Multi-Source Indicator Modeling and Spatiotemporal Evolution of Spring Sowing Agricultural Risk Along the Great Wall Belt, China

Abstract

The spatiotemporal heterogeneity of hydrothermal conditions during the spring sowing period profoundly shapes cropping layouts and sowing strategies. Using NASA’s GLDAS remote sensing reanalysis, we developed a continuous agricultural climate risk index that integrates three remotely driven indicators—spring sowing window days (SWDs) derived from a “continuous suitable-day” logic, the hydrothermal coordination degree (D value), and a comprehensive suitability index (SSH_SI)—thus advancing risk assessment from single metrics to a multidimensional framework. Methodologically, dominant periodic structures of spring sowing hydrothermal risk were extracted via a combination of wavelet power spectra and the global wavelet spectrum (GWS), while spatial trend-surface fitting and three-dimensional directional analysis captured spatial non-stationarity. The index’s spatial migration trajectories and centroid-evolution paths were then quantified. Results reveal pronounced gradients along the Great Wall Belt: SWD displays a “central-high, terminal-low” pattern, with sowing windows restricted to only 3–6 days in northeastern Inner Mongolia and western Liaoning but extending to 11–13 days in the central plains of Inner Mongolia and Shanxi; SSH_SI and D values form an overall “south-west high, north-east low” pattern, indicating more favorable hydrothermal coordination in southwestern areas. Temporally, although SWD and SSH_SI show no significant downward trend, their interannual variability has increased, signaling rising instability, whereas the D value declines markedly in most regions, reflecting intensified hydrothermal imbalance. The integrated risk index identifies high-risk hotspots in eastern Inner Mongolia and northern North China, and low-risk zones in western provinces such as Gansu and Ningxia. Centroid-shift analysis further uncovers a dynamic regional adjustment in optimal sowing patterns, offering scientific evidence for addressing spring sowing climate risks. These findings provide a theoretical foundation and decision support for optimizing regional cropping structures, issuing climate risk warnings, and precisely regulating spring sowing schedules.

1. Introduction

Against the dual backdrop of global warming and high-quality agricultural development, accurately identifying crop sowing windows and optimizing sowing schedules has become essential for enhancing climate resilience and ensuring food se-curity [1,2,3]. As the onset of the growing season, spring sowing is highly influenced by hydrothermal conditions, which directly affect seed germination, emergence, and early growth, exerting a significant “window effect” on final yields [4,5]. Determining a suitable spring sowing window can help avoid frost and spring drought, improve resource-use efficiency, and support disaster mitigation as well as stable, high yields [6].

Globally, research increasingly relies on crop models (e.g., DSSAT, AquaCrop, APSIM) to simulate optimal sowing dates under climate change [7,8]. Recently, re-mote sensing and reanalysis datasets have been integrated to improve the spatiotemporal resolution of sowing assessments [9]. In China, however, studies still primarily depend on meteorological thresholds—such as average air or soil temperature and soil moisture—to define sowing criteria [10,11,12,13]. Composite indicators, such as the hydrothermal coordination degree (D value), are also emerging [14]. Yet, most studies remain limited to single regions or variables and rarely account for climate variability or associated risks [15,16]. Overall, current methods tend to be model-dependent, poorly transferable, and insufficiently sensitive to climate fluctuations, highlighting the need for a multi-source, multi-ecozone framework for sowing window detection and risk evaluation [17,18,19].

Spanning semi-arid, semi-humid, and agro-pastoral ecotones across northern China, the Great Wall Belt serves as both a dryland agricultural base and an ecological buffer zone. Spring maize, millet, and potato are the dominant crops, and their sowing and emergence are highly sensitive to springtime hydrothermal conditions [20]. In recent years, the region has experienced warmer and drier springs, with earlier soil thaw and more frequent drought events, creating new challenges for sowing suitability [21,22]. Current studies in the region still rely on crop models or empirical thresholds [23,24,25,26,27,28,29], which are difficult to generalize across its complex landscapes and diverse cropping systems. Therefore, there is an urgent need for a high-resolution, systematic, and transferable framework to assess spring sowing suitability.

Existing studies face several key limitations: (1) Most current methods rely on single-factor thresholds or crop model simulations, making it difficult to comprehensively characterize the synergistic effects of multiple variables such as temperature, moisture, radiation, and wind speed. (2) There is a lack of effective identification of “consecutive suitable days” to define the spring sowing window, which limits the ability to quantify the window length (SWD) and its temporal stability. (3) Quantitative assessment of hydrothermal coupling coordination (D value) between meteorological and soil conditions remains underdeveloped. (4) A comprehensive risk index (CRI) to represent the transition from decreasing suitability to increasing uncertainty is still lacking. (5) Key information on data preprocessing, weight determination, and reproducibility of the workflow is often missing, limiting the scalability and transferability of these methods across regions.

To address these issues, this study proposes a spring sowing agricultural suitability evaluation framework based on GLDAS multi-source data, fuzzy comprehensive evaluation, and objective weighting. The framework is designed for cross-crop and cross-region applicability. The main innovations of this study are as follows: (i) The development of a Spring Sowing Hydrothermal Suitability Index (SSH_SI) that integrates seven key meteorological and soil variables; (ii) introduction of a Hydrothermal Coupling Coordination Index (D value) to characterize the alignment between thermal and moisture conditions; (iii) proposal of a novel method to identify spring sowing windows based on the logic of “consecutive suitable days” (SWD); (iv) construction of a Comprehensive Risk Index (CRI) to represent declining suitability and increasing uncertainty; (v) application of multiple spatiotemporal analysis methods—including Sen’s slope, Mann–Kendall test, wavelet analysis, trend surface analysis, and centroid migration—to reveal the temporal-spatial evolution of spring sowing suitability along the Great Wall region from 2000 to 2024; and (vi) clear documentation of critical technical steps such as data filtering thresholds, spatial masking, temporal aggregation, interpolation, and weight calculation, with supplementary scripts and flowcharts provided to enhance reproducibility.

This study not only offers a quantitative tool for spring sowing management and climate risk prevention along the Great Wall Belt, but also provides a replicable technical path for sowing window identification, coupling diagnosis, and risk zoning in ecologically complex regions.

The technical framework for agricultural suitability and risk assessment is illustrated in Figure 1.

The findings offer theoretical support and methodological guidance for adaptive cropping layouts, climate risk management and precise sowing date regulation in northern China under climate change.

2. Materials and Methods

2.1. Overview of the Study Area

The study area lies along the Great Wall in northern China, spanning the provinces (or autonomous regions and municipalities) of Gansu, Ningxia, Shaanxi, Shanxi, Inner Mongolia, Hebei, Beijing between roughly 100°–125° E and 35°–42° N in a northwest to southeast orientation (Figure 2). This belt forms a classic transition zone between arid and semi-arid regions and is dominated by a continental monsoon climate: spring temperatures rise rapidly, yet precipitation is scarce and spring droughts are frequent, yielding strong interannual climate variability. Mean annual temperatures range from 5 °C to 10 °C, while annual precipitation averages about 200–500 mm—most of which falls in summer—leaving the spring sowing period (mid-April to mid-May) particularly sensitive to hydrothermal deficits [30].

Dryland farming prevails, and the region is a key production zone for spring-sown wheat, millet, maize and potato, as well as specialty crops such as oat, flax and cool-season vegetables. Given limited water resources, hydrothermal suitability during sowing is critical for crop emergence and yield [30]. To evaluate spring hydrothermal coupling and agricultural risk, we established more than 2000 sampling points (blue dots in Figure 1) within the study area and derived three core indicators—spring sowing window days (SWD), hydrothermal coordination degree (D value) and comprehensive suitability index (SSH_SI). The red boundary in Figure 1 delineates the study domain, providing the spatial framework for constructing the sowing-period climate risk index and analyzing its spatiotemporal evolution.

2.2. Data Source

The meteorological and soil variables used in this study are derived primarily from NASA’s Global Land Data Assimilation System (GLDAS) product suite. GLDAS couples the NOAH land-surface model with remotely sensed retrievals and ground observations, providing high spatial–temporal resolution and internally consistent data that are widely employed in hydrological and agro-meteorological research [31,32,33,34,35,36,37].

The meteorological and soil variables used in this study are primarily sourced from NASA’s Global Land Data Assimilation System (GLDAS) product suite. GLDAS integrates the NOAH land surface model with remote sensing retrievals and ground-based observations to produce data products with high spatial–temporal resolution and physical consistency, widely used in hydrological and agrometeorological research [31,32,33,34,35,36,37]. In this study, the GLDAS_NOAH dataset with a spatial resolution of 0.25° × 0.25° and a daily temporal resolution was used to extract data for the period from 20 April to 20 May each year between 2000 and 2024, and seven core variable groups were obtained:

Thermal factors include surface soil temperature (SoilTMP0_10cm_inst and SoilTMP10_40cm_inst, in K) and 2 m air temperature (Tair_f_inst, in K), reflecting near-surface thermal conditions during the spring sowing period.

Moisture factors comprise precipitation rate (Rainf_f_tavg, in kg·m⁻²·s⁻¹), actual and potential evapotranspiration (Evap_tavg and PotEvap_tavg, both in kg·m⁻²·s⁻¹), as well as surface and sub-surface soil moisture (SoilMoi0_10cm_inst and SoilMoi10_40cm_inst, in kg·m⁻²) and root-zone soil moisture (RootMoist_inst, in kg·m⁻²), which together characterize water availability and soil water retention during early crop development.

Radiation and wind factors include downward shortwave radiation at the surface (SWdown_f_tavg, in W·m⁻²) and near-surface wind speed (Wind_f_inst, in m·s⁻¹), providing insight into energy inputs and wind disturbances during the sowing period.

To ensure the scientific rigor and reproducibility of data preprocessing, the workflow included the following key steps:

(1)

Outlier detection and removal were performed by applying physical-based thresholds to each key variable to ensure data reliability. Specifically, 2 m air temperatures were excluded if they exceeded 50 °C or fell below −50 °C (i.e., outside the range of 223–323 K). Soil temperatures at 0–10 cm and 10–40 cm were removed if greater than 333 K or lower than 223 K. Negative values were eliminated from variables such as wind speed, precipitation rate, evapotranspiration, and soil moisture. Shortwave radiation values above 1400 W·m⁻² were identified as extreme and excluded. Soil moisture values—both surface (0–10 cm) and sub-surface (10–40 cm)—as well as root-zone moisture, were considered invalid if they implied a volumetric water content greater than 0.6 m³/m³. These outlier filters ensured the physical consistency and quality of the input data before subsequent analysis.

(2)

To ensure consistency across variables, all GLDAS data were standardized as follows: Precipitation flux and actual evapotranspiration (kg·m⁻²·s⁻¹) were converted to mm/day by multiplying by 86,400. Potential evapotranspiration (W·m⁻²) was transformed to mm/day using a factor of 0.0352. Soil moisture and root-zone moisture (kg·m⁻²) were treated as equivalent to mm. Soil and air temperatures (K) were converted to °C using °C = K − 273.15. Downward shortwave radiation (W·m⁻²) was scaled to MJ·m⁻²·day⁻¹ by multiplying by 0.0864. Wind speed (m·s⁻¹) was retained without conversion. This process ensured unit compatibility for subsequent analysis and visualization.

(3)

Administrative boundary masking: The administrative boundary shapefile of the study area was imported into ArcGIS(v10.7). The “Clip” tool was applied to spatially constrain the interpolated GLDAS data, retaining only valid grid cells and sampling points within the study region.

(4)

Data completeness filtering: For each grid cell, the completeness of data during the spring sowing period over the 25-year span was evaluated. Cells with less than 90% annual completeness were excluded to ensure temporal continuity and consistency in time-series analysis.

(5)

Variable derivation and point sampling: Daily sequences were retained for all valid latitude–longitude grid points. Based on these, key hydro-meteorological indicators such as evapotranspiration deficit and hydrothermal coordination index (D value) were calculated. Sampling points were spatially matched with the study area shapefile to extract valid sites located within the region.

(6)

Spatial interpolation analysis: Using ArcGIS(v10.7) spatial analysis tools, the inverse distance weighting (IDW) method was applied to generate continuous spatial distribution maps for each variable. These outputs support the subsequent spatial heterogeneity analysis and data visualization.

2.3. Construction of the Indicator System

2.3.1. Logic for Defining Single-Factor Suitability Indices

(1)

Precipitation Suitability Index (PSI).

The Precipitation Suitability Index (PSI) is evaluated within a 7–10-day window centered on the sowing date. Cumulative rainfall < 5 mm denotes severe drought and a dry-sowing failure risk (score 0.0); 5–10 mm is slightly dry (0.4); 10–20 mm represents optimal soil moisture (1.0); 20–30 mm is slightly wet (0.8); and >30 mm poses risks of seed rot or surface crusting (0.3), reflecting the eco-hydrological principle that moderation is best. Temporal synchrony is assessed using daily rainfall ± 3 days around sowing (effective rain ≥ 1 mm∙day⁻¹): rain 1–3 days before sowing provides ideal soil wetting (1.0); rain 1–3 days after sowing still aids emergence (0.8); rain occurring ≥ 5 days before or after sowing risks moisture loss or delayed germination (0.4); and no effective rainfall in either window indicates unsuitable dry sowing (0.0).

The final PSI combines these components as: PSI = 0.6 × (total amount score) + 0.4 × (temporal sequence score), balancing water supply with its timing relative to sowing.

(2)

Air Temperature Suitability Index (TSI).

Based on the daily mean air temperature (Tair_f_inst), thermal conditions are deemed suitable only when the sowing day itself—or three consecutive days—exceed a specified threshold. If the temperature is ≥10 °C, conditions are ideal (score 1.0); 8–10 °C is marginally suitable (score 0.7); 5–8 °C is borderline acceptable (score 0.4); and below 5 °C the site is considered unsuitable for sowing (score 0.0).

(3)

Wind Speed Suitability Index (WSI).

Using the maximum wind speed measured 3–7 days after sowing: a speed ≤ 3 m s⁻¹ is considered optimal (score 1.0); 3–5 m∙s⁻¹ is acceptable (score 0.8); 5–7 m∙s⁻¹ entails wind-erosion risk and is only marginally suitable (score 0.5); and any speed > 7 m∙s⁻¹ is deemed unsuitable (score 0.0).

(4)

Soil Temperature Suitability Index (SSI)

Based on 0–20 cm soil temperature, the soil temperature suitability index (SSI) is assigned as follows: three consecutive days ≥ 10 °C indicate optimal conditions (score 1.0); if any two days reach ≥ 10 °C and the three-day mean is ≥9.5 °C, conditions are acceptable (0.8); if only one day is ≥9 °C or temperatures fluctuate widely, suitability is poor (0.4); when all three days remain <9 °C, sowing is unsuitable (0.0).

(5)

Evapotranspiration Balance Index (ETBI)

The evapotranspiration balance index (ETBI) uses the ratio of actual to potential evapotranspiration.

$$ETBI={\displaystyle \frac{Evap}{PotEvap}}$$

(1)

A ratio of 0.75–1.00 denotes a balanced water supply (score 1.0); 0.50–0.75 is acceptable (0.8); 0.30–0.50 is poor (0.5); <0.30 is unsuitable (0.3); and ratios > 1.00 are treated as possible data errors (score 0.0).

(6)

Radiation Suitability Index (RSI)

For the radiation suitability index (RSI), the three-day mean downward shortwave radiation (two days before and the sowing day) is classified by long-term percentiles: >75% is optimal (1.0); 50%–75% is acceptable (0.7); 25%–50% is poor (0.4); and <25% is unsuitable (0.1).

(7)

Soil Moisture Suitability Index (SMSI)

The soil moisture suitability index (SMSI) is based on 0–20 cm volumetric soil water content. Values < 12% (or <15.6 kg∙m⁻²) are unsuitable (0.0); exactly 12% marks the sowing threshold (0.4); 12–25% (15.6–32.5 kg∙m⁻²) defines the optimal sowing window (1.0); and >25% indicates overly wet soil, lowering the score to 0.3 and hindering field operations.

2.3.2. The Construction of a Composite Index

(1)

Spring Sowing Hydrothermal Suitability Index (SSH_SI)

On the basis of the seven single-factor suitability indices, we constructed the “Spring Sowing Hydrothermal Suitability Index” (SSH-SI). This index employs fuzzy comprehensive evaluation theory [38,39] and integrates the seven single-factor suitability indices using a weighted average operator. The fundamental model is expressed as follows:

$${SSH}_{SI}=\sum _{i=1}^n{w}_i·{\mu }_i$$

(2)

where ${\mu }_i$ represents the suitability score of the i-th factor, and $w_i$ denotes its corresponding weight, satisfying

$$\sum _{i=1}^n{w}_i=1$$

(3)

To achieve objectivity and data-driven characteristics in weight allocation, this study employs the entropy weight method [40,41] to determine weights for the seven indicators. A trapezoidal membership function is applied to calculate the membership degree of each grade for every indicator.

The entropy weight method is an objective weighting approach based on the principle of information entropy, which determines the weight of each indicator by measuring the variability of the indicator across the sample. The calculation method is as follows:

$$w_j={\displaystyle \frac{1-e_j}{{\sum }_{j=1}^m\left(1-e_j\right)}},$$

(4)

$$e_j=-{\displaystyle \frac{1}{\ln \left(n\right)}}{\sum }_{i=1}^n{\rho }_{ij}\ln \left({\rho }_{ij}\right),$$

(5)

$${\rho }_{ij}={\displaystyle \frac{x_{ij}}{{\sum }_{i=1}^n{x}_{ij}}},$$

(6)

where $w_j$ is the objective weight of the $j$-th indicator; $e_j$ is the information entropy of the $j$-th indicator; and ${\rho }_{ij}$ is the proportion of the normalized value of the i-th sample under the $j$-th indicator.

The weight coefficients of the SSH_SI index corresponding to wind speed, radiation, precipitation, air temperature, soil moisture, soil temperature, and evapotranspiration balance are as follows: W = [0.0612, 0.1473, 0.2308, 0.1898, 0.1336, 0.1097, 0.1275].

The trapezoidal membership function defines a trapezoidal region and assigns different membership degrees across different intervals of input values, thereby describing the degree to which a variable belongs to a given fuzzy set. The shape of the trapezoidal membership function is typically determined by four parameters ($a$, $b$, $c$, $d$), which correspond to key positions on the horizontal axis (i.e., the range of indicator values). Its mathematical expression is as follows:

$$\mu \left(x\right)=\left\{\begin{array}{c}0, x\le a\\ {\displaystyle \frac{x-a}{b-a}}, a<x\le b\\ 1, b<x\le c\\ {\displaystyle \frac{d-x}{d-c}}, c<x\le d\\ 0, x>d\end{array}\right.$$

(7)

where $\mu \left(x\right)$ represents the membership degree of the input value (x) for a specific fuzzy grade. Parameters $a$ and $d$ denote the start and end points of the trapezoid’s base, while $b$ and $c$ define the interval where the membership degree equals 1. This function can be flexibly adjusted to accommodate different evaluation grade classifications.

(2)

Meteorological Sub-index (MI)

To quantitatively describe the meteorological suitability during the spring sowing period, the Meteorological Sub-index (MI) was introduced. MI primarily comprises the following three elements—temperature, wind speed, and radiation indicators—with its fundamental model expressed as below:

$$MI=w_1·{\mu }_T+w_2·{\mu }_w+w_3·{\mu }_R$$

(8)

where ${\mu }_T$ represents the temperature suitability evaluation index value; ${\mu }_w$ represents the wind speed suitability evaluation index value; ${\mu }_R$ represents radiation suitability evaluation index value; and $w_1{, w}_2, w_3$ are the corresponding weighting coefficients.

(3)

Hydrological–Thermal Sub-index (WHI)

To quantitatively describe the coordination and suitability of subsurface thermal and moisture conditions during the spring sowing period, the hydrological–thermal sub-index (WHI) is introduced. WHI primarily comprises the following four elements—soil temperature, soil moisture, precipitation conditions, and evapotranspiration equilibrium—with its fundamental model expressed as below:

$$WHI=w_4·{\mu }_{ST}+w_5·{\mu }_{SM}+w_6·{\mu }_P+w_7·{\mu }_{ETB}$$

(9)

where ${\mu }_{ST}$ represents the soil temperature suitability evaluation index value (°C); ${\mu }_{SM}$ represents soil moisture suitability evaluation index value (% vol); ${\mu }_P$ represents the precipitation suitability indicator value (mm); ${\mu }_{ETB}$ represents the evapotranspiration suitability evaluation indicator value (dimensionless); and $w_4{,w}_5,w_6{,w}_7$ are the corresponding weighting coefficients.

(4)

Meteorological–Hydrothermal Coupling Coordination Degree (D)

To quantitatively describe the synergistic matching and high spatiotemporal consistency between hydrothermal and meteorological conditions, the meteorological–hydrothermal coupling coordination degree model (D) is introduced [42,43], with its fundamental model expressed as follows:

$$D_i=\sqrt{C_i·T_i}$$

(10)

$$T_i=\alpha {WHI}_i+\beta {MI}_i$$

(11)

$$C_i={\displaystyle \frac{2·\sqrt{{WHI}_i·{MI}_i}}{{WHI}_i+{MI}_i}}$$

(12)

where $T_i$ represents comprehensive benefit index ($\alpha =\beta =0.5$), $D_i$ represents coupling coordination degree on day i (synergy intensity), ${WHI}_i$ represents the hydrological–thermal sub-index on day i-th (soil moisture temperature suitability), and ${MI}_i$ represents the meteorological sub-index on day i-th (atmospheric condition suitability), C ∈ [0, 1]. Higher values indicate stronger coupling. High C and low D indicates a strong interaction but low developmental level between the systems, while high C and high D signifies high suitability, robust coupling, coordinated meteorological–hydrothermal conditions, and optimal sowing suitability.

(5)

Spring Sowing Window Days (SWD)

The spring sowing window is defined as a consecutive period within a given year at a specific site during which the meteorological–hydrothermal coupling coordination degree (D) exceeds a defined threshold, indicating that climatic conditions meet optimal agricultural sowing. To account for regional and interannual fluctuations in D value distributions, a non-parametric threshold-setting approach is adopted for enhanced robustness. The methodology proceeds as follows for each sampling point annually:

Threshold Application: Filter dates satisfying the D ≥ global D value threshold;

Window Identification: Identify the longest consecutive date segment (“window period”);

Metrics Extraction: Record the start/end dates and duration (days) of this window;

Annual Evaluation: Use the window length as the annual metric for spring sowing suitability.

(6)

Risk Index Identification

To comprehensively evaluate agricultural suitability risks during the spring sowing period, this study constructs a continuous comprehensive risk index (CRI) quantifying region-specific planting risks under meteorological–hydrothermal matching conditions. The index integrates three key factors: spring sowing window days (SWD, $x_1$), composite suitability index (SSH_SI, $x_2$), meteorological–hydrothermal coordination degree (D value, $x_3$). First, each factor undergoes min–max normalization to eliminate dimensional differences, followed by directional consistency adjustment (i.e., converting the spring sowing window days from a positive- to a negative-oriented indicator) [44,45].

$$x_i^{\prime }={\displaystyle \frac{x_{max}-x_i}{x_{max}-x_{min}}}$$

(13)

Next, the comprehensive risk index (CRI) is calculated as follows:

$$CRI=w_1·x_1^{\prime }+w_2·x_2^{\prime }+w_3·x_3^{\prime }$$

(14)

where weights $w_1$, $w_2$, $w_3$ are set as equal ($w_1$ = $w_2$ = $w_3$ = 1/3).

A higher risk index indicates less suitable regional agricultural hydrothermal conditions, poorer spring sowing flexibility, and weaker factor coordination, thus implying greater risk. This index can not only quantitatively identify high-risk areas but also be applied for spatial clustering analysis and screening of high-potential zones, providing decision-making support for precision agriculture management and crop structure adjustment.

2.4. Research Methods

2.4.1. Trend Analysis

Employing the nonparametric Sen’s slope estimator (Theil-Sen Median) [46] to compute the median of slopes between all point pairs.

$$\beta =\mathrm{m}\mathrm{e}\mathrm{d}\mathrm{i}\mathrm{a}\mathrm{n}\left({\displaystyle \frac{x_j-x_i}{j-i}}\right),\forall j>i$$

(15)

where $x_j$ is the observation value on day j-th, $x_i$ is the observation value on day i-th; β > 0 indicates a rising trend, while β < 0 indicates a falling trend.

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

(16)

where

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

(17)

When the sample size n ≥ 10, S can be approximated by a normal distribution. The standard normal variate Z is then constructed as follows:

If $\left|Z\right|>Z_{0.05/2}$, then a significant trend exists.

2.4.2. Wavelet Analysis

Morlet continuous wavelet analysis (CWT) is a time–frequency analysis method that decomposes signals across different frequencies and temporal scales to reveal local features. The procedural steps are as follows [47,48]:

Morlet continuous wavelet analysis (CWT) comprises signal preprocessing, the selection of appropriate wavelet parameters, computation of the continuous wavelet transform, generation of time–frequency plots, and feature extraction. The Morlet wavelet is a complex-valued mother wavelet formed by combining a sinusoidal wave with a Gaussian function, providing balanced time and frequency resolution in the time–frequency plane. Its mathematical expression is defined as

$$\psi \left(t\right)={\pi }^{-{\displaystyle \frac{1}{4}}}{e}^{i{\omega }_{0}t}{e}^{-{\displaystyle \frac{t^2}{2}}}$$

(18)

where $\psi \left(t\right)$ denotes the wavelet function; ${\omega }_0$ represents the central frequency parameter, typically set to 5 or 6 to balance temporal and spectral resolution; $e^{i{\omega }_{0}t}$ constitutes the complex sinusoidal component responsible for capturing frequency information of the signa; and $e^{-{\displaystyle \frac{t^2}{2}}}$ is the Gaussian window ensures superior time–frequency localization properties.

The continuous wavelet transform (CWT) of the Morlet wavelet is mathematically defined as:

$$W\left(a,b\right)={\int }_{-\infty }^{+\infty }x\left(t\right){\psi }_{a,b}\left(t\right)dt$$

(19)

where $W\left(a,b\right)$ denotes the wavelet coefficients, representing signal characteristics at scale a and time location b; $x\left(t\right)$ is the original signal; ${\psi }_{a,b}\left(t\right)$ represents the Morlet wavelet function adjusted by scale $a$ and translation $b$, expression is ${\psi }_{a,b}\left(t\right)={\displaystyle \frac{1}{\sqrt{a}}}\psi \left({\displaystyle \frac{t-b}{a}}\right)$, By varying $a$ (scale) and $b$ (translation), the Morlet wavelet enables the analysis of signal variations across different scales (i.e., frequencies) and temporal positions.

Subsequently, the power spectrum is computed [49]:

$$Power={\left|W\left(a,b\right)\right|}^2$$

(20)

2.4.3. Center of Gravity Migration

Shifts in the drought frequency centroid reveal the spatial variation trends and patterns of drought events, with the centroid calculation method given by [50]:

$$X_i={\displaystyle \frac{{\sum }_{i=1}^n{w}_i{x}_i}{{\sum }_{i=1}^n{w}_i}}{ Y}_i={\displaystyle \frac{{\sum }_{i=1}^n{w}_i{y}_i}{{\sum }_{i=1}^n{w}_i}}$$

(21)

In the formula, $X_i$ and $Y_i$ represent the longitude and latitude coordinates of the drought centroid, respectively; n is the number of stations; $w_i$ denotes the drought frequency at the i-th station; and $X_i$, $Y_i$ indicates the location of the i-th point. For each time point, the centroid positions (X₁,Y₁), (X₂,Y₂), …, (X_n,Y_n}) are calculated using the same method. If the data spans multiple time periods, the centroid position shifts over time, reflecting the dynamic migration of the system.

3. Results

3.1. Temporal Variation Characteristics of Indicators

To reveal long-term trends in key meteorological and suitability indicators during the spring sowing period, this study conducted linear regression analysis on the annual average D value (meteorological hydrothermal coordination degree), SSH_SI (comprehensive suitability index), and spring sowing window days (SWD) from 2000 to 2024. The results are illustrated in Figure 3.

As shown in Figure 3, the D value exhibits a statistically significant downward trend, indicating a continuous decline in hydrothermal coordination over the study period. This suggests a progressive reduction in the suitability of meteorological hydrothermal conditions, potentially linked to enhanced climate variability or increased extreme weather events, thereby threatening agricultural sowing safety. The SSH_SI index also displays a gradual declining trend, albeit with smaller magnitude, yet remains statistically significant (p < 0.05), implying a year-by-year weakening of overall agricultural suitability. This subtle decline may reflect the combined effects of reduced regional ecological stability or intensified climate fluctuations during the sowing period. In contrast, SWD shows greater interannual variability without a clear linear trend, suggesting that despite annual fluctuations in sowing window length, no definitive long-term shortening pattern exists. It can be inferred that SWD is strongly driven by climatic conditions but has not yet manifested a persistent reduction trend, possibly due to regional or interannual regulatory mechanisms. Collectively, the significant downward trends in D value and SSH_SI reveal deteriorating risks in meteorological suitability for spring sowing, while the non-significant change in SWD may indicate stronger influences from other nonlinear factors. The coordinated variations in these three indicators provide a foundation for subsequent coupled-relationship analysis and comprehensive risk assessment.

Further daily-scale analysis of the meteorological hydrothermal coordination degree (D value) and comprehensive suitability index (SSH_SI) trends during 2000–2024 is presented in Figure 4. The D value shows a statistically significant declining trend over the study period, indicating a deterioration in the coordination between meteorological and moisture conditions during the spring sowing period. Although SSH_SI exhibits a slight downward tendency, this trend lacks statistical significance. Overall, the D value displays high-amplitude fluctuations with strong interannual instability, while SSH_SI maintains high stability and consistency, consistently remaining near 0.42. This divergence suggests that the dynamic coupling relationships considered in the D value calculation are more susceptible to disturbances from extreme weather and moisture imbalances, whereas SSH_SI—as a weighted composite of multiple factors—demonstrates greater smoothing capacity against fluctuations in individual variables. Consequently, combining both indicators provides a more holistic representation of hydrothermal environmental changes during spring sowing: the D value emphasizes hydrothermal coordination dynamics, while SSH_SI reflects integrated suitability. These findings establish the basis for the in-depth assessment of agroclimatic risks and optimal sowing window shifts in the region.

To reveal the periodic characteristics of key indicators during the spring sowing period, this study employs the continuous wavelet transform (CWT) method to analyze the cyclical patterns of spring sowing window days (SWD), comprehensive suitability index (SSH_SI), and hydrothermal coordination degree index (D value). The results are presented in Figure 5.

The periodicity of spring sowing window days (SWD) is more concentrated, with a dominant cycle of approximately 4 years. High-value regions in the power spectrum primarily appear in the early study period (2000–2005), indicating significant fluctuations in sowing window length during this phase, likely reflecting adverse impacts of climatic instability on agricultural activities [13]. The power spectrum of the SSH_SI index reveals a dominant cycle of about 6 years, exhibiting stable power distribution across most years. High-power zones are predominantly distributed within the scale range of 5–8, suggesting a potential association with mesoscale climate oscillations such as ENSO34. This implies that its variation may be modulated by longer-term climatic processes [13]. For the D value, the wavelet power spectrum shows distinct high-power regions around scale 3, particularly concentrated in the early 2000s and around 2010. This indicates intense fluctuations in the coordination between meteorological and hydrothermal conditions during these periods [34]. In the corresponding global power spectrum, the power values at this dominant scale are relatively high. Although not all exceed the 90% significance level, they demonstrate discernible periodic characteristics [34]. In summary, the D value, SSH_SI, and SWD all exhibit periodic fluctuations at the 3–6-year scale, highlighting clear interannual variability in hydrothermal climatic conditions and suitability during the spring sowing period [13]. These findings provide a basis for optimizing sowing schedules and formulating medium-to-long-term agroclimatic adaptation strategies.

3.2. Spatial Variation Characteristics of Indicators

To investigate the variation patterns of climatic suitability during the spring sowing period along the Great Wall, this study selects the annual mean SWD values for each sample point from 2000 to 2024 to analyze spatial distribution characteristics. Interpolation mapping enables visualization of the spatial heterogeneity in spring sowing window length across different regions, providing a basis for identifying key high-suitability or high-risk zones. The results are presented in Figure 6.

Figure 6 reveals a spatial pattern of SWD (spring sowing window days) in the study area characterized by lower values in the northeast and southwest, with higher values in the central region. The northeastern corner of Inner Mongolia and western Liaoning exhibit extremely limited spring sowing windows, with only 3–6 days meeting sowing conditions. Central Inner Mongolia, northern Hebei, areas west of Beijing, northern Shanxi, and the mountainous/hilly regions of central Gansu show slightly extended but still short sowing windows (6–9 days). Western Inner Mongolia, northern Shaanxi, and northern Ningxia demonstrate moderate sowing windows (9–11 days), while the central Shanxi plains and central Inner Mongolia enjoy relatively favorable conditions (11–13 days). Areas with >15-day sowing windows are exceptionally rare. The 3D trend analysis shows SWD initially increases then decreases from west to east, with faster reduction rates at higher longitudes, while displaying rapid increases along the southward latitudinal direction.

Sen’s slope of SWD illustrates the spatial distribution of Sen’s slope values along the Great Wall region (2000–2024), reflecting temporal trends in spring sowing windows. Deep brown coloration dominates most areas (particularly Hebei, Liaoning, eastern Inner Mongolia, and central/northern Shanxi), indicating declining SWD trends and compressed suitable sowing periods—likely attributable to intensified spring droughts and asynchronous early warming with insufficient moisture availability. Limited light-yellow zones (e.g., western Shanxi) suggest marginal SWD improvements, possibly from localized hydrothermal optimization. The 3D trend shows gentle longitudinal variation but pronounced latitudinal decline, revealing stronger negative impacts in high-latitude regions (e.g., Inner Mongolia, Liaoning) where advancing spring phenology lacks a matching moisture supply.

CV of SWD demonstrates significant “low-west, high-east” variability in interannual SWD stability (2000–2024). The southwestern region (Gansu, Ningxia, southeastern Gansu) maintains low CV values with stable sowing windows, while northeastern areas (Liaoning, Jilin, eastern Inner Mongolia) exhibit high CV values and agricultural uncertainty. The 3D analysis confirms CV increases eastward, reflecting greater climatic volatility impacts on eastern sowing windows compared to the hydrothermally stable west.

Spatial analysis of coupling coordination degree (D value) and SSH_SI index (Figure 7 and Figure 8) reveals an east-to-west increasing gradient in D values. Eastern regions (Liaoning, Hebei, eastern Inner Mongolia) show poorer coordination, while western areas (Gansu, Ningxia, western Shanxi) demonstrate superior agroclimatic suitability.

Figure 7 illustrates that the D value exhibits a spatially decreasing pattern from the southwest to the northeast. In the southwestern regions (southern Gansu, Ningxia, Shaanxi), hydrothermal coordination is relatively high, with D values predominantly ranging between 0.40 and 0.47. This indicates favorable matching of hydrothermal resources during the spring sowing period, supporting coordinated meteorological conditions conducive to crop sowing and emergence. In the central transitional zone (central-northern Shanxi, western Hebei, central Inner Mongolia), D values gradually increase, reflecting progressive improvement in hydrothermal conditions. Conversely, the northeastern areas (Liaoning, Jilin, eastern Inner Mongolia) show lower D values, signifying poor hydrothermal coordination and higher meteorological risks for agricultural production. The 3D trend plot further reveals directional characteristics: D values decrease progressively from west to east and decline steadily from south to north with increasing latitude.

Figure 8 demonstrates that the SSH_SI index displays a distinct “high-southwest, low-northeast” spatial gradient. Higher values occur in the southwest (Gansu, Ningxia, central-western Inner Mongolia), while lower values dominate the northeast (eastern Inner Mongolia, northern Hebei, northern Shanxi). The 3D trend analysis highlights directional variations: SSH_SI exhibits a clear decreasing trend longitudinally from west to east, indicating weakening hydrothermal coordination eastward. Additionally, SSH_SI decreases latitudinally from south to north as latitude increases.

Further analysis tracks the centroid migration trajectories of SWD, SSH_SI, and D value indicators from 2000 to 2024, as shown in Figure 9.

Figure 9 reveals that the centroids of all three indicators exhibit a migration trend from the northeast (Inner Mongolia and northern Hebei) toward the southwest or south (central Shanxi), indicating a gradual southward shift of optimal spring sowing regions. Specifically, the SWD centroid demonstrates the most pronounced shift, showing consistent southwestward displacement. The SSH_SI centroid has progressively moved southward since 2000, reflecting improved hydrothermal conditions in southern regions. The D value trajectory remains relatively concentrated, signaling the southward migration of high hydrothermal coordination zones. The consistent directional shift highlights a structural transformation in agricultural suitability driven by climate change-induced reorganization of hydrothermal resources. These findings provide critical insights for optimizing agricultural layouts, identifying high-risk areas, and dynamically adjusting sowing schedules.

Further quantitative assessment evaluates spatial displacements and change rates of indicator centroids across different periods, supplying key quantifiable metrics for spatiotemporal evolution (

Table 1. Spatiotemporal migration characteristics of spring sowing indicators.

Indicators	Time	Migration Distance (km)	Migration Velocity (km/year)	Migration Bearing
SWD	2000–2005	26.25	5.25	West-southwest
	2005–2010	10.3	2.06	West
	2010–2015	9.01	1.8	South
	2015–2020	5.35	1.07	North-northwest
	2020–2024	9.53	2.38	West-southwest
SSH_SI	2000–2005	111.58	22.32	North-northeast
	2005–2010	29.58	5.92	East-northeast
	2010–2015	16.37	3.27	South-southwest
	2015–2020	11.12	2.22	West
	2020–2024	21.81	5.45	Southwest
D Value	2000–2005	196.84	39.37	West-southwest
	2005–2010	31.86	6.37	West-southwest
	2010–2015	35.65	7.13	South-southeast
	2015–2020	65.09	13.02	South-southwest
	2020–2024	79.79	19.95	Southwest

).

Based on the data presented in Table 1, a quantitative comparison was conducted on the migration distances and velocities of three key spring sowing indicators—SWD, SSH_SI and D value to identify their distinct spatial response characteristics.

Overall, the D value exhibits the most rapid and directionally consistent migration. Over the five time periods from 2000 to 2024, it accumulates a total migration distance of approximately 409.23 km, with an average of 81.85 km per period and a migration velocity of 15.83 km/year, far exceeding the other two indicators. Particularly strong responses are observed in 2000–2005 (196.84 km, 39.37 km/year) and 2020–2024 (79.79 km, 19.95 km/year). Its dominant migration direction is consistently toward the southwest or southeast, reflecting its high sensitivity to regional-scale hydrothermal changes. As a composite metric, the D value demonstrates strong adaptability and representativeness in capturing large-scale climatic variation.

In contrast, SWD shows the most stable and conservative migration behavior. Its total migration distance over five stages is only 60.44 km, with an average of 12.09 km per stage and a velocity of just 2.11 km/year. The migration direction is mainly toward the west-southwest and west, indicating strong geographical and climatic inertia. A temporary northward shift occurred during 2015–2020, likely due to short-term climatic fluctuations. This suggests that SWD mainly responds to local fine-scale climate regulation, with relatively low spatial variability and high regional stability.

SSH_SI shows the most fluctuating and complex migration patterns. It experienced its most intense migration in 2000–2005 (111.58 km, 22.32 km/year), indicating a rapid initial response to climate disturbances. In the following periods, migration speed decreased significantly (2–6 km/year), but its migration path remained unstable, shifting from northeast → east-northeast → southwest → west → southwest. The cumulative migration distance reached 190.46 km, suggesting that SSH_SI is highly sensitive to the combined effects of meteorological factors and terrain conditions.

In summary, the three indicators exhibit distinct spatial response mechanisms: the D value is sensitive, fast-moving, and directionally stable, making it the most suitable indicator for reflecting large-scale hydrothermal coordination changes; SWD is stable and conservative, reflecting the foundational pattern of regional sowing suitability; and SSH_SI is both responsive and volatile, highlighting the impacts of complex environmental interactions. In-depth analysis and visualization of these migration patterns enhance the depth of spatial quantification and provide critical scientific support for designing regionally tailored spring sowing adaptation strategies.

3.3. Risk Index Analysis

The temporal variation trend and periodic characteristics of the spring sowing risk index are shown in Figure 10 and Figure 11. During the period from 2000 to 2024, the risk index exhibited an overall slight upward trend with significant interannual fluctuations, reflecting the uncertainty in spring sowing climate risk. As seen from the wavelet power diagram, after 2000, there are continuous strong power regions around the approximately 4-year cycle, indicating that sowing risk exhibits significant medium-short-term periodic fluctuations. The global power spectrum diagram further verifies the significance of this dominant cycle; the power value at the 4-year scale is much higher than at other scales and exceeds the 95% significance test threshold, indicating that this periodic change is statistically significant. This result suggests that sowing risk is significantly influenced by certain periodic climate factors and may be correlated with climate systems such as the El Niño–Southern Oscillation (ENSO), providing a basis for agricultural risk prediction and the periodic arrangement of sowing strategies.

The spatial distribution of spring sowing risk generally exhibits a pattern of “lower in the west and south, higher in the east and north”. Low-risk areas include Gansu, Ningxia, western Shaanxi, etc., while high-risk areas include Hebei and eastern Inner Mongolia (Figure 12). A three-dimensional trend diagram further reveals directional trends in the risk index along longitudinal and latitudinal gradients: risks show an increasing trend from west to east along the longitude and a gradual rise from south to north along the latitude. Overall, spring sowing risk demonstrates significant temporal fluctuations and spatially non-stationary distribution. The spatial centroid of the sowing risk index displays a distinct northward migration trend, indicating that agricultural sowing risks progressively advance northward with climatic or environmental changes (Figure 13). Since 2010, the centroid has gradually stabilized in the northern region of the map, reflecting a tendency toward risk pattern stabilization during this period. This figure reveals the dynamic evolutionary characteristics of sowing risk in space, providing crucial evidence for identifying high-risk areas and formulating regional agricultural management strategies, which holds significant importance for agricultural risk assessment and adaptive planning during spring sowing under climate change.

4. Discussion

This study constructs a spring sowing risk index based on three indicators—Spring Sowing Window Days (SWD), Hydrothermal Coordination Degree (D value), and Spring Sowing Hydrothermal Suitability Index (SSH_SI)—and systematically analyzes its spatiotemporal evolution characteristics. Results reveal a rising trend in spring sowing risk from 2000 to 2024, with significant regional heterogeneity and spatial regularity.

Temporal trends: During 2000–2024, the hydrothermal coordination degree (D value) exhibited a significant decline, reflecting intensified climate-driven mismatches between thermal and hydrological conditions [14]. Although the composite suitability index (SSH_SI) and spring sowing window days (SWD) showed no statistically significant overall trends, their interannual volatility increased markedly, highlighting a “high fluctuation, low certainty” pattern in sowing suitability [21]. This phenomenon was particularly pronounced in typical arid and semi-arid regions along the Great Wall (Ningxia, Inner Mongolia, northern Shaanxi, northern Shanxi, and northwestern Hebei), where severe D value declines significantly elevated sowing risks in the Hexi Corridor and central Gansu due to scarce spring precipitation and enhanced evaporation [51].

Cyclical patterns: SWD, D value, SSH_SI, and the spring sowing risk index all displayed dominant 3–6-year cyclical fluctuations, indicating substantial regulation by medium-term climate processes (e.g., ENSO) [52,53,54]. Concurrently, some interannual variations correlated closely with climatic oscillations such as seasonal droughts, precipitation anomalies, and the Pacific Decadal Oscillation (PDO) [55]. Notably, these cyclical features primarily occurred in the early study period and weakened in recent years—a shift potentially linked to amplified climate system volatility and frequent extreme weather events [56]. This further exacerbated interannual instability in spring sowing suitability, particularly sensitivity to micro-disturbances like “late spring coldness” and “spring droughts” [56].

Spatial distribution: SSH_SI, D value, and SWD generally exhibited a “higher in southwest, lower in northeast” spatial pattern. Early-onset spring precipitation and synchronized thermal–hydrological conditions in the southwest (e.g., southern Gansu, Ningxia, northern Shaanxi) resulted in higher hydrothermal coordination. Conversely, the northeast (eastern Inner Mongolia, Liaoning, Jilin) frequently experienced thermal–hydrological asynchrony (“earlier temperature rise, delayed soil moisture availability”), elevating sowing risks [14,21]. This spatial heterogeneity arises from interactions among thermal accumulation, soil moisture, topography, and meteorological factors, reflecting significant spatial complexity in sowing conditions across arid/semi-arid regions.

The derived spring sowing risk index indicates relatively stable, low-risk conditions in the North China Plain and eastern Inner Mongolia. In contrast, arid/semi-arid regions show a “higher in west/north, lower in east/south” risk pattern, aligning with fundamental agroclimatic resource distribution in spring China. While the current normalized and equal-weight indexing method offers strong interpretability and operability, its subjectivity warrants refinement through objective weighting approaches (e.g., principal component analysis, entropy weight method). Integrating multi-source data—remote sensing (e.g., MODIS LST), soil moisture monitoring, and crop growth models—could enable dynamic risk early-warning systems at higher spatiotemporal resolutions [57].

In summary, spring sowing suitability exhibits declining temporal trends, amplified interannual volatility, weakening cyclical signals, and pronounced spatial disparities. Establishing a climate-adaptive framework centered on “continuous suitable windows” and multi-scale risk identification will provide a robust scientific basis for precise adjustments to farming schedules, crop variety deployment, and disaster resilience strategies in arid/semi-arid regions, effectively enhancing regional agricultural climate resilience.

Although this study has constructed a multi-source indicator system—comprising SSH_SI, the D index, and SWD—to reflect agricultural suitability during the spring sowing period, direct validation of these indicators remains limited due to the lack of uniform and continuous ground-based observational data (such as actual sowing dates, emergence times, or crop yields). At present, indices such as SSH_SI and SWD primarily characterize the potential suitability of climatic conditions during the spring sowing period, rather than actual crop responses.

Nevertheless, numerous regional studies show strong consistency with the outputs of this study, indirectly supporting the robustness and practical relevance of the proposed model. For instance, Wang et al. [30], based on daily-scale MCI data from 1960 to 2023, observed an increasing frequency of moderate to severe droughts along the Great Wall region, with the drought center shifting northeastward—demonstrating a spatiotemporal decline in hydrothermal coordination during spring sowing. The Shanxi Climate Center (2024) [58] also reported that the optimal sowing window for spring maize advanced by 5–10 days compared to the climatological norm, indirectly reflecting a reduction in SWD. Wang Wanzhao et al. (2021) [59] and Song Yingnan et al. [60] both analyzed the climatic suitability of maize sowing in western Liaoning and found that the average suitability values were relatively low and showed a declining trend—findings that align closely with the spatial and temporal distribution of SSH_SI in this study. Moreover, a series of national agrometeorological bulletins from 2018 to 2024 [61,62,63,64,65,66,67] consistently reported dual risks of “pre-rain drought followed by post-rain cold spells” across the study region, further confirming the spring sowing risk patterns identified in this study.

Building upon the existing work, future efforts will focus on improving and validating the model in three key directions: (1) incorporating actual sowing date records from agricultural meteorological stations to conduct temporal alignment and bias analysis with simulated SWD values; (2) utilizing annual sowing service bulletins issued by local agricultural authorities to indirectly validate interannual fluctuations in SSH_SI; and (3) integrating remotely sensed crop phenology data (e.g., NDVI inflection point detection) or UAV-based field observations to enhance spatiotemporal cross-validation. These improvements are expected to strengthen the empirical foundation and practical applicability of the proposed indicator system.

5. Conclusions

Based on systematic analysis of key hydrothermal indicators (SWD, SSH_SI, D value) and their composite risk index across the Great Wall region during the 2000–2024 spring sowing period, the following primary conclusions emerge:

Spring sowing conditions along the Great Wall Belt display clear hydrothermal gradients and evolving risks. The sowing window is longest (11–13 days) in the central plains of Inner Mongolia and Shanxi but shrinks to only 3–6 days in northeastern Inner Mongolia, the West Liaohe Plain and western Liaoning; meanwhile, the comprehensive suitability index (SSH_SI) and hydrothermal coordination degree (D) trace a “south-west high, north-east low” pattern, signaling favorable heat–moisture matching in southern Gansu, Ningxia and the Guanzhong–northern Shaanxi corridor and pronounced mismatches farther north-east. From 2000 to 2024, SWD and SSH_SI show no significant decline yet exhibit far greater interannual variability, whereas D values fall significantly at more than 70% of stations, indicating a growing hydrothermal imbalance. An integrated risk index accordingly delineates high-risk zones (eastern Inner Mongolia, northern North China), moderate-risk transition belts (northern Shanxi, northern Shaanxi) and low-risk areas (Gansu, Ningxia, Guanzhong Plain). Wavelet analysis identifies a 3–6-year resonance in all three indicators—linked to ENSO/PDOs—while centroid tracking reveals a northwest-to-southeast-and-back migration of optimal sowing areas under warming. Because the framework relies solely on remote sensing variables and a “continuous suitable-day” logic, it is readily transferable to other arid and semi-arid regions, providing quantitative guidance for precise sowing schedules, drought risk mitigation and cropping pattern optimization.

In summary, climate suitability and risks during spring sowing manifest distinct spatiotemporal heterogeneity in the Great Wall region. The constructed SSH_SI and spring sowing risk index provide scientific sowing schedule references for diverse subregions, offering theoretical foundations and decision-making support for optimizing sowing timelines, agricultural structure, and climate resilience enhancement in arid/semi-arid zones.