Key points are not available for this paper at this time.
Climate change is linked to changes in rainfall patterns and an increased susceptibility to both floods and droughts. There is a pressing need to enhance forecasting and monitoring technologies in response to the occurrence of flash floods and severe flooding. Rainfall prediction is a challenging but critical issue owing to the complexity of atmospheric processes, the spatial and temporal variability of rainfall, and the dependency of this variability on several nonlinear factors. Accurate real-time rainfall nowcast is crucial for taking necessary precautions, implementing control measures, and planning, as excessive rainfall is responsible for disasters including floods and landslides. This study examines the application of NASA Giovanni satellite-derived precipitation products with the ConvLSTM method, a variant of LSTM, for the purpose of nowcasting rainfall. Data augmentation is employed using interpolation techniques including nearest neighbor, bilinear, bicubic, distance weighted average, first and second order conservative and largest area fraction techniques to meet the data requirements of deep learning-based prediction algorithms. The objective of the study is to evaluate the influence of data augmentation on flood nowcasting. Thanks to these methods, the study utilized three types of satellite-derived rainfall data, including spatial, temporal, and spatiotemporal interpolated rainfall data, to conduct a comparative analysis of the results obtained through nowcasting rainfall. This research examines two catastrophic floods that transpired in Trkiye Marmara Region in 2009 and Central Black Sea Region in 2021, which are selected as the focal case studies. The Marmara and Black Sea regions are prone to frequent flooding, which, due to the dense population, has devastating consequences. Furthermore, these regions exhibit distinct topographical characteristics and precipitation patterns, and the frontal systems that impact them are also dissimilar. The results of nowcasts for floods exhibit substantial variations, both between the two regions and across nowcasts generated using raw and augmented data.
Baydaroğlu et al. (Fri,) studied this question.