Spatiotemporal Dynamics of Annual Precipitation and Future Projections of China’s 400 mm Isohyet

Highlights

What are the main findings?

• From 2001 to 2017, China’s precipitation generally increased, rising in the southeast while decreasing in Southwest China.
• The 400 mm isohyet moved northwestward in East China and northeastward in West China. This overall migration is projected to continue westward and northward by 2100.

What is the implication of the main finding?

• The study provides crucial insights into the spatial distribution and dynamic changes of China’s precipitation patterns.
• The study provides valuable theoretical support for regional water resource planning and decision-making, promoting sustainable agricultural development, and ensuring ecological balance.

Abstract

The 400 mm isohyet in China serves as a critical geographical demarcation of dry and wet regions. Amidst intensifying global warming, this climatic boundary has undergone notable shifts, with significant implications for China’s agriculture, water resources, and ecosystems. This study integrates meteorological station data, the China Gridded Daily Precipitation dataset (CN05.1), and Integrated Multi-satellite Retrievals for the Global Precipitation Measurement (GPM IMERG) satellite observations to assess the spatiotemporal distribution of precipitation across mainland China and analyze the migration trend of the 400 mm isohyet. Furthermore, utilizing outputs from five models of the Coupled Model Intercomparison Project Phase 6 (CMIP6), we projected future trends of China’s annual mean precipitation and the 400 mm isohyet’s migration under three Shared Socioeconomic Pathways (SSPs: low, medium, and high radiative forcing scenarios) until the end of this century (2100). Results reveal that from 2001 to 2017, the 400 mm isohyet exhibited a prominent northwestward migration trend. This trend is projected to continue in the future. These findings provide a crucial reference for understanding the spatial distribution and changing dynamics of precipitation patterns in China, offering vital decision support for land resource planning and water resource management.

1. Introduction

The Sixth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC) highlights that global average temperatures have risen by approximately 0.9 to 1.2 °C above pre-industrial levels since the Industrial Revolution. This warming trend is projected to intensify further in the coming decades [1]. Amidst global warming, the global hydrological cycle is undergoing profound alterations, often characterized by a pronounced polarization where ‘dry regions become drier and wet regions become wetter [2,3]. As a crucial component of this cycle, precipitation profoundly influences surface and groundwater resources, as well as the structure and functions of regional ecosystems [4]. Consequently, accurately assessing these changes in precipitation holds paramount importance for China’s socio-economic development [5].

The migration of climatic demarcation lines, including isohyets, represents a significant indicator of regional climate change and has garnered substantial research attention [6,7]. In China, the 400 mm isohyet serves as a pivotal geographic demarcation, delineating the boundary between semi-humid and semi-arid zones. Beyond its climatic definition, this line inherently marks crucial ecological, agricultural, and socio-economic transitions, such as the shift from forest to grassland, from farming to nomadic practices, and from cultivation to pastoralism [8]. In recent years, intensified global warming has driven a notable northwestward movement of this important climatic boundary [9]. This observed shift raises critical questions regarding the continued efficacy of the static 400 mm isohyet as a functional boundary for the aforementioned ecological and societal transitions. Indeed, recent eco-hydrological modeling suggests that, under mid-century warming scenarios consistent with SSP245, the functional dry–wet boundary in northern China may be more accurately represented by an annual precipitation band of 430–450 mm, rather than the traditional 400 mm threshold [10,11]. Further supporting this, Huang et al. [12] demonstrated that every 1 °C of warming tends to shift the semi-humid/semi-arid boundary by 15–20 km per year towards wetter values, based on analyses of 35 regional studies from 1991 to 2020. An accurate and dynamic analysis of the 400 mm isohyet’s actual migration, rather than just its static definition, becomes an indispensable reference for understanding and managing changes in China’s ecological landscape.

Building upon this recognition, numerous previous studies have explored various aspects of isohyet changes in China. For instance, Che et al. [13] analyzed the spatial changes in the 800 mm isohyet in China’s eastern monsoon region (1961 to 2015), reporting a notable eastward and southward movement. Yanhong Gao et al. [8] used three sets of precipitation data from Station Observation Gridded Precipitation (CNgrid), GPCC (Global Precipitation Climatology Center), and CRU (Climate Research Unite) to reveal the 400 mm isohyet general northeast-southwest trend and distinct interannual and interdecadal segmentation characteristics during 1961 to 2019. Li et al. [9] investigated the 400 mm isohyet on the Tibetan Plateau (1982 to 2021), identifying an overall northwestward shift driven primarily by increased precipitation. Zheng et al. [14,15] examined the relationship between alpine meadows, alpine grassland transition zones, and the 400 mm isohyet in northern Tibet, indicating a general westward shift in the isohyet in that region. Despite these valuable regional and thematic studies [16], a comprehensive, in-depth, and quantitative analysis of the 400 mm isohyet’s migration across all of China remains limited, particularly concerning its future projection trends under various climate change scenarios.

To address these gaps and assess future changes, numerous studies have utilized global climate models. The Coupled Model Intercomparison Project Phase 6 (CMIP6) models have demonstrated general reliability in simulating China’s precipitation patterns, albeit with some recognized regional and seasonal variations. For instance, Yang et al. [17] assessed 20 CMIP6 models, concluding that while they accurately reproduce China’s precipitation distribution, their temperature simulation capability generally surpasses that for precipitation. Lu et al. [18] further evaluated the historical performance of 46 CMIP6 Global Climate Models (GCMs), identifying 10 top-performing models that effectively simulate the distribution of both precipitation and temperature, showing strong correlation coefficients (0.8 to 0.99). Additionally, Li et al. [19] found that the CMIP6 multi-model ensemble reasonably simulates current temperature and precipitation extremes. A comparative analysis by Jiang et al. [20] also indicated that CMIP6 models, when compared to CMIP5, exhibit improvements in simulating climatological state temperature, precipitation, and winter winds for the Chinese climate and East Asian monsoon, although with only minor enhancements in interannual variability for temperature, precipitation, and summer winds.

This study focuses on a comprehensive analysis of precipitation patterns in China, with particular emphasis on the 400 mm isohyet. Utilizing a robust dataset comprising measured precipitation data from 2074 stations, the China Gridded Daily Precipitation dataset (CN05.1), and Integrated Multi-satellite Retrievals for the Global Precipitation Measurement (GPM IMERG) satellite observations, alongside CMIP6 model projections, this research systematically investigates the spatiotemporal distribution and future trends of China’s precipitation. Specifically, the objectives of this study are twofold: (1) To analyze the spatial and temporal characteristics of precipitation in China from 2001 to 2017, focusing on the historical trend and migration of the 400 mm isohyet. (2) To predict future precipitation changes using CMIP6 data under various Shared Socioeconomic Pathways (SSPs: SSP126, SSP245, and SSP585) and project the future trends and variability of the 400 mm isohyet’s migration. The findings of this study are anticipated to provide valuable theoretical support for regional water resource planning and decision-making. They contribute significantly to promoting the rational utilization and conservation of water resources, fostering sustainable agricultural development, and ensuring ecological balance. Ultimately, this research holds profound practical implications for enhancing China’s ecological security and supporting stable socio-economic development.

2. Materials and Methods

2.1. Study Area

The study area is located in mainland China, spanning geographic coordinates from approximately 73° to 135°E longitude and 18° to 53°N latitude (Figure 1). Elevation within the study area exhibits significant variation, generally decreasing progressively from west to east. This topographic distinctiveness allows for its broad classification into three major steps or terraces (labeled I, II, and III in Figure 1a, respectively). The first terrace (I) is predominantly represented by the Tibetan Plateau, boasting an average elevation exceeding 4000 m. The second terrace (II) features an average elevation of 1000 to 2000 m, encompassing regions such as Northwest China, North China, and parts of Southwest China. The third terrace is characterized by China’s extensive plains and hilly regions, with relatively lower elevations.

Combining topographic and climatic features, the Chinese mainland region can be further subdivided into eight distinct subregions [21] (Figure 1b,c). The Xinjiang (XJ) region is predominantly characterized by a basin-and-mountain topography, experiencing a temperate continental climate with relatively sparse precipitation. The Tibetan Plateau (TP) is marked by rugged and complex terrain, experiencing generally sparse precipitation and a limited network of observation stations. Conversely, the Southwest (SW) region, dominated by the Yungui Plateau and the Hengduan Mountains, features high and complex topography with numerous canyons, mountains, and basins, along with abundant precipitation. The Northwest (NW) encompasses the Loess Plateau and Inner Mongolia Plateau, including portions of the Gobi Desert, and is characterized by significantly less precipitation. Further to the east, the Northeast (NE) region, situated at a higher latitude, experiences a cold winter climate with precipitation predominantly concentrated during the summer months. Adjacent to it, North China (NC) is characterized by a temperate monsoon climate. Its main terrain is the North China Plain, interspersed with mountainous areas, and it exhibits distinct seasonal precipitation patterns—dry winters and greater rainfall in summer. The East China (EC) region presents a predominantly plain topography, with hilly areas along its coast, and receives sufficient precipitation. Finally, the South China (SC) region is marked by numerous hilly basins and a tropical and subtropical monsoon climate, resulting in high temperatures and exceptionally abundant precipitation.

2.2. Datasets Sources

The precipitation data used in this study include in situ meteorological station observations, the daily gridded precipitation dataset (CN05.1), GPM IMERG remote sensing data, and future precipitation data from CMIP6 models (

Table 1. Precipitation data used in this study.

Data	Data Description	Spatial Resolution	Time Span
Observed precipitation data	2074 meteorological observation stations	Point observations	2001–2017
CN05.1	Gridded data interpolated from over 2416 ground-based meteorological stations in China	0.25° × 0.25°	2001–2017
GPM IMERG	GPM satellite mission	0.1° × 0.1°	2001–2017
CMIP6	The NASA Earth Exchange Global Daily Downscaled Projections	0.25° × 0.25°	Historical data: 1985–2014 Future projections: 2015–2100

). Nevertheless, as our analysis relies solely on the gridded output of CN05.1, it is treated in the same manner as the GPM dataset, rather than as individual station observations. For consistency and comparability across diverse data sources, a common time span of 2001–2017 was uniformly applied to all observational and historical data. Although some datasets extend beyond 2017, the CN05.1 dataset ends in 2017; therefore, to maintain a strict common analysis window across all products, data from the 2020s were not included.

2.2.1. In Situ Station Observations

For in situ station observations, precipitation records from 2074 stably operating stations across China were utilized. These data were extracted from the China Surface Meteorological Dataset (CMDC, 2023), which is managed by the China Meteorological Data Service Center (CMDC) under the China Meteorological Administration (CMA). These daily precipitation records, spanning 2001–2017, feature an observation accuracy of 0.1 mm/day. The stations are geographically distributed across a wide range of environments, from urban to rural, plains to mountains, and coastal to inland regions. Their distribution is notably dense in the southeast, contrasting with lower densities in the northwest and the Tibetan Plateau (Figure 1c).

2.2.2. Gridded Precipitation Data

The gridded precipitation data used in this study include CN05.1 and GPM. The CN05.1 dataset, developed by Wu et al. [22], is a high-resolution gridded product (0.25° × 0.25°) interpolated from observations of 2416 ground-based meteorological stations across China. It is generated using a combination of “distance approximation” and “angular distance weighting method” to ensure uniform data generation even in regions with sparse station networks. Renowned for its high accuracy within China, this dataset has been widely adopted in numerous precipitation analysis [23,24,25].

GPM is an international satellite mission jointly sponsored by NASA and JAXA to provide global precipitation observations. The IMERG algorithm combines information from the GPM satellite constellation to estimate precipitation over a large portion of the Earth’s surface, providing the most advanced precipitation estimation to date, with the finest spatial and temporal resolution to properly estimate and detect regional precipitation patterns and their spatial averages [23,26]. The GPM IMERG Final precipitation product is a high-precision precipitation estimation product generated based on the observation data from the GPM satellite constellation, with a raw spatial resolution of 0.1° × 0.1° and global coverage [27].

It should be noted that the raw station observations represent point-scale measurements, whereas CN05.1 provides spatially continuous gridded fields through interpolation of station data. Consequently, in regions with dense observational coverage, such as eastern China, the differences between the two datasets are minimal. However, in data-sparse regions—particularly over high-altitude areas like the Tibetan Plateau—interpolation inevitably introduces larger uncertainties. Part of the discrepancies observed between the raw station data and the gridded CN05.1 product can therefore be attributed to interpolation uncertainties associated with limited station coverage.

2.2.3. CMIP6 Data

The CMIP6 model precipitation data used in this study are based on the multi-model output within the framework of the Sixth Phase of the Coupled Model Intercomparison Project (CMIP6). The selected models include CanESM5, MPI-ESM1-2-HR, MPI-ESM2-0, EC-Earth3-Veg-LR, and NorESM2-LM.These five climate model precipitation data have been demonstrated to have high simulation accuracy in China [28,29]. The data time span includes historical data from 1985 to 2014 and future projections from 2015 to 2100. The future scenarios are based on the Shared Socioeconomic Pathways (SSP) framework [28] and include three future development pathways: the Low Forcing Scenario (SSP126), the Medium Forcing Scenario (SSP245), and the High Forcing Scenario (SSP585). SSP585 is the scenario that brings the radiative forcing up to 8.5 W/m² in 2100, which is a scenario that will result in a drastic change in the global temperature and a significant change in the amount of precipitation. SSP126 is characterized by a radiative forcing of 2.6 W/m² by 2100, often referred to as a sustainable development scenario where global climate change is projected to be minimal. SSP245, on the other hand, reaches a radiative forcing of 4.5 W/m² by 2100, aligning with scenarios in which current socio-economic, scientific, and technological development trends are maintained, leading to moderate global climate changes. The CMIP6 model data used in this study are from the NEX-GDDP-CMIP6 (Next Generation Emissions and Climate Projection Integrated Dataset-CMIP6) dataset, which has been statistically downscaled and has a spatial resolution of 0.25° × 0.25°.

2.3. Method

In this study, multiple precipitation data are used to analyze the migration and future change characteristics of isohyet. Firstly, we analyze the multi-year average precipitation, the trend of precipitation changes, and the spatial and temporal characteristics of the 400 mm isohyet in China based on station precipitation, GPM precipitation, and grid-point precipitation data. Considering that the CMIP6 model data still exhibit a significant bias in simulating the precipitation index in western China [30], the CMIP6 model data are corrected using the quantile mapping method, and the characteristics of future changes in the isohyet are analyzed based on the correction results. The specific technical route is shown in Figure 2.

2.3.1. 400 mm Isohyet Extraction

The 400 mm isohyet was extracted from the CN05.1 and GPM datasets on an annual basis. For CMIP6 data, the extraction was performed at five-year intervals. In situ point data needed to be converted to a gridded format before isohyet extraction. This study employed the Co-Kriging (collaborative kriging interpolation) method for this conversion, leveraging its demonstrated higher accuracy in previous studies [31]. Additionally, a Digital Elevation Model (DEM) was integrated as an auxiliary variable during interpolation. This approach aimed to more effectively capture the topographic influences on precipitation patterns and thereby enhance interpolation accuracy.

To quantify the spatial position change in the extracted isohyet, it was first discretized into individual points. The latitude and longitude coordinates of all these discrete points were then obtained, allowing for the calculation of the isohyet’s average position. This was determined as follows [32]:

$$\overline{X}={\displaystyle \frac{\sum _{i=1}^n{X}_i}{n}}$$

(1)

$$\overline{Y}={\displaystyle \frac{\sum _{i=1}^n{Y}_i}{n}}$$

(2)

where $n$ is the total number of discrete points, and $\overline{X}$ and $\overline{Y}$ represent the average position of the isohyet.

2.3.2. Linear Trend Estimation

In this study, a one-dimensional linear regression model was employed to estimate trends in precipitation and the latitudinal and longitudinal shifts in the spatial position of the 400 mm isohyet. This model can be expressed as follows:

$$x_i=\mathrm{a}·t_i +\mathrm{b}$$

(3)

where $a$ is the regression coefficient, $b$ is a constant, $i$ indicates the time index, $t$ is time, and $x$ is the value of the time series. The coefficients $a$ and $b$ have been obtained using the least squares method:

$$a={\displaystyle \frac{{\sum }_{i=1}^n{x}_i{t}_i - \frac{1}{n}\left({\sum }_{i=1}^n{x}_i\right)\left({\sum }_{i=1}^n{t}_i\right)}{{{\sum }_{i=1}^n{t}_i^2 - \frac{1}{n}\left({\sum }_{i=1}^n{t}_i\right)}^2}}$$

(4)

$$b={\displaystyle \frac{1}{n}}{\sum }_{i=1}^n{x}_i-a{\displaystyle \frac{1}{n}}{\sum }_{i=1}^n{t}_i$$

(5)

where $n$ is the total number of data points. The coefficient $a$ (i.e., the slope) represents the linear trend: $a>0$ indicates an increasing trend, $a<0$ indicates a decreasing trend, and the magnitude of $a$ indicates the strength of the trend.

2.3.3. Quantile Mapping Method

For the bias correction of raw CMIP6 data, the Quantile Mapping (QM) method was employed. This method operates by first calculating the Cumulative Distribution Function (CDF) for both observed and simulated values during a defined calibration period. Subsequently, a transfer function is established to map the simulated quantiles to the observed quantiles. This derived transfer function is then applied to adjust the simulated values during the validation period, ensuring their statistical distribution aligns with that of the observations. Ultimately, this process effectively reduces model simulation errors and significantly enhances the accuracy of precipitation simulations. The effectiveness of the QM method for correcting simulated precipitation data has been well-established in previous studies [33]. The calculation method is as follows:

$$PDF\left(x\right)={\int }_a^{b}f\left(x\right)dx$$

(6)

$$CDF=\underset{0}{\overset{x_P}{\int }}PDF\left(x\right)dx$$

(7)

where $x$ represents precipitation, $PDF\left(x\right)$ is the probability density function of precipitation within the range $[a,b]$, and $CDF\left(x\right)$ is the cumulative probability distribution of precipitation less than a given threshold $x_P$. When $x_P$ equals the maximum precipitation value, the cumulative probability is 1 [34].

To establish the mapping relationship during calibration, specific quantile steps were utilized, corresponding to the 1st, 2nd, …, 99th percentiles. These inter-percentile intervals defined the calibration quantiles at which the mapping relationship was established. The uncertainty of the calibration parameters was estimated using a resampling (bootstrap) method. Confidence intervals were then calculated from 10 resamples. The training period spanned 1985 to 2004, and the validation period covered 2005 to 2014. CN05.1 was used as the reference dataset.

2.3.4. Evaluation Metrics and Uncertainty Estimation

The indicators for assessing the results of CMIP6 data calibration are the normalized standard deviation (STD), the correlation coefficient with observed data (CC), and the root-mean-square error (RMSE). The STD reflects the degree of dispersion of the data, and a larger CC indicates a stronger linear relationship between the model and the observed data, and the model predictions better reflect the actual situation. The smaller the RMSE is, the lower the model prediction error is, and the more accurate the model prediction is. The uncertainty of the 400 mm isohyet derived from CMIP6 precipitation data is quantified by the standard deviation of the results from five CMIP6 models. This standard deviation is calculated as follows:

$$std=\sqrt{{\displaystyle \frac{1}{n-1}}\sum _{i=1}^n(x_i-\overline{x}{)}^2}$$

(8)

where $\overline{x}$ is the average position of the 400 mm isohyet obtained from the five models, and $x_i$ is the average position computed by each model. Given the small sample size ($n$ = 5), the calculated spread is associated with considerable sampling uncertainty. Therefore, it is presented purely as a descriptive measure rather than a formal confidence interval; however, it can still indicate the isohyet’s uncertainty.

3. Results

3.1. Spatio-Temporal Changes in Precipitation Distribution over Mainland China

Figure 3 illustrates the annual precipitation trends derived from station, CN05.1, and GPM_IMERG datasets across mainland China from 2001 to 2017. These trends demonstrate general spatial consistency, typically indicating increasing precipitation in the southeast and decreasing trends in southwest China. Slight precipitation increases are also observed in the XJ and NW regions. Specifically, significant increases in annual precipitation are apparent in the EC and SC regions. Conversely, significant decreases, exceeding −10.0 mm/a, are observed in the southern Tibetan Plateau and the southwestern region along the Hengduan Mountains. Notably, despite the general increasing trend in eastern parts of China, a widespread decrease in precipitation is observed in the NC region, specifically along the northern bank of the Huaihe River and the Shandong Peninsula.

Distinct differences are also observed in the spatial patterns of precipitation trends among the different datasets. Generally, CN05.1 data exhibit more pronounced trends than GPM data. Specifically, CN05.1 reveals that the Kunlun and Altai Mountains are characterized by a strong precipitation gradient: the northern foothills show a modest increasing trend, whereas the southern foothills exhibit a clear decreasing trend. The GPM data, conversely, show a slight increasing trend in precipitation in this region. Given the sparse meteorological stations in this region and supported by previous studies [22], we infer that the GPM pattern is more convincing. The discrepancies in CN05.1 may primarily result from the interpolation method applied to limited station data, which could introduce significant errors in this specific area. In particular, the strong precipitation gradient near the Kunlun Mountains in Figure 3b, which appears prominently in the CN05.1 dataset, is likely influenced by interpolation uncertainties due to sparse station coverage in this high-altitude region, as this feature is less evident in satellite-based products.

Figure 4 illustrates the temporal characteristics of annual average precipitation across eight subregions from 2001 to 2017. For in situ observation station data, annual values within a specific region were obtained by averaging the station observations. Conversely, for gridded data (CN05.1 and GPM-IMERG), regional annual values were derived by averaging the grid-cell values within the same specific region. Across these subregions, the annual average precipitation generally ranges from 550 mm to 750 mm. Overall, precipitation trends mostly exhibit an increasing pattern, with rates typically ranging from 4.24 to 6.05 mm/a. The precipitation trends in most subregions are statistically significant. Specifically, significant increasing trends are observed in the NE, EC, and SC regions, with rates of 8.41–10.83 mm/a, 15.75–24.80 mm/a, and 12.55–24.04 mm/a, respectively. Meanwhile, the NC, XJ, and NW regions show slight increases at rates of 1.45–4.17 mm/a, 1.18–1.49 mm/a, and 3.35–4.25 mm/a, respectively. In contrast, the TP region exhibits a notable decreasing trend, specifically from the GPM data, at a rate of −1.95 mm/a. Given the relatively short 17-year duration of this study, and as evidenced by comparisons with longer-period trends from station data (see Table S1 in the Supplementary Materials), it is crucial to interpret these trend values as reflecting changes exclusively during the study period, rather than as indicators of long-term trends.

Despite general similarities in results across different datasets, significant discrepancies are observed in certain regions. Compared to the station data, CN05.1 data tend to overestimate precipitation, while GPM data show a clear underestimation across China. This pattern is consistent in the NE, EC, and SW regions. Conversely, the NC and SC regions exhibit very high consistency among the three datasets. Both gridded datasets (CN05.1 and GPM) consistently show underestimation in the XJ and TP regions. The notable differences between in situ data and the gridded datasets in the TP, can be largely attributed to the sparse meteorological station network, where limited point observations may not adequately represent the complex regional precipitation patterns. Therefore, a thorough assessment of a precipitation product’s regional applicability is crucial before its selection for specific studies.

3.2. Features of the 400 mm Isohyet Distribution

Figure 5 presents the migration characteristics of the 400 mm isohyet, derived from station, CN05.1, and GPM data, superimposed on the spatial distribution of multi-year average annual precipitation. Overall, the annual mean precipitation across these different datasets exhibits similar spatial patterns, characterized by a prominent precipitation gradient decreasing from southeast to northwest. This clearly underscores the highly uneven spatial distribution of annual precipitation across China. Specifically, maximum annual precipitation, reaching up to 2200 mm, is observed in the southeastern coastal areas (EC and SC regions). In stark contrast, precipitation is notably scarce in the western and northern inland areas, with the Tarim Basin in Northwest China receiving less than 50 mm annually.

The 400 mm isohyet follows a distinctive path from the western slope of the Daxinganling, traversing the Yinshan, Luliang, Bayan Kara, Tanggula, and Gundisi Mountains, and ultimately reaching the Yarlung Zangbo River Valley. The line roughly divides China into two regions, the southeast and the northwest. This vital geographical boundary effectively delineates China into two major climatic and land-use zones: the humid southeast and the arid northwest. The southeastern part is characterized by its humid climate, which fosters forest growth and supports extensive agricultural activities, representing China’s primary farming belt. In contrast, the northwestern part, with its arid climate, is dominated by grasslands and serves as the country’s principal pastoral area. Over time, this isohyet generally tends to shift towards the northwest.

To more accurately characterize the migration of the 400 mm isohyet, its average position was determined following the methodology outlined in Section 2.3.1 (Figure 6). Given the significant differences between East and West China, we divided the isohyet’s analysis into two regions: East China (east of 105°E) and West China (west of 105°E). For East China, the isohyet exhibits a consistent northwestward migration, though with varying speeds across the different datasets. Specifically, station data indicate a westward movement at 0.007°/a and a northward movement at 0.190°/a. In contrast, the CN05.1 data show the same directions of movement, but at speeds of 0.051°/a (westward) and 0.055°/a (northward), respectively. For West China, the isohyet shows a consistent northward shift with rates ranging from 0.013 to 0.036°/a. However, the longitudinal migration direction differs among the products: station and GPM data show an eastward shift, while CN05.1 data indicate a westward shift. Considering the uncertainties inherent in CN05.1’s interpolation method due to limited station data in the western region and its relatively slight westward speed, we suggest that the eastward shift observed in station and GPM data is more convincing.

3.3. Analysis of Future Precipitation Changes in China

3.3.1. Evaluation of Bias Correction for CMIP6 Model Data

Considering the strong relationship between precipitation and elevation, CMIP6 data in this study underwent bias correction stratified by elevation levels (Figure 1a). Figure 7 presents Taylor diagrams illustrating the performance metrics of the five model datasets before and after bias correction. Overall, the corrected model datasets show improved agreement with the reference data. Specifically, significant reductions were observed in STD and RMSE, while CC consistently exhibited a slight increase. The effectiveness of bias correction varied among the different model datasets. For instance, the CC for the EC-Earth3-Veg-LR model increased from 0.73 to 0.75 across three terrain steps. The RMSE of the MPI-ESM2-0 model was reduced from 0.77 to 0.72. Notably, the STD of the CanESM5 model on the first terrain step increased from 0.87 to 1.00. This indicates that the corrected model’s simulated variability became fully aligned with that of the reference data, representing a significant improvement in capturing the magnitude of fluctuations.

3.3.2. Future Precipitation Changes in China Under Different Scenarios

Figure 8 presents the projected spatial trends of annual precipitation from five CMIP6 models under SSP126, SSP245, and SSP585 scenarios. The models exhibit pronounced variations across different scenarios and regions. With increasing emission intensity, the spatial manifestation of precipitation change becomes more drastic. Under the low forcing scenario (SSP126), future precipitation generally decreases across the vast majority of China. Conversely, under the high forcing scenario (SSP585), the CanESM5 and EC-Earth3-Veg-LR models project widespread precipitation growth across almost the entire China region (excluding the Northwest), with a trend exceeding +4 mm/a. In contrast, the MPI-ESM1-2-HR and MPI-ESM2-0 models indicate precipitation growth concentrated in eastern and southern China, and the southern Tibetan Plateau. Under the medium emission scenario (SSP245), a notable divergence among models begins to appear. Most models (e.g., CanESM5, EC-Earth3-Veg-LR) exhibit a clear trend of precipitation increase in eastern China and coastal southern China, with increases generally ranging from 1 to 3 mm/a. However, some models (e.g., MPI-ESM1-2-HR) project a strong drying trend in the southwest mountainous region, suggesting a considerable degree of uncertainty in this specific area under future climate change.

The multi-model ensemble mean provides a more robust representation of the spatial precipitation patterns (Figure 8p–r). Under the low forcing scenario (SSP126), most regions in China show a precipitation decrease, with slight increases observed only in the SC and EC regions. This general pattern of decreasing precipitation intensifies as the emission scenario progresses. Under SSP245, only the EC region continues to present a slight precipitation increase. However, under the high emission scenario (SSP585), western parts of the TP region exhibit a significant precipitation increase, which contrasts sharply with the trends observed under the other two scenarios. Given the complex climate and geography of this region, the exact patterns and precise characteristics will require more comprehensive data and further investigation in future studies.

3.3.3. Projected Shifts in the 400 mm Isohyet

Figure 9 illustrates the future shifts in the 400 mm isohyet, superimposed on the spatial distribution of multi-year average annual precipitation. Under the three scenarios, the projected average annual precipitation across China consistently exhibits a clear spatial distribution, characterized by prominent south-to-north and east-to-west gradients. Specifically, eastern and southern China are projected to receive substantial precipitation, with annual totals generally exceeding 1000 mm, and even reaching 2000 mm in some localized areas. Conversely, precipitation is notably scarce in NC, particularly in XJ, and extends to the northern part of the TP. Here, annual precipitation is generally below 400 mm, indicative of typical arid and semi-arid climate characteristics.

Analyzing different scenarios, a consistent trend emerges: the projected average precipitation shows a drying trend in the north and northeast, and a wetting trend in the south and southeast. Concurrently, the 400 mm isohyet is projected to shift predominantly westward across all scenarios, with an increasingly pronounced northwestward tendency along the northern boundary of the TP as emission intensity rises. Beyond this overall shift, the inter-annual variability of the 400 mm isohyet also shows distinct patterns. Under the SSP126 and SSP245 scenarios, the spatial location of the isohyet demonstrates greater inter-annual consistency. In contrast, under the high-emission SSP585 scenario, the isohyet exhibits significantly higher instability, particularly evident along the northern boundary of the TP. Despite this increased variability, the overarching tendency is a westward displacement, with northwestward components along the TP’s northern margin.

Figure 10 quantitative evaluation of the projected changes in the 400 mm isohyet over East and West China. Under the low forcing scenario (SSP126), the isohyet exhibits a slight northwestward migration in East China, with migration rates of 0.001°/10a for longitude and 0.088°/10a for latitude, respectively. West China shows a similar northwestward shift but at a faster speed under SSP126. With increasing radiative forcing (SSP245), the northwestward shift becomes more significant. Specifically, East China experiences migration speeds of 0.014°/10a northward and 0.103°/10a westward, while West China shows rates of 0.105°/10a northward and 0.245°/10a westward. However, this situation becomes more complex under the high forcing scenario (SSP585), with a southwestward shift observed in East China and a rapid northwestward shift in West China.

4. Discussion

From 2001 to 2017, the spatial pattern of precipitation in China exhibited the well-known “more in the southeast and less in the northwest” distribution. During this period, annual precipitation increased in the southeast coastal area, while decreasing in the southern Tibetan Plateau and the southwestern Hengduan Mountains. Looking ahead to future changes, projections indicate a significant precipitation decrease across most parts of China by the end of the 21st century, even under relatively positive mitigation scenarios (such as SSP126 and SSP245) [35]. Conversely, the southeastern coastal areas are anticipated to continue experiencing increasing precipitation trends in the future. However, it is important to note that due to the use of only five CMIP6 models in this study and the significant differences observed among them, the future precipitation change projections contain substantial uncertainties. Future research should incorporate more models to obtain more accurate results.

Analysis and calculation of the 400 mm isohyet in China reveal that this climatic dividing line has consistently shifted northwestward over the last two decades, with its component migration rates showing distinct differences. In East China, the isohyet exhibits a northwestward shift at varying rates across different datasets. Conversely, in West China, the isohyet shows a consistent northward and eastward shift. Future projections indicate that the 400 mm isohyet will continue its westward and northward migration under various emission pathways. Notably, the migration rate is projected to be significantly faster under the high radiative forcing scenario (SSP585) compared to low and medium scenarios. These findings strongly suggest an ongoing and accelerated structural transformation of China’s precipitation patterns. Consequently, this 400 mm isohyet may no longer serve as the definitive dry–wet boundary in the future, as stated in previous studies [10,11,12]. Furthermore, if no effective emission reduction measures are taken, the rate of change for China’s climate structural transformation is projected to accelerate even faster than current trends, which may have profound impacts on the distribution of agriculture, livestock farming, ecosystems, and water resources.

This study primarily focused on precipitation changes and the migration of the 400 mm isohyet. However, it does not yet integrate other influencing climatic factors such as temperature, wind speed, and evaporation. To address this limitation, future work will develop a multi-factor climate-aridity index that blends precipitation, temperature, PET, wind speed, and soil-moisture metrics, providing a more realistic depiction of semi-humid and semi-arid dynamics. Furthermore, integrating advanced algorithms, such as machine learning and deep learning, could significantly enhance the monitoring of the isohyet’s dynamic morphological changes and abnormal fluctuations. This approach would further improve spatial and temporal accuracy, as well as the overall analytical intelligence of the assessment. Future studies will focus on developing a more integrated model to evaluate changes in China’s dry–wet patterns and assess their impacts on hydrological and ecological systems. Machine-learning and deep-learning approaches will be explored to capture non-linear interactions and track the isohyet’s morphology with finer spatial–temporal resolution.

5. Conclusions

This study, leveraging multi-source precipitation data and CMIP6 model outputs, systematically analyzed the spatiotemporal variation characteristics of China’s 400 mm isohyet from 2001 to 2017 and projected its future changes until the end of the 21st century (2100). The key conclusions are as follows:

From 2001 to 2017, precipitation trends in China mostly exhibited an increasing pattern, with rates ranging from 4.24 to 6.05 mm/a. Specifically, increasing precipitation was observed in the southeast, while decreasing trends occurred in Southwest China. During this period, the 400 mm isohyet exhibited a spatially complex shift. The East China section moved northwestward, showing average rates of 0.075°/a northward and 0.055°/a westward. In contrast, the West China section shifted northeastward, with average rates of 0.028°/a northward and 0.032°/a eastward. By 2100, under low (SSP126), medium (SSP245), and high (SSP585) radiative forcing scenarios, the 400 mm isohyet is projected to maintain its westward and northward migration, with significantly faster rates under intensifying radiative forcing scenarios.

These findings offer robust theoretical support for regional land resource planning, water resource management, and ecological environment protection. They are crucial for promoting the rational utilization and conservation of water resources, fostering sustainable agricultural development, and ensuring ecological balance. This research holds significant practical implications for enhancing China’s ecological security and fostering stable socioeconomic development.