Model Construction and Evaluation of Flood Area Estimation Based on SAR and GPS Data

In this study, SAR data, DEM, basic geographic data, and mobile phone GPS location data were used to construct flood area estimation models for the city of Nagano and the surrounding area in case of the October 2019 flood in the Chikuma River Basin in Nagano Prefecture, Japan, as an example. The effect of mobile phone GPS location data on the flood area estimation model was evaluated by taking advantage of the ability of SAR to pass through clouds and the accurate recording of population movement by mobile phone GPS location data. The results of the study showed that the combined data could construct a flood prediction model with an accuracy of approximately 0.8. Specifically, the accuracy of the flood area estimation model without GPS data was 0.799, and the accuracy of the flood area estimation model with GPS data was 0.815. It was worth noting that GPS data could have a positive impact on the accuracy of the model in more populated building areas. This study was extended to the flood area estimation models and determines the importance of GPS data in improving model accuracy, which has practical implications for improving flood disaster monitoring and response.