Predicting Playa Inundation Using a Long Short‐Term Memory Neural Network

In the Great Plains, playas are critical wetland habitats for migratory birds and a source of recharge agriculturally-important High Plains aquifer. The temporary wetlands exhibit complex hydrology, filling rapidly via local rain storms then drying through evaporation groundwater infiltration. Using long short-term memory (LSTM) neural network to account these processes, we modeled playa inundation 71,842 in from 1984-2018. At level individual playas, model achieved an F1-score 0.538 on withheld test set, displaying ability predict patterns. When averaging over all entire region, is able very closely track trends, even during periods drought. Our results demonstrate potential using LSTMs hydrological dynamics. modeling approach could be used into future under different climate scenarios better understand how will impacted by changing climate.