Probabilistic rainfall thresholds in Chibo, India: estimation and validation using monitoring system

The Himalayan region has been severely affected by landslides especially during the monsoons. In particular, Kalimpong region in Darjeeling Himalayas has recorded several landslides and has caused significant loss of life, property and agricultural land. The study region, Chibo has experienced several landslides in the past which were mainly debris and earth slide. Globally, several types of rainfall thresholds have been used to determine rainfall-induced landslide incidents. In this paper, probabilistic thresholds have been defined as it would provide a better understanding compared to deterministic thresholds which provide binary results, i.e., either landslide or no landslide for a particular rainfall event. Not much research has been carried out towards validation of rainfall thresholds using an effective and robust monitoring system. The thresholds are then validated using a reliable system utilizing Microelectromechanical Systems(MEMS) tilt sensor and volumetric water content sensor installed in the region. The system measures the tilt of the instrument which is installed at shallow depths and is ideal for an early warning system for shallow landslides. The change in observed tilt angles due to rainfall would give an understanding of the applicability of the probabilistic model. The probabilities determined using Bayes' theorem have been calculated using the rainfall parameters and landslide data in 2010-2016. The rainfall values were collected from an automatic rain gauge setup near the Chibo region. The probabilities were validated using the MEMS based monitoring system setup in Chibo for the monsoon season of 2017. This is the first attempt to determine probabilities and validate it with a robust and effective monitoring system in Darjeeling Himalayas. This study would help in developing an early warning system for regions where the installation of monitoring systems may not be feasible.

Spatial flood susceptibility mapping using an explainable artificial intelligence(XAI)model

Floods are natural hazards that lead to devastating financial losses and large displacements of people.Flood susceptibility maps can improve mitigation measures according to the specific conditions of a study area.The design of flood susceptibility maps has been enhanced through use of hybrid machine learning and deep learning models.Although these models have achieved better accuracy than traditional models,they are not widely used by stakeholders due to their black-box nature.In this study,we propose the application of an explainable artificial intelligence(XAI)model that incorporates the Shapley additive explanation(SHAP)model to interpret the outcomes of convolutional neural network(CNN)deep learning models,and analyze the impact of variables on flood susceptibility mapping.This study was conducted in Jinju Province,South Korea,which has a long history of flood events.Model performance was evaluated using the area under the receiver operating characteristic curve(AUROC),which showed a prediction accuracy of 88.4%.SHAP plots showed that land use and various soil attributes significantly affected flood susceptibility in the study area.In light of these findings,we recommend the use of XAIbased models in future flood susceptibility mapping studies to improve interpretations of model outcomes,and build trust among stakeholders during the flood-related decision-making process.

Artificial Intelligence:A new era for spatial modelling and interpreting climate-induced hazard assessment

The application of Artificial Intelligence in various fields has witnessed tremendous progress in the recent years.The field of geosciences and natural hazard modelling has also benefitted immensely from the introduction of novel algorithms,the availability of large quantities of data,and the increase in computational capacity.The enhancement in algorithms can be largely attributed to the elevated complexity of the network architecture and the heightened level of abstraction found in the network's later layers.As a result,AI models lack transparency and accountability,often being dubbed as"black box"models.Explainable AI(XAI)is emerging as a solution to make AI models more transparent,especially in domains where transparency is essential.Much discussion surrounds the use of XAI for diverse purposes,as researchers explore its applications across various domains.With the growing body of research papers on XAI case studies,it has become increasingly important to address existing gaps in the literature.The current literature lacks a comprehensive understanding of the capabilities,limitations,and practical implications of XAI.This study provides a comprehensive overview of what constitutes XAI,how it is being used and potential applications in hydrometeorological natural hazards.It aims to serve as a useful reference for researchers,practitioners,and stakeholders who are currently using or intending to adopt XAI,thereby contributing to the advancements for wider acceptance of XAI in the future.