Fig 6.

Nash-Sutcliffe modeling efficiency comparison between the proposed LSTM-based models and the other model set-ups for (a) deciduous and (b) evergreen forests. Nash-Sutcliffe modeling efficiency values have been calculated based on the mean ensemble ±sd of the 50 model runs. LSTM = LSTM model using the full depth of the Landsat time series and climate data; LSTM_perm = LSTM model but the temporal patterns of both the predictive and the target variables were randomly permuted while instantaneous relationships between predictive and target variables were kept; LSTM_msc = LSTM model but the Landsat time series for each band were replaced by their mean seasonal cycle, while using the actual values of air temperature (T_air), precipitation (P), global radiation (Rg), and vapor pressure deficit (VPD); LSTM_annual = LSTM model but the Landsat time series for each band were replaced by their annual mean, while using the actual values of T_air, P, Rg, and VPD, RF = Random Forest model using the actual values of the Landsat time series and climate data.