Fig. 2: Multi-year averages and trends of total feature importance over a seven-day window for snowmelt, glacier meltwater, and rainfall in AMF (1960–2018) and AMFSp (1961–2018) events using LSTM-SG and LSTM-DDF models. | npj Climate and Atmospheric Science

Fig. 2: Multi-year averages and trends of total feature importance over a seven-day window for snowmelt, glacier meltwater, and rainfall in AMF (1960–2018) and AMFSp (1961–2018) events using LSTM-SG and LSTM-DDF models.

From: Shifted dominant flood drivers of an alpine glacierized catchment in the Tianshan region revealed through interpretable deep learning

Fig. 2

a–c Total feature importance for snowmelt, glacier meltwater, and rainfall for AMF events in LSTM-SG model. d–f Total feature importance for snowmelt, glacier meltwater, and rainfall for AMF events in LSTM-DDF model. g–i Total feature importance for snowmelt, glacier meltwater, and rainfall for AMFSp events in LSTM-SG model. j–l Total feature importance for snowmelt, glacier meltwater, and rainfall for AMFSp events in LSTM-DDF model. The Zc value indicates the significance of the Mann-Kendall trend test, with * at the 0.10 level, ** at the 0.05 level, and *** at the 0.01 level.

Back to article page