Assessing multi-year-drought vulnerability in dense Mediterranean-climate forests using water-balance-based indicators

• Water stress is expressed based on precipitation ( P ) minus evapotranspiration ET ). Cumulative P–ET relates to vulnerability reflected by moisture loss and mortality. Dense lower-elevation forests are more vulnerable during multi-year droughts. An LSTM model predicts (testing r 2 = 0.72) water availability (r 0.99). Historical-minimum annual portends in mountain affects vegetation response drought, which turn changes We investigated water-balance indicators assess Mediterranean-climate forest used the drought-vulnerable dense mixed-conifer of California’s Sierra Nevada, includes 78 groves giant sequoia as study area. With long-term Landsat-based data 1985–2018, water-stress patterns at 30-m resolution two historical droughts (1987–92 2012–15) were analyzed. Canopy tree mortality indices drought vulnerability. Using cumulative an indicator, that stressed 1987–92 unprecedented 2012–15 drought. indicator stress, explaining 32% 29% variances canopy mortality, respectively. As extreme test explore potential response, we trained a deep-learning Long Short-Term Memory (LSTM) project hypothetical extended-drought scenarios. The reasonably predicted with 0.72 for testing period. Annual using LSTM-based agreed 0.99) values from Landsat. Historical water-stress-prone areas projected suffer larger decreases experience more-severe 12-yr scenario. severe forests, versus mid-to-high have higher shorter growing season under current climate. Our provides water-balance-based effects disturbance warming