Choosing downscaling techniques is crucial in obtaining accurate and reliable climate change predictions, allowing for detailed impact assessments of climate change at regional and local scales. Traditional statistical methods are likely inefficient in downscaling precipitation data from multiple sources or complex data patterns, so using deep learning, a form of nonlinear models, could be a promising solution. In this study, we proposed to use deep learning models, the so-called long short-term memory and feedforward neural network methods, for precipitation downscaling for the Vietnamese Mekong Delta. Model performances were assessed for 2036–2065 period, using original climate projections from five climate models under the Coupled Model Intercomparison Project Phase 5, for two Representative Concentration Pathway scenarios (RCP 4.5 and RCP 8.5). The results exhibited that there were good correlations between the modelled and observed values of the testing and validating periods at two long-term meteorological stations (Can Tho and Chau Doc). We then analysed extreme indices of precipitation, including the annual maximum wet day frequency (Prcp), 95th percentile of precipitation (P95p), maximum 5-day consecutive rain (R5d), total number of wet days (Ptot), wet day precipitation (SDII) and annual maximum dry day frequency (Pcdd) to evaluate changes in extreme precipitation events. All the five models under the two scenarios predicted that precipitation would increase in the wet season (June–October) and decrease in the dry season (November–May) in the future compared to the present-day scenario. On average, the means of multiannual wet season precipitation would increase by 20.4 and 25.4% at Can Tho and Chau Doc, respectively, but in the dry season, these values were projected to decrease by 10 and 5.3%. All the climate extreme indices would increase in the period of 2036–2065 in comparison to the baseline. Overall, the developed downscaling models can successfully reproduce historical rainfall patterns and downscale projected precipitation data.     «

Choosing downscaling techniques is crucial in obtaining accurate and reliable climate change predictions, allowing for detailed impact assessments of climate change at regional and local scales. Traditional statistical methods are likely inefficient in downscaling precipitation data from multiple sources or complex data patterns, so using deep learning, a form of nonlinear models, could be a promising solution. In this study, we proposed to use deep learning models, the so-called long shor...     »