datasets with resolutions coarser than 100 m (e.g., 1000 m), ensuring smooth transitions between pixels. Up-scaling for fine-resolution data was achieved by using the nearest neighbor method to aggregate datasets with resolutions finer than 100 m (e.g., 30 m), preserving original data integrity.

2.3. Methods

To explore the future complexities of supply and demand matching in the HYRB, this study follows several main steps (Figure 2). First, the PLUS model utilizes various scenario parameters to project land use shifts in the study region for 2030 and 2050. Based on these simulations, quantitative predictions are made regarding the distribution of ESs, with a focus on WY, CS, and FP. A thorough evaluation of supply–demand dynamics for ESs in the region is performed over the timeframe spanning 2020 to 2050, using supply–demand matching metrics and spatial autocorrelation models.

2.3.1. Land Use Simulation

This study employed the patch-generating land use simulation (PLUS) model to simulate land-use change patterns in the HYRB for 2030 and 2050 in various scenarios. The PLUS model incorporates an innovative approach to land expansion analysis along with a cellular automaton (CA) model that utilizes diverse random patch seeds. This combination facilitates the simulation of the creation and development of different land-use patches across both time and space. In comparison to alternative models, the PLUS model is superior in pinpointing the factors influencing land-use changes over particular time periods, offering higher precision in simulation results.

First, land expansion data were extracted using land use data from the years 2000 and 2010. Next, 10 driving factors were selected based on natural, socioeconomic, and accessibility factors (Table S1). The LEAS module, using a random forest algorithm, was employed to create development potential maps for different land use types in the study area. On this basis, the 2010 land use data of the study area was used in the CARS module to combine the relevant parameters such as land use demand and neighborhood weight (Table S2) to obtain the 2020 land use simulation map of the study area [36], which was juxtaposed with the real land use pattern in 2020. The accuracy of the PLUS model was validated, yielding a Kappa coefficient of 0.93 and an overall accuracy of 0.95, signifying the model’s strong reliability. Finally, with 2020 land use as the baseline, and using a land use transition matrix (Table S3), parameters corresponding to various scenarios were input to simulate land use changes for 2030 and 2050.

$${\mathrm{I}}_{\mathrm{P}}^{\mathrm{t}}\left\{\begin{array}{c}{\mathrm{I}}_{\mathrm{P}}^{\mathrm{t}-1},\left|{\mathrm{D}}_{\mathrm{P}}^{\mathrm{t}-1}\right|\le \left|{\mathrm{D}}_{\mathrm{P}}^{\mathrm{t}-2}\right|\\ {\mathrm{I}}_{\mathrm{P}}^{\mathrm{t}-1}\times {\displaystyle \frac{{\mathrm{D}}_{\mathrm{P}}^{\mathrm{t}-2}}{{\mathrm{D}}_{\mathrm{P}}^{\mathrm{t}-1}}},0>{\mathrm{D}}_{\mathrm{P}}^{\mathrm{t}-2}>{\mathrm{D}}_{\mathrm{P}}^{\mathrm{t}-1}\\ {\mathrm{I}}_{\mathrm{P}}^{\mathrm{t}-1}\times {\displaystyle \frac{{\mathrm{D}}_{\mathrm{P}}^{\mathrm{t}-1}}{{\mathrm{D}}_{\mathrm{P}}^{\mathrm{t}-2}}},0>{\mathrm{D}}_{\mathrm{P}}^{\mathrm{t}-1}>{\mathrm{D}}_{\mathrm{P}}^{\mathrm{t}-2}\end{array}\right.$$

(1)

where ${\mathrm{I}}_{\mathrm{P}}^{\mathrm{t}}$ denotes the inertia coefficient of land use type p at moment t of iteration, and ${\mathrm{D}}_{\mathrm{P}}^{\mathrm{t}-1}$ is the difference between the demand for land use type p and the actual amount of land use at moment t − 1.

The simulation results for the period 2020–2050 show that, under the SSP126 scenario (Figure S1), strict ecological protection policies lead to the largest increase in forested land, rising from 12,292 km² in 2020 to 13,744 km² in 2050, an increase of 1492 km² (11.8%). Under the SSP245 scenario, the extent of arable land attains its maximum, totaling 31,779 km² in 2050. In the SSP5-8.5 scenario, propelled by the swift advancement of the secondary and tertiary sectors, along with changes in dietary structure and consumption patterns, the land area designated for construction expands significantly by 7010 km² from 2020 to 2050.

2.3.2. Estimates of ESSD

• Water yield

This study employs the InVEST model’s water yield module to measure the ecosystem’s water yield in the research area (Tables S4 and S5). The calculation formula is as follows:

$$\mathrm{Supply}: \mathrm{y}\left(\mathrm{x}\right)=\left(1-{\displaystyle \frac{{\mathrm{AET}}_{\mathrm{x}}}{{\mathrm{P}}_{\mathrm{x}}}}\right)\times {\mathrm{P}}_{\mathrm{x}}$$

(2)

In the above formula, $\mathrm{y}\left(\mathrm{x}\right)$ represents the water yield (m³) for grid cell x, ${\mathrm{AET}}_{\mathrm{x}}$ is the actual annual evapotranspiration for cell x, and ${\mathrm{P}}_{\mathrm{x}}$ is the annual precipitation of pixel x (mm/yr). The WY for 2030 and 2050 is calculated based on projected precipitation, evapotranspiration, and land use under different scenarios.

Water demand is calculated based on water use data from Henan Province’s socio-economic system. The study area’s water usage is divided into three primary categories: agricultural, industrial, and domestic. These categories are then allocated to raster layers of cropland, GDP, and population density to generate a spatial distribution map of water demand. The formula for calculation is as follows:

$$\mathrm{Demand}:{ \mathrm{W}}_{\mathrm{D}}={\mathrm{WY}}_{\mathrm{agr},\mathrm{d}}+{\mathrm{WY}}_{\mathrm{dom},\mathrm{d}}+{\mathrm{WY}}_{\mathrm{ind},\mathrm{d}}$$

(3)

In the above formula, ${\mathrm{W}}_{\mathrm{D}}$ represents the water demand for grid cell D, which is calculated as the sum of agricultural, industrial, and domestic water use, based on cropland, GDP, and population raster data. ${\mathrm{WY}}_{\mathrm{agr},\mathrm{d}}$, ${\mathrm{WY}}_{\mathrm{dom},\mathrm{d}}$, and ${\mathrm{WY}}_{\mathrm{ind},\mathrm{d}}$ represent the agricultural, domestic, and industrial water use, respectively. The demand for water yield services is predicted using the GWR model [37]. The core objective of this study using GWR is to reveal the spatially non-stationary relationship between water demand and driving factors in HYRB. Compared with traditional global regression models (e.g., OLS) that assume the regression coefficients to be spatially constant, the GWR allows the influence weights of population density, GDP, and farmland area on water demand to change dynamically with geographic location, which offers a high degree of compatibility with the HYRB’s characteristics of significant topographic gradients and strong spatial heterogeneity of socioeconomic activities. Specifically, derived from the evaluation of water yield requirements in 2020, the GWR method is used to derive the local regression functions of water demand for each county with respect to population density, GDP, and cropland area. The model’s accuracy was assessed through cross-validation, yielding a strong R² value of 0.94, indicating a high degree of predictive accuracy (Table S6) [38]. It is assumed that water use efficiency will remain constant from 2020 to 2030 and 2050.

2.

Food production

Food production and provisioning services are the ability of ecosystems to supply food crops for human beings and are the foundation for human existence and advancement. It not only sustains regional food security and social stability while ensuring fundamental livelihood requirements, but is also a significant determinant of human well-being. The study utilized NDVI and land use/cover data, and the total one-year production of arable land (comprising vegetables, wheat, rice, oilseeds, and maize, etc.) was allocated to each raster according to the ratio of the raster NDVI to the total NDVI of the arable land to represent the food availability in each 100 m × 100 m arable land raster. This study operates under two critical assumptions: (1) the Normalized Difference Vegetation Index (NDVI) remains constant from 2020 to 2030 and 2050, and (2) key determinants influencing grain yields (e.g., technological advancements, climate shifts, or policy interventions) maintain relative stability throughout the projection period. The specific formula is as follows [39]:

$$\mathrm{Supply}:{ \mathrm{FS}}_{\mathrm{x}}={\displaystyle \frac{{\mathrm{NDVI}}_{\mathrm{x}}}{{\mathrm{NDVI}}_{\mathrm{sum}}}}\times {\mathrm{S}}_{\mathrm{sum}}$$

(4)

where ${\mathrm{FS}}_{\mathrm{x}}$ is the food supply on grid x (t), ${\mathrm{NDVI}}_{\mathrm{x}}$ is the NDVI value on grid x, ${\mathrm{NDVI}}_{\mathrm{sum}}$ is the sum of NDVIs of the cropland, and ${\mathrm{S}}_{\mathrm{sum}}$ is the total food production of the cropland (t). As the actual food supply in the forecast year is not available, this study characterizes the supply capacity of cropland in the forecast year by linear regression analysis of the yield per unit area of cropland in previous years.

The computation of food demand is predicated on the methodology of “people-based demand”, and food demand is expressed by food consumption. Taking into account the population density and individual food consumption of city and rural dwellers, the food demand for urban and rural populations is determined using the following formula:

$$\mathrm{Demand}:{ \mathrm{FP}}_{\mathrm{x}}^{\mathrm{d}}=\left\{\begin{array}{cc}{\mathrm{P}}_{\mathrm{t}}\times {\mathrm{AF}}_{\mathrm{urban}}& \mathrm{In} \mathrm{urban} \mathrm{areas}\hfill \\ { \mathrm{P}}_{\mathrm{t}}\times {\mathrm{AF}}_{\mathrm{rural}}& \mathrm{In} \mathrm{rural} \mathrm{areas}\hfill \end{array}\right.$$

(5)

In the above equation, ${\mathrm{FP}}_{\mathrm{x}}^{\mathrm{d}}$ denotes the food demand of raster cell x, and ${\mathrm{AF}}_{\mathrm{urban}}$ and ${\mathrm{AF}}_{\mathrm{rural}}$ are the respective per capita food demands in urban and rural areas of Henan Province in 2020 from the China Statistical Yearbook, which are 136.6 kg and 163.4 kg, respectively. It is assumed that the demand for food by urban and rural dwellers remains constant from 2020 to 2030 and 2050. The food demands of 2030 and 2050 were projected based on the population projections of different scenarios.

3.

Carbon storage

Terrestrial ecosystem CS is essential in alleviating climate change. It is not only an essential component of the Earth’s overall carbon storage but also provides a foundation for assessing the impact of climate change on ecosystem productivity. The formula for calculating the carbon module in the InVEST model is as follows [40]:

$$\mathrm{Supply}:{ \mathrm{CS}}_{\mathrm{x}}^{\mathrm{s}}={\mathrm{C}}_{\mathrm{iabove}}+{\mathrm{C}}_{\mathrm{ibelow}}+{\mathrm{C}}_{\mathrm{isoil}}+{\mathrm{C}}_{\mathrm{idead}}$$

(6)

where ${\mathrm{CS}}_{\mathrm{x}}^{\mathrm{s}}$ is the total supply of CS at pixel x; ${\mathrm{C}}_{\mathrm{iabove}}$ is the above-ground biomass carbon pool (t/km²); ${\mathrm{C}}_{\mathrm{ibelow}}$ is the below-ground biomass carbon pool (t/km²); ${\mathrm{C}}_{\mathrm{isoil}}$ is the soil organic carbon pool (t/km²); and ${\mathrm{C}}_{\mathrm{idead}}$ is the dead organic matter carbon pool (t/km²) (Table S7). CS services in 2030 and 2050 were based on the different scenarios of land-use data that were calculated.

The actual CO₂ emissions usually represent the demand for CS, and the per capita carbon emissions and grid population density are utilized to evaluate the demand for CS. The 2020 carbon emission data of counties (districts) were extrapolated by linear fitting through CEADs data [41], with the R² value exceeding 0.85 (Table S8). The formula for its calculation is as follows.:

$$\mathrm{Demand}:{ \mathrm{CS}}_{\mathrm{x}}^{\mathrm{D}}={\mathrm{CS}}_{\mathrm{x}}\times {\displaystyle \frac{{\mathrm{Pop}}_{\mathrm{i}}}{{\mathrm{Pop}}_{\mathrm{x}}}}$$

(7)

where ${\mathrm{CS}}_{\mathrm{x}}^{\mathrm{D}}$ represents the CS demand for grid cell x, ${\mathrm{CS}}_{\mathrm{x}}$ is the carbon emissions of county x, ${\mathrm{Pop}}_{\mathrm{x}}$ is the total population of county x, and ${\mathrm{Pop}}_{\mathrm{i}}$ is the population of pixel i. CS demand for 2030 and 2050 is estimated based on population projections under different scenarios. It is assumed that the per capita carbon emissions remain constant from 2020 to 2030 and 2050.

2.3.3. Matching Supply and Demand for ESs

The ecosystem service demand–supply ratio (ESDR) is utilized to evaluate the spatial differences between the availability and need for ESs, which helps illustrate both the spatial congruence and variances in supply and demand for each ES [42]. Calculation is performed using the following formula:

$$\mathrm{ESDR}={\displaystyle \frac{{\mathrm{supply}}_{\mathrm{actual}}-{\mathrm{demand}}_{\mathrm{human}}}{\left({\mathrm{supply}}_{\max }+{\mathrm{demand}}_{\max }\right)/2}}$$

(8)

where ${\mathrm{supply}}_{\max }$ and ${\mathrm{demand}}_{\max }$ represent the maximum values of real ecosystem service provision and human demand, respectively, inside the research region. ${\mathrm{supply}}_{\mathrm{actual}}$ and ${\mathrm{demand}}_{\mathrm{human}}$ denote the supply of ecosystem services and human demand, respectively, on the raster cells. $\mathrm{ESDR}$ is greater than 0 to indicate an oversupply of system services, while values of less than 0 indicate a shortage and values of 0 indicate an equilibrium.

Bivariate spatial autocorrelation indices (local indicators of spatial association (LISA)) were utilized to characterize spatial matching patterns of ESs [43]. The supply and demand values of ESs across several scenarios were input into the GeoDa program, a bivariate local spatial autocorrelation analysis was conducted between them, and the LISA clustering map was calculated. Bivariate spatial autocorrelation analysis divides the ESSD into five spatial clusters: low supply–low demand (L–L), low supply–high demand (L–H), high supply–high demand (H–H), high supply–low demand (H–L), and non-significant areas.

3. Results

3.1. Supply and Demand Analysis of ESs in 2020

In the year 2020, the ESSD exhibited significant spatial disparities (Figure 3). The aggregate supply of CS and FP exceeded their total demand, although the aggregate supply of WY fell short of its demand. The total WY supply was 7.24 × 10⁹ m³, with high-value areas predominant in the regions of Sanmenxia, southern and northwestern Luoyang, and the Taihang Mountains. Its spatial distribution demonstrated a general trend of decreasing from southwest to northeast. The total CS supply amounted to 443.09 × 10⁶ t, with high-value areas located in the northern edge’s banded forestlands and the mountainous regions of Sanmenxia and Luoyang. Meanwhile, the total FP supply reached 1.78 × 10⁷ t, with high-value regions mainly distributed in the plains outside urbanized zones, where flat terrain and fertile soil favor crop cultivation. The total demand for the three services was 1.03 × 10⁷ m³ (WY), 2.7 × 10⁸ t (CS), and 3.74 × 10⁶ t (FP), High-demand areas are regularly found in highly populated metropolitan centers.

3.2. Forecasts of ESSD for Different Scenarios, 2020–2050

We simulated and calculated the ESSD under various SSP-RCP scenarios (

Table 2. Scenarios forecast of total ESSD in HYRB, 2030-2050.

Year	Scenarios	Water Yield		Carbon Storage		Food Production
		Supply/ (×109 m3)	Demand/ (×109 m3)	Supply/ (×106 t)	Demand/ (×106 t)	Supply/ (×106 t)	Demand/ (×106 t)
2030	SSP1-RCP2.6	7.4	12.1	446.9	324.1	24.3	4.1
	SSP2-RCP4.5	7.1	12.1	439.6	333.5	25.3	4.7
	SSP5-RCP8.5	8.5	12.5	437.6	326.7	24.9	4.6
2050	SSP1-RCP2.6	8.7	13.6	451.1	359.2	23.3	4.4
	SSP2-RCP4.5	2.9	13.7	445.1	387.7	28.4	5.4
	SSP5-RCP8.5	5.7	15.2	407.2	364.8	25.3	5.1

). Between 2020 and 2050, WY and CS exhibited continuous growth under the SSP1-2.6 scenario, increasing by 2.4% and 0.8% from 2020 to 2030, and by 17.8% and 0.9% from 2030 to 2050, respectively. Under the SSP2-4.5 scenario, CS grew slightly by 0.4%, while WY showed a declining trend, decreasing by 59.9%. In the SSP5-8.5 scenario, both WY and CS experienced sustained declines of 21.9% and 8.1%, respectively, between 2020 and 2050. In contrast, FP supply exhibited continuous growth under all three scenarios, with the largest increase of 59.8% observed under the SSP2-4.5 scenario. Regarding demand, WY, CS, and FP all showed a continuous upward trend between 2020 and 2050. The SSP2-4.5 scenario recorded the highest growth in demand for CS (44.1%) and FP (45.1%), while the SSP5-8.5 scenario saw the most significant rise in WY demand at 47.2%.

In future scenarios, the ESSD exhibited significant spatial heterogeneity (Figure 4 and Figure 5). The spatial distribution of WY supply showed considerable variation, with high-value areas concentrated in the northwest, southwest, and central mountainous regions. FP supply exhibited a distribution of elevated values in the northeast and diminished values in the southwest, with high-value regions predominantly situated in the eastern and northern plains. Conversely, the CS supply demonstrated elevated values in the southwest and diminished values in the northeast, with high-value areas centered in the northwest, southwest, and central mountainous regions. The regional patterns of demand for the three ESs were analogous, with high demand areas mainly located in densely populated urban centers.

3.3. Matching ESSD in Different Scenarios, 2020–2050 Analysis

The supply–demand balance of ESs in HYRB displayed notable spatiotemporal mismatches under different scenarios (Figure 6). The surplus regions for WY and CS were centered in the forested portions of the northwest, southwest, and central mountainous zones, whilst the deficit regions were predominantly situated in highly urbanized areas. Conversely, FP surplus regions were predominantly located in the croplands of the eastern and northern plains, whilst deficit areas were chiefly found in non-cropland zones. Between 2020 and 2050, WY had the largest deficit area among the three services, which continued to increase over time. In the SSP2-4.5 scenario, the deficit area reached its maximum in 2050, with an increase of 27.35%, covering 71.04% of the basin’s total area. Similarly, the deficit areas of CS and FP expanded under all three scenarios during the same period, with the SSP5-8.5 scenario recording the largest increases of 29.3% and 56.1%, respectively.

At the county level (Figure 7), the deficit areas for the three services were mainly distributed in Xigong District and Jianxi District of Luoyang City, as well as Jinshui District and Erqi District of Zhengzhou City. Surplus areas for WY and CS were concentrated in Luanchuan County and Song County of Luoyang City and Lushi County of Sanmenxia City, while FP surplus areas were primarily located in Fengqiu County of Zhoukou City and Yuanyang, Yanjin, and Huojia Counties of Xinxiang City. Compared with 2020, the average supply–demand ratios for WY, CS, and FP reached their maximum values in 2030 under different scenarios: −0.01 for WY under SSP5-8.5, −0.036 for CS under SSP1-2.6, and 0.076 for FP under SSP2-4.5. By 2050, the highest average supply–demand ratios for WY and CS were observed under SSP5-8.5 (−0.009) and SSP1-2.6 (−0.060), respectively, while FP reached its peak under SSP1-2.6 (0.139). From 2020 to 2050, the average supply–demand ratios of WY and CS showed a declining trend under all three scenarios. The most significant decrease for WY occurred under SSP2-4.5, with an average reduction of 0.015, while CS experienced its largest decline under SSP5-8.5, with an average decrease of 0.044. In contrast, FP exhibited continuous growth across all scenarios, particularly under SSP1-2.6, where the average value increased by 0.071 by 2050.

From 2020 to 2050, the geographical alignment patterns of supply and demand for the three services displayed significant differences (Figure 8). For WY, the dominant cluster type was the high–low cluster, primarily distributed in the northwest, southwest, and central mountainous regions. Notably, under SSP2-4.5 and SSP5-8.5 in 2050, the low–low cluster areas expanded to some extent. CS were characterized mainly by high–low clusters, concentrated in the hilly areas of the northwest and southwest. Conversely, FP largely demonstrated low–low clusters, chiefly situated in the southern and northwestern hilly portions of the research area.

4. Discussion

4.1. Comparison with Other Studies

HYRB is situated in the transitional area between the middle and lower sections of the Yellow River and is densely populated, is a significant primary agricultural zone in China and serves as the central area of the Central Plains Urban Agglomeration. Recent years have witnessed growing urbanization and industrialization, resulting in disputes over water, energy, and food. Therefore, simulating the matching of supply and demand under different scenarios in the future is of great significance for realizing sustainable ecosystem management and advancing environmental conservation and sustainable growth in the HYRB. In this study, we integrated multi-source data and refined localized parameters to effectively enhance the accuracy of ecosystem service estimations. Specifically, our approach to assessing ecosystem service demand differs from previous studies, which often relied on simplistic calculations using indicators such as land development intensity, population density, and GDP [44,45]. Instead, we adopted distinct quantitative metrics based on the actual consumption of different ESs to improve the precision of our results. Furthermore, while most existing supply–demand studies are grounded in historical evolutionary characteristics [46], few have explored future supply–demand matching, and even fewer have accounted for more than one influencing factor. Many such studies consider only land-use changes [47]. In contrast, our research couples land-use change with climate change to simulate and predict ESSD, thereby increasing both the accuracy and depth of supply–demand studies.

The supply and demand results indicate that water production services are significantly influenced by climate change, through substantial variations in rainfall amounts and patterns across different scenarios, which aligns with findings from other studies. By 2050, under the SSP126 scenario, WY and CS values reach their maximum, indicating that, within a sustainable development scenario, the government may enact conservation regulations to restrict land use intensity, thus effectively safeguarding natural ecosystems such as forests and grassland. In the SSP585 scenario, the growth of built-up areas results in a steady decrease in CS from 2020 to 2050, aligning with previous research [48]. In the SSP245 scenario, FP supply is the largest due to the gradual reduction of arable land. The demand for all three ESs is anticipated to rise across all scenarios, with small variations.

The outcomes of supply–demand matching analysis reveal a distinct spatial mismatch between supply and demand for WY, CS and FP. Surplus areas for WY and CS are primarily concentrated in mountainous and forested regions, which are sparsely populated and rich in natural resources, while surplus areas for FP are mainly found in plains. Wang et al. [49] conducted a univariate analysis on seven factors influencing the spatial differentiation and temporal variation of carbon stocks. Their results show that areas with abundant vegetation cover exhibited higher carbon sequestration potential compared with other land types. Meanwhile, population density was identified as a key indicator of human activities and had the strongest impact on carbon stocks among anthropogenic factors, which aligns with our findings. Deficit areas for all three services are concentrated in the construction zones with high human activities, congruent with the results of previous studies [50]. From 2020 to 2050, the deficit areas of CS and FP continued to expand, with the most substantial shortfall seen under the SSP585 scenario [51]. During the same period, the deficit areas for WY increase in both the SSP245 and SSP585 scenarios but show an initial rise before a fall in the SSP126 scenario. Overall, the ES deficit area is relatively small in the SSP126 scenario, while it is larger in the SSP585 scenario and across all three scenarios.

4.2. Implications for Ecological Management

The HYRB is among the water-deficient areas in central China, where studies have highlighted a significant water resource gap [52]. While climate change may increase precipitation, potentially alleviating water shortages in areas such as Kaifeng and Luoyang under the SSP245 scenario, water scarcity in the HYRB is expected to increase significantly, posing a major challenge for future sustainable development. Spatial cluster analysis of ESDR distribution and supply–demand matching has highlighted the ecological significance of the forested areas in the northwest, southwest, and central parts of the study area, as well as the agricultural importance of the grain-producing plains in the east. Based on the specific conditions of each region, differentiated ecological management strategies have been proposed. For the forested regions in the northwest, southwest, and central parts of the HYRB, where supply exceeds demand, an ecology-first approach should be maintained, optimizing ecological space. Rehabilitation of the biological barriers in the Taihang Mountains to the northwest and the Funiu Mountains to the southwest, along with the protection of the ecological core of the Song Mountains in the central region, should be strengthened to form an “ecological safety framework” of “one core, two barriers”. Strengthening the protection of natural forests, optimizing vegetation structure, and enforcing ecological protection redlines will enhance the water conservation function of ecosystems, ensuring the provision of high-quality ecosystem services. For water bodies with high ecological service supply, an “ecological corridor network” of “one belt with multiple corridors” should be established. This includes implementing the “Yellow River Basin Ecological Protection and High-Quality Development Plan” and preserving the national ecological belt along the Yellow River mainstream, as well as the ecological corridors of the Shaying and Yiluo rivers, to form a regional ecological protection network and improve ecosystem quality. Additionally, efforts should be made to advance ecological management along both banks of the Yellow River, restore the ecological conditions of floodplain areas, and protect small watersheds, thereby improving the overall ecological health of the basin.

As the primary agricultural production area, the eastern plains must strictly protect the basic farmland redline and prevent the conversion of farmland for non-agricultural purposes. Scenario simulations suggest that, while the area of farmland decreases, FP will continue to grow, emphasizing the importance of increasing yields on limited farmland. To achieve this, efforts should focus on constructing high-standard farmland in concentrated, well-equipped zones, improve soil quality, enhance resilience to disasters, and create permanent farmland suited to modern agricultural practices. Advancing breeding technology and fostering innovation will help develop high-yield, stable, and resilient crop varieties. Establishing quality seed breeding bases will support the modernization of the seed industry and strengthen food production foundations. Additionally, farmers should be educated on the scientific use of pesticides, fertilizers, and plastic films to reduce soil pollution and protect microbial activity. Increased investment in research on safe pesticides will help improve soil conditions, optimizing food production on limited farmland.

In the study area, ecosystem service deficits are concentrated in highly urbanized regions, where the supply–demand imbalance is particularly pronounced. In these areas, it is imperative to prioritize ecological factors while facilitating coordinated development. Efforts should focus on overcoming administrative boundaries to enhance regional ecological connectivity. This includes improving ecological restoration, expanding green infrastructure, and advancing urban greening to improve the supply of ESs within cities. Simultaneously, a sound ecological compensation mechanism must be established, with increased financial investment in ecological protection projects. Economic incentives should be leveraged to promote environmental governance and encourage local governments to act. Strengthening penalties for environmental harm and fostering sustainable industries will be essential for supporting high-quality development in the HYRB.

4.3. Limitations and Improvements

This research examines the alignment of ES supply and demand in various future scenarios; however, notable limitations remain. First, this study predominantly employed the InVEST model to assess ES supply. While the InVEST model is widely recognized for its applicability in multi-scale ES assessments, it has certain limitations [53]. For example, although the InVEST model has high application value in estimating carbon stocks, its assumption of constant future carbon density may be an oversimplification. As land-use changes and changes in carbon density directly affect the prediction of carbon stocks, future studies should consider dynamic changes in carbon density and improve the model to reflect these changes more accurately. Furthermore, future research should adopt an integrated approach from the perspective of ecosystem structure, function, and services, incorporating land surface process models (e.g., Jules model) to provide more accurate estimations of ESs based on ecosystem mechanisms. This would enhance the precision of estimations and offer stronger support for future studies. Second, the method used to calculate ecosystem service demand in scenario projections is relatively simple, with limited factors considered. For example, the study’s assumptions regarding constant water efficiency and linear regression for food production could introduce uncertainties in future projections. Specifically, the assumption that water efficiency will remain constant does not account for potential technological advancements in water-saving practices, which could affect future water demand. Similarly, the use of linear regression to predict food production may not capture future changes in crop yields due to climate change, technological improvements, or policy interventions. These assumptions should be revisited in future studies to incorporate a range of scenarios that reflect potential changes in water efficiency and crop productivity. Future studies should account for more variables, such as ecosystem complexity, uncertainties, and socioeconomic and cultural factors, to improve the accuracy of demand calculations. Moreover, discrepancies in historical and prospective socioeconomic data, the inadequate geographic resolution of meteorological data in the SSP-RCP scenarios, and the application of diverse models may engender uncertainty in the results. Consequently, subsequent research must concentrate on acquiring higher-resolution data and enhancing model precision to tackle these issues.

5. Conclusions

This study quantitatively simulated and predicted the supply and demand of key ESs in the HYRB from 2020 to 2050 under various scenarios. The main findings are as follows: (1) Spatiotemporal heterogeneity of ESSD: Significant spatiotemporal disparities were observed in ESSD distributions. From 2020 to 2050, CS and FP consistently exhibited supply surpluses, whereas WY demonstrated persistent supply deficits. High-capacity WY supply zones clustered in Sanmenxia City, southern/northwestern Luoyang, and the Taihang Mountain range, displaying a southwest-to-northeast declining gradient. CS hotspots predominantly occurred in northern forested zones and the mountainous territories of Sanmenxia–Luoyang, while FP production cores were concentrated in peri-urban plains. Demand for all three services was highest in urban centers. (2) Scenario-dependent supply–demand trajectories: temporal analysis revealed divergent trajectories: WY and CS supply–demand ratios exhibited progressive decline across scenarios, contrasting with FP’s sustained growth. Surplus distributions demonstrated spatial specificity: WY and CS surpluses aggregated in northwestern, southwestern, and central highland regions, whereas FP surpluses dominated eastern and northern plains. Deficit zones for all ESs consistently concentrated in urban cores characterized by intensive anthropogenic activity. Distinct spatial autocorrelation patterns emerged: WY manifested high–low clustering in northwestern, southwestern, and central mountainous regions and CS displayed analogous high–low clustering, which was particularly pronounced in southwestern/northwestern hilly terrain. Conversely, FP exhibited low–low clustering predominance within southwestern/northwestern upland areas. (3) Spatial association typology variations: the dominant spatial matching types showed notable variation. WY mainly displayed a high–low clustering pattern, mainly distributed in the northwest, southwest, and central mountainous regions. CS displayed a similar high–low clustering pattern, focused in the mountainous areas of the southwestern and northwestern regions. In marked contrast, FP spatial organization adhered to a low–low clustering model, predominantly localized in southwestern/northwestern mountainous zones.