Global agricultural drought is evolving into a compound crisis threatening food security and ecological stability that is characterized by increased frequency, intensity, duration, and spatial extent. Since 2000, the global number of drought events has increased by 29% compared with the previous two decades, and 82% of drought-related losses in developing countries are concentrated in agriculture. The UNCCD (2022) projects that drought may affect up to three quarters of the world’s population by 2050. Climate change and human activities jointly drive this escalation through higher atmospheric evaporative demand, altered precipitation regimes, land use change, groundwater overexploitation, and pollution emissions. Their interaction forms amplifying feedback loops across ecology-agriculture and climate–agriculture systems, resulting in ecosystem degradation, crop yield loss, and rising socioeconomic inequality. Addressing this crisis requires a three-dimensional framework integrating mitigation, adaptation, and collaborative governance. This review synthesizes coupled driving mechanisms, cross-system impacts, and response pathways, and it also highlights priorities in compound-drought attribution, region-specific technology scaling, cross-scale governance, and resilience building in vulnerable regions.

Leaf water content (LWC) is a key physiological indicator for assessing crop water status. However, its spectral response may vary under different irrigation practices, which limits the general applicability of existing models. This study aims to develop irrigation-specific LWC estimation models for spring wheat based on UAV multispectral imagery. Field experiments were conducted during two growing seasons (2023–2024) under three irrigation methods, with five water treatments and three replicates, resulting in a total of 45 experimental plots. Multispectral data and in situ measurements were collected at key growth stages. Irrigation-dependent sensitive vegetation indices were identified through correlation analysis, and machine learning models, including Random Forest (RF), Multiple Linear Regression (MLR), and Backpropagation Neural Network (BPNN), were constructed and evaluated using a five-fold cross-validation framework. The results showed that spectral sensitivity to LWC varied significantly across irrigation methods, with different dominant indicators under FD, ND, and MD. Model performance also exhibited irrigation-dependent differences. Among the three models, RF showed the most stable performance, achieving mean R² values of 0.70, 0.74, and 0.62 and corresponding RMSE values of 0.04, 0.06, and 0.08 under FD, ND, and MD, respectively. In contrast, MLR showed lower predictive accuracy, while BPNN exhibited limited robustness under the current dataset, particularly under ND. These findings highlight the importance of irrigation-specific modeling strategies for improving LWC estimation reliability.

To clarify the optimal water regulation strategy for spring wheat in arid areas, this study set up three irrigation methods [film-mulched drip irrigation (FD), non-mulched drip irrigation (ND), non-mulched subsurface drip irrigation (MD)] and five water treatments [CK: 80% field capacity; W1–W4: irrigation amounts were 90%, 80%, 70%, and 60% of CK, respectively] in the Shiyang River Basin during 2023–2024. The effects of these treatments on the phenotype, yield, and water use efficiency (WUE) of spring wheat were investigated. The results showed that under the same water treatment, the leaf area index (LAI), SPAD value, and stem diameter (SD) significantly decreased with the reduction in irrigation amount (p < 0.05), while plant height (HC) was less affected. FD performed optimally under the W1 treatment: its yield reached 11,868.93 kg·ha⁻¹, which was 54.88% and 38.72% higher than that of ND and MD, respectively; and its WUE reached 4.36 kg/m³, which was 123.19% and 100.83% higher than that of ND and MD, respectively. ND performed better under the CK treatment: its yield was 10,044.33 kg·ha⁻¹, which was 27.07% and 12.25% higher than that of FD and MD, respectively. Annual precipitation had a significant impact: when precipitation was 175 mm in 2023, ND showed an obvious advantage; when precipitation decreased to 110 mm in 2024, FD exhibited stronger stress resistance. The study concludes that FD is suitable for moderate to severe water stress, while ND is suitable for sufficient water conditions or mild stress. This can provide a basis for water-saving and the high-yield production of spring wheat in arid areas.

Global warming is reshaping terrestrial water cycling and near-surface climate risks through atmospheric moistening, enhanced precipitation variability, rising evaporative demand, and more frequent compound extremes. This narrative review synthesizes recent advances in land–atmosphere interactions and atmospheric water cycle feedbacks, and its distinctive contribution is to connect physical feedback chains with human land surface perturbations, compound risk, and observation model machine learning evidence. We reviewed the literature from Web of Science, Scopus, Google Scholar, publisher databases, and Crossref metadata, prioritizing peer-reviewed studies published mainly during 2010–2026 while retaining foundational work on soil moisture feedbacks, moisture recycling, irrigation, aerosols, and boundary-layer processes. The synthesis emphasizes where evidence is robust, where feedback signs are regime dependent, and where uncertainty still propagates from evapotranspiration partitioning, boundary-layer diagnosis, aerosol–cloud interactions, human water management, and nonstationary climate conditions. The review concludes that the same land surface perturbation may cool locally, increase humid heat exposure, alter downwind precipitation, or intensify water depletion, depending on the climate regime, season, scale, and management. Future research should therefore move beyond single-variable correlation analyses toward causal, cross-scale, and risk-oriented attribution frameworks that integrate multi-source observations, process models, moisture tracking, and physically constrained machine learning.