Search

• Based on the principles of SPEI, we develop a GDAI. • The relationship between GDAI and AI is closer. • The correlation between GDAI and SPEI is also better. Climate change has increased the risk of drought in both arid and non-arid regions of the globe. Therefore, in order to monitor and evaluate global drought and aridity, based on the principle of standardized precipitation evapotranspiration index (SPEI), this paper selected precipitation and potential evapotranspiration data and used a frequency calculation method to develop a multi-scale global drought aridity index (GDAI) that concurrently considers spatial and temporal variation characteristics: the spatiotemporal standardized precipitation evapotranspiration index. The index was compared with the aridity index (AI) and SPEI. The results showed that the correlation coefficients between GDAI and AI were greater than 0.8 in most areas, and the relationship was closer and could reveal the global aridity and wetness zoning. In addition, the correlation between GDAI and SPEI was also better, which could capture the temporal change characteristics of drought. Moreover, this study also compared the grades of GDAI and SPEI with the actual drought grades in Yunnan from September 2009 to August 2010, and found that GDAI was better than SPEI, and GDAI was the most accurately evaluated at the 3 month time scale, with an accuracy rate of 86.11%. Therefore, this index is more helpful for the assessment and monitoring of global drought and aridity. [ABSTRACT FROM AUTHOR]

• A Decision Support System for flash-flood risk management in the arid areas is designed. • An integrated GIS and Watershed Modelling System is applied for flood flow simulation. • AI and Classic Delphi are employed for smart sustainable management of flash-flood risks. • Flood risk is analysed using Failure Mode and Effects Analysis & Hazard & Operability techniques. This paper presents a novel framework for smart integrated risk management in arid regions. The framework combines flash flood modelling, statistical methods, artificial intelligence (AI), geographic evaluations, risk analysis, and decision-making modules to enhance community resilience. Flash flood is simulated by using Watershed Modelling System (WMS). Statistical methods are also used to trim outlier data from physical systems and climatic data. Furthermore, three AI methods, including Support Vector Machine (SVM), Artificial Neural Network (ANN), and Nearest Neighbours Classification (NNC), are used to predict and classify flash flood occurrences. Geographic Information System (GIS) is also utilised to assess potential risks in vulnerable regions, together with Failure Mode and Effects Analysis (FMEA) and Hazard and Operability Study (HAZOP) methods. The decision-making module employs the Classic Delphi technique to classify the appropriate solutions for flood risk control. The methodology is demonstrated by its application to the real case study of the Khosf region in Iran, which suffers from both drought and severe floods simultaneously, exacerbated by recent climate changes. The results show high Coefficient of determination (R2) scores for the three AI methods, with SVM at 0.88, ANN at 0.79, and NNC at 0.89. FMEA results indicate that over 50% of scenarios are at high flood risk, while HAZOP indicates 30% of scenarios with the same risk rate. Additionally, peak flows of over 24 m3/s are considered flood occurrences that can cause financial damage in all scenarios and risk techniques of the case study. Finally, our research findings indicate a practical decision support system that is compatible with sustainable development concepts and can enhance community resilience in arid regions. [ABSTRACT FROM AUTHOR]