Droughts are increasingly threatening ecological balance, agricultural productivity, and socio-economic resilience—especially in semi-arid regions like the Inner Mongolia segment of China’s Yellow River Basin. This study presents a hybrid drought forecasting framework integrating machine learning (ML) and deep learning (DL) models with high-resolution historical and downscaled future climate data. TerraClimate observations (1985–2014) and bias-corrected CMIP6 projections (2030–2050) under SSP2-4.5 and SSP5-8.5 scenarios were utilized to develop and evaluate the models. Among the tested ML algorithms, Random Forest (RF) demonstrated the best trade-off between accuracy and interpretability and was selected for feature importance analysis. The top-ranked predictors—precipitation, solar radiation, and maximum temperature—were used to train a Long Short-Term Memory (LSTM) network. The LSTM outperformed all ML models, achieving high predictive skill (R² = 0.766, CC = 0.880, RMSE = 0.885). Scenario-based projections revealed increasing drought severity and variability under SSP5-8.5, with mean PDSI values dropping below −3 after 2040 and deepening toward −4 by 2049. The high-emission scenario also exhibited broader uncertainty bands and amplified interannual anomalies. These findings highlight the value of hybrid AI–climate modeling approaches in capturing complex drought dynamics and supporting anticipatory water resource planning in vulnerable dryland environments.

Agricultural and livestock production plays an important role in Inner Mongolia, China, but its continued development poses challenges for greenhouse gas mitigation. Using multi-source data from 2005 to 2023, this study integrates the IPCC emission-factor approach, panel STIRPAT modeling, uncertainty and sensitivity analyses, and Ridge regression forecasting to quantify carbon-equivalent emissions and emission intensity from selected agricultural and livestock sources, examine emission changes and associated factors, and project future emissions under a trend-continuation scenario (TCS) and a policy-target scenario (PTS). Total emissions increased from 9.8472 Mt C-eq in 2005 to 14.5223 Mt C-eq in 2023, while carbon-equivalent emission intensity declined over the same period. Livestock-related emissions remained dominant, although the contribution of agricultural emissions increased over time, with cattle and sheep accounting for most livestock-related emissions. Panel estimates showed significant positive associations between emissions and year-end rural population, year-end large-livestock inventory, and real agricultural and animal husbandry output value per capita at constant 2005 prices. Monte Carlo analysis showed that emission-factor uncertainty had a notable influence on absolute emission levels, with a coefficient of variation of 9.04% in 2023; enteric-fermentation CH₄ emission factors for cattle and sheep were the dominant sources of uncertainty. Among the models compared, Ridge regression showed the strongest overall validation performance. Emissions are projected to continue increasing through 2035 under both the TCS and PTS, reaching 16.5974 and 16.5012 Mt C-eq, respectively, indicating that the currently quantifiable policy constraints alone are insufficient to reverse the projected upward trend. Future mitigation should therefore prioritize major livestock-related emission sources and regionally differentiated management, while further refinement of the emission inventory should focus on developing more regionally representative livestock emission factors.

The Gaizi River Basin, an alpine region in China crossed by the Karakoram Highway, is highly prone to glacier-related debris flows (GDF). Accurate debris flow susceptibility assessment in this high-altitude area remains challenging due to complex terrain, active tectonics, and dynamic glacial processes. This study develops a hybrid model integrating statistical methods and machine learning-based anomaly detection for debris flow susceptibility mapping. To address data noise, certainty factor (CF) distributions of debris flow predisposing factors (DFPFs) were derived via Locally Weighted Scatterplot Smoothing (LOWESS). The strength of the association between DFPFs and GDF susceptibility was evaluated using the mean residual between the raw and LOWESS-smoothed CF values. Multiple anomaly detection algorithms, including distance-based (L2 Norm), density-based (One-Class SVM), ensemble (Isolation Forest, RandNet), and GAN-based (WBiGAN-GP) methods, were tested on raw and CF-transformed data, using only the GDF inventory as the label. The CF-WBiGAN-GP model delivers the most balanced performance, excelling at identifying both high- and low-susceptibility zones. Results show that distance to stream, slope, and the topographic roughness and wetness indices are strongly associated with GDF susceptibility. Distance to glacier and precipitation appear less informative for direct susceptibility inference under our specific dataset and analytical setup.