Scholarlink

<p>Given the prevailing context of climate change, this study presents a multi-index response modeling framework to quantify the population-averaged effects of meteorological and topographic covariates on long-term, month-specific drought events. The framework explicitly accommodates intra-station temporal dependence, addresses multicollinearity among predictors, and evaluates the sensitivity of results across alternative drought indices, including the Standardized Precipitation Index (SPI), the Standardized Precipitation Temperature Index (SPTI), and the Standardized Precipitation Evapotranspiration Index (SPEI). The population-averaged Autocorrelation Function (ACF) and Partial Autocorrelation Function (PACF) plots reveal strong serial correlation in drought counts, indicating that short-term seasonal transitions dominate the temporal dependence structure. Complementing this, month-specific Global Moran’s I statistics demonstrate significant spatial clustering of drought counts across all months, with slightly weaker clustering observed during the monsoon transition, highlighting the joint influence of large-scale atmospheric processes and local physiographic factors in shaping drought dynamics. Building on these insights, Lasso regression effectively addresses multicollinearity and facilitates variable selection by consistently retaining the key domain-relevant predictors across all drought indices, including temperature at 2&#xa0;m (T2M), absolute humidity at 2&#xa0;m (QV2M), wind speed at 10&#xa0;m (WS10M), wind direction at 10&#xa0;m (WD10M), and topographic relief. Functional‑form of Generalized Estimating Equation (GEE) models was explored across alternative working correlation structures, viz. Exchangeable and AR(1) based on QIC, QICu, QICC, and CIC. Results from the final GEE models indicate that rising temperature and changes in prevailing wind direction are associated with increased month-specific drought counts, whereas higher absolute humidity and greater topographic relief are associated with reduced drought count. Notably, SPEI-based GEE models exhibit superior performance across all performance matrices, highlighting the necessity of incorporating evaporative demand when modeling drought in the climate-sensitive regions. This study advances drought risk assessment by integrating climatic, physiographic, and temporal dynamics of drought within a unified framework, offering methodological template for researchers and actionable insights to decision makers for effective climate adaptation, water resource management and agricultural planning in climate‑sensitive regions.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Integrating climatic and topographic drivers in a poisson-GEE framework for modelling drought dynamics

  • 期刊论文

  • 发表日期：

    2026年5月4日

查看全文
  • Rizwan Farooq,
  • Maysaa Elmahi Abd Elwahab,
  • Rizwan Niaz,
  • Hefa Cheng,
  • Mhassen. E. E. Dalam,
  • Mohammed M. A. Almazah,
  • Ijaz Hussain

摘要

Given the prevailing context of climate change, this study presents a multi-index response modeling framework to quantify the population-averaged effects of meteorological and topographic covariates on long-term, month-specific drought events. The framework explicitly accommodates intra-station temporal dependence, addresses multicollinearity among predictors, and evaluates the sensitivity of results across alternative drought indices, including the Standardized Precipitation Index (SPI), the Standardized Precipitation Temperature Index (SPTI), and the Standardized Precipitation Evapotranspiration Index (SPEI). The population-averaged Autocorrelation Function (ACF) and Partial Autocorrelation Function (PACF) plots reveal strong serial correlation in drought counts, indicating that short-term seasonal transitions dominate the temporal dependence structure. Complementing this, month-specific Global Moran’s I statistics demonstrate significant spatial clustering of drought counts across all months, with slightly weaker clustering observed during the monsoon transition, highlighting the joint influence of large-scale atmospheric processes and local physiographic factors in shaping drought dynamics. Building on these insights, Lasso regression effectively addresses multicollinearity and facilitates variable selection by consistently retaining the key domain-relevant predictors across all drought indices, including temperature at 2 m (T2M), absolute humidity at 2 m (QV2M), wind speed at 10 m (WS10M), wind direction at 10 m (WD10M), and topographic relief. Functional‑form of Generalized Estimating Equation (GEE) models was explored across alternative working correlation structures, viz. Exchangeable and AR(1) based on QIC, QICu, QICC, and CIC. Results from the final GEE models indicate that rising temperature and changes in prevailing wind direction are associated with increased month-specific drought counts, whereas higher absolute humidity and greater topographic relief are associated with reduced drought count. Notably, SPEI-based GEE models exhibit superior performance across all performance matrices, highlighting the necessity of incorporating evaporative demand when modeling drought in the climate-sensitive regions. This study advances drought risk assessment by integrating climatic, physiographic, and temporal dynamics of drought within a unified framework, offering methodological template for researchers and actionable insights to decision makers for effective climate adaptation, water resource management and agricultural planning in climate‑sensitive regions.