As global climate change intensifies, flood disasters occur frequently, causing severe impacts on agriculture and the socioeconomic environment. Accurate prediction of the affected area of flood disasters is crucial. Considering that historical flood disaster years exhibit characteristics such as non-equidistant intervals, multivariable influences and strong nonlinearity – while existing studies mainly focus on improvements to equidistant grey multivariable models or non-equigap GM (1,1) and MGM (1, m) models – this study proposes a combined forecasting approach integrating the grey disaster model with a non-equigap multivariable grey prediction model (NE-GM (1, N)).
First, the grey disaster model is used to predict future disaster years, determining the specific time points for potential flood disasters. Then, by introducing factors such as precipitation and population density, the NE-GM (1, N) model is applied to predict the affected area of flood disasters for those years. This model integrates the principle of giving priority to new information and polynomial expansion techniques, enhancing the response capability to real-time changes and improving prediction accuracy by capturing nonlinear relationships.
Verification with data from Hunan and Hubei Provinces shows that this model outperforms traditional methods in terms of prediction accuracy and stability.
Providing more accurate information for disaster early warning and emergency management, while laying the foundation for further research.
