Landslides, a prevalent natural disaster, wreak havoc on both human lives and vital infrastructure, making them a significant global concern. Their devastating impact is immeasurable, necessitating proactive measures to minimize their occurrence. The ability to accurately forecast the severity of a landslide, including its potential fatality rate and the scale of destruction it may cause, holds tremendous potential for prevention and mitigation to reduce the risk and the damage caused by a landslide to infrastructure and life. In this study, the spatial variability in severity of landslides (in terms of mortality rates) and its dependence on various meteorological, geographical and soil composition has been attempted to be established. To do this, Ordinary Least Squares (global) and various Geographically Weighted (local) models have been employed to observe the varying relation between mortality rates and its various causative factors. Existence of geographical heterogeneity in the relationships is also investigated. The spatial pattern of landslide mortality and its associations with various causative variables in the South Asian Region are investigated and analysed. Through this, insights into targeting of prevention and mitigation measures for landslides based on a given location can be obtained by studying the various forms of heterogeneous spatial associations observed. The outcomes highlight that the local models in the form of Gaussian GWR and Poisson GWR outperform their global counterparts by a huge margin with better R2 and Adj R2 values. In comparison with Poisson GWR and Gaussian GWR, it is seen that Poisson GWR outperforms Gaussian GWR in terms of Mean Absolute Error, Mean Squared Error and Corrected Akaike Information Criterion. Furthermore, several intriguing local relationships patterns are also noted.

Land subsidence is an environmental geological phenomenon mainly caused by groundwater overexploitation. Long-term overexploitation of groundwater not only causes compaction of aquifer thickness and surface deformation but also leads to the loss of aquifer water storage capacity. The skeleton water storage coefficient (S-k) is an important parameter for evaluating the water storage capacity of aquifer groups. This article proposes a new research framework for obtaining the S-k of different aquifer groups: combining permanent scatter for SAR interferometry technology and a multiscale geographic weighted regression model to obtain subsidence information for different aquifer groups, inverting the S-k of different aquifer groups from the spatial scale, and discussing the deformation characteristics of soil layers under different water head change modes to evaluate the deformation and water storage characteristics of different aquifer groups. This framework is applied to the land subsidence region of the Beijing Plain. We calculated that the settlement proportions of different compression layer groups were 14.75%, 23.65%, 33.44%, and 28.16%. Due to the different lithological compositions and groundwater exploitation of different aquifers, the S-k values exhibit different spatial distribution characteristics. With the continuous development of subsidence, the water storage performance of the aquifer group is continuously declining. These findings contribute to managing the sustainable use of groundwater resources and controlling subsidence. It is demonstrated that the research framework proposed in this article can serve as an effective tool for obtaining settlement information and the S-k of different aquifer groups.

研究冻土地温空间分布，有助于探索冻土活动层厚度的变化特征，为冻土灾害防治提供科学依据。以青藏铁路昆仑山至尺曲谷地段多年冻土覆盖区域为研究区域，采用地理加权岭回归克里金（GWRRK）方法对该区域2001年7月至9月的地温空间分布进行了模拟，揭示了该区域多年冻土融化深度的变化特征。结果表明：研究区域内多年冻土地温总体表现为山区地温低于平原和盆地地区地温；地温随深度的增加而降低，在0～5m的深度区间内温度变化较大，平均温差为10.3℃，而在5～15m的深度区间内基本保持不变，平均温差仅0.2℃。通过将GWRRK方法与具有外部漂移克里金（KED）方法和地理加权岭回归（GWRR）方法的模拟效果进行对比，发现前者的模拟精度优于后两种方法。

研究冻土地温空间分布，有助于探索冻土活动层厚度的变化特征，为冻土灾害防治提供科学依据。以青藏铁路昆仑山至尺曲谷地段多年冻土覆盖区域为研究区域，采用地理加权岭回归克里金（GWRRK）方法对该区域2001年7月至9月的地温空间分布进行了模拟，揭示了该区域多年冻土融化深度的变化特征。结果表明：研究区域内多年冻土地温总体表现为山区地温低于平原和盆地地区地温；地温随深度的增加而降低，在0～5m的深度区间内温度变化较大，平均温差为10.3℃，而在5～15m的深度区间内基本保持不变，平均温差仅0.2℃。通过将GWRRK方法与具有外部漂移克里金（KED）方法和地理加权岭回归（GWRR）方法的模拟效果进行对比，发现前者的模拟精度优于后两种方法。