Soil pH is crucial for the quality of the farmland and crop growth. The objective of this study is to analyze the spatial and temporal variations of farmland soil pH in Xinjiang (XJ), and to provide a scientific basis for soil improvement and agricultural production optimization. Based on soil pH data from XJ farmland in two periods, 2008~2010 and 2019~2021, geostatistical methods and kriging spatial interpolation techniques were employed to analyze the spatiotemporal changes in soil pH and to investigate the relationship between relevant influencing factors and pH over more than a decade. The results indicate that the spatiotemporal distribution of soil pH in XJ farmlands is uneven. Between 2019 and 2021, the average soil pH was 8.11, generally being on the higher side, with a coefficient of variation (CV) of 3.85%, indicating minimal spatial variability. In the farmland soil of Southern Xinjiang (S-XJ), the average pH value was 8.17, significantly higher than that of Northern Xinjiang (N-XJ), which was 8.10, demonstrating a spatial pattern of “higher in the south, lower in the north”. Over the past decade, soil pH in XJ has significantly increased from 8.11 to 8.13, with a 0.05 unit increase in the north and a 0.01 unit increase in the south (p < 0.05). Regionally, Altay saw the largest increase of 0.22 units, while Bortala Mongol Autonomous Prefecture (Bortala) experienced the most significant decrease, dropping by 0.59 units. Furthermore, this study found that factors such as topography, nutrients, and irrigation methods all have certain influences on the spatial distribution of soil pH in XJ farmland, while variations in climate factors and fertilization levels may affect its long-term temporal changes. These research findings will provide new insights for adjusting and updating agricultural management measures related to soil pH regulation in XJ.

Spatiotemporal variability of soil salinity and the driving factors of cultivated land in Xinjiang, China

Nitrogen (N), phosphorus (P), and potassium (K) are important nutrients limiting cotton yield. Although N, P, and K fertilization is an important practice for improving cotton productivity, their fertilizer use efficiencies remain insufficiently evaluated in Xinjiang, China. In this study, field experiments were conducted at 57 sites across the major cotton-growing areas of Xinjiang during 2021–2022. Four fertilization treatments, PK, NK, NP, and NPK, were established to evaluate cotton yield response, nutrient accumulation, and fertilizer use efficiencies of N, P, and K under different soil fertility conditions. The results indicated that the cotton yield in both southern and northern Xinjiang significantly increased under NPK treatment, with mean yields of 6300 and 7361 kg ha⁻¹, respectively. The use efficiencies of N, P, and K were 39.8%, 19.5%, and 51.4% in southern Xinjiang, and 40.3%, 20.2%, and 66.1% in northern Xinjiang, respectively. For every 100 kg of seed cotton produced, nutrient uptake was 4.64 kg N, 1.43 kg P₂O₅, and 3.67 kg K₂O in southern Xinjiang, and 4.74 kg N, 1.48 kg P₂O₅, and 4.58 kg K₂O in northern Xinjiang, indicating relatively high N and K requirements but lower P demand in cotton production. Cotton yield in southern Xinjiang was significantly and positively correlated with soil alkali-hydrolyzable N. In northern Xinjiang, cotton yield was significantly and positively correlated with soil organic matter, alkali-hydrolyzable N, and available K, but negatively correlated with available P. Overall, cotton yield was mainly positively affected by soil organic matter, alkali-hydrolyzable N, and available K, whereas K use efficiency was significantly and negatively correlated with N and K rates. The suitable N, P, and K rates for cotton were 285, 120, and 106 kg ha⁻¹ in southern Xinjiang, and 243, 101, and 91 kg ha⁻¹ in northern Xinjiang, respectively. Based on soil fertility classification of cotton fields, the recommended N, P, and K rates were 256–313, 96–117, and 125–153 kg ha⁻¹ in southern Xinjiang, and 219–267, 97–119, and 136–166 kg ha⁻¹ in northern Xinjiang, respectively. Accordingly, reducing N and P inputs in southern Xinjiang, maintaining N and P at appropriate levels in northern Xinjiang, and moderately increasing K are recommended to achieve coordinated improvements in yield stability and N, P, and K use efficiencies.

Soil salinization constrains the sustainability of irrigated oasis agriculture in arid regions. Using repeated post-harvest monitoring of 125 fixed cropland sites in Bachu County, southern Xinjiang, from 2018 to 2021, this study investigated the short-term spatiotemporal variability of topsoil total salt content (TSC) and pH. Descriptive statistics, one-way ANOVA with Tukey’s HSD test, Universal Kriging interpolation, class-transition analysis, hotspot recurrence, centroid migration, and principal component analysis were used to characterize temporal variation, spatial structure, and environmental gradient associations. TSC showed a mitigation–rebound sequence, decreasing to 4.88 ± 5.21 g kg⁻¹ in 2020 and increasing to 6.90 ± 5.93 g kg⁻¹ in 2021, whereas pH increased first and then declined. Salinity remained consistently concentrated in downstream cropland, while pH showed weaker and more year-dependent zonal differentiation. Class-transition analysis revealed marked salinity reorganization in 2021, mainly driven by conversion from lower-salinity classes to moderately and severely saline classes. Severe-salinity hotspots were temporally intermittent but spatially recurrent in the downstream zone, whereas high-pH hotspots were short-lived and mainly confined to the upstream zone. PCA further showed that TSC and pH were aligned with different environmental gradient combinations. Overall, the four-year sequence should be interpreted as short-term interannual variability rather than a robust long-term sequence. These results indicate that TSC and pH should not be treated as interchangeable indicators in oasis cropland assessment, and they provide a transferable basis for zone-specific salinity monitoring and management, with priority given to persistent downstream sink areas.

Accurate and timely cotton-field mapping is essential for irrigation management, water resource allocation, and regional yield assessment in arid irrigated agroecosystems. However, existing deep-learning-based crop mapping approaches generally lack interpretability and often exhibit performance variability across phenological stages, thereby limiting their reliability for operational deployment. To address these limitations, we developed an interpretable semantic segmentation framework for cotton mapping in the Wei-Ku Oasis, Xinjiang, China, under multi-source remote sensing conditions. The proposed model integrates Sentinel-2 surface reflectance, Sentinel-1 VV/VH backscatter, DEM, vegetation indices, and GLCM texture features. By incorporating a receptive-field enhancement mechanism together with an embedded feature-attribution module, the framework enables importance estimation of multi-source predictors within the network architecture, thereby providing intrinsic model interpretability. Under a unified training and evaluation protocol, the proposed model achieved an mIoU of 85.62% and an F1-score of 92.96% on the test set, outperforming U-Net, DeepLabV3+, and SegFormer baselines. Monthly classification results indicated that August provided the most discriminative acquisition window (mIoU = 85.54%, F1 = 92.83%), while June–July also maintained high recognition accuracy. Feature attribution results indicate that the importance of different predictors varies across phenological stages: Sentinel-2 red-edge bands remained highly influential throughout the growing season, NDVI/EVI exhibited increased contributions during June–August, SAR VH showed relatively higher importance during peak canopy development, and DEM maintained stable information contribution across all stages. Cross-year and cross-region experiments further demonstrated the model’s generalization capability, achieving an mIoU of 82.81% in same-region cross-year evaluation and 74.56% under cross-region transfer. Overall, the proposed segmentation framework improves classification accuracy while explicitly modeling and quantifying feature importance, providing a methodological reference for cotton-field mapping and acquisition timing selection in arid irrigated regions.