Deep Learning-Based Analyses of Feedback Mechanisms in the Land-Atmosphere Interactions during Droughts over the Vietnamese Mekong Delta

The Vietnamese Mekong Delta (VMD) is the most productive region in Vietnam in terms of agriculture and aquaculture. Unsurprisingly, droughts have emerged as a persistent concern for stakeholders throughout the VMD in recent decades. In the evolution and intensification of droughts, local feedbacks in the Land-Atmosphere (LA) interactions were considered to play a crucial role. Previous studies mainly focused on the water cycle feedback loop (e.g., soil moisture-evaporation-precipitation) in the LA interactions. However, there is a noticeable gap in the feedback loop of coupled water and energy balances (e.g., soil moisture-sensible heat-precipitation) associated with the anomalies in sensible heat and precipitation. Therefore, deep understanding of the roles of key variables and their inter-relationships in the LA interactions is of great significance for local communities and authorities. In this study, a deep learning model, named Long- and Short-term Time-series Network (LSTNet), was applied to simulate the LA interactions over the VMD. With the ERA5 data as modelling inputs, the role of each key variable (e.g., soil moisture, sensible and latent heat) in the LA interactions over the past decade (2011-2020) was investigated, and the variations of these variables and their inter-relationships in the future period (2015-2099) were also analyzed based on the Coupled Model Intercomparison Project Phase 6 (CMIP6) data. The LSTNet model has demonstrated that the deep learning algorithm can effectively capture the relative importance of key variables in the LA interactions. We found it is crucial to evaluate the effect of coupled temperature and sensible heat on the LA interactions, particularly for the regions that are susceptible to concurrent droughts and heatwaves, as the co-occurrence of dry and hot weather conditions would inhibit the formation of precipitation and intensify the drought severity. Moreover, the decline in soil moisture and the rise in sensible heat under a changing climate are anticipated to further diminish precipitation in the future. This study would not only enhance our knowledge of the feedback mechanisms in the LA interactions during the drought evolution and intensification, but also provide valuable insights for further development and advancement of hydrologic models for drought monitoring and forecasting.