Extended Data Fig. 1: Climate-change detection using the ridge regression method in the CESM2 LE.
From: Anthropogenic fingerprints in daily precipitation revealed by deep learning
a, Time series of the simulated AGMT from 1850 to 2100 in the CESM2 LE (black line) and the annual average of the estimated daily AGMT by prescribing 2 m temperature (T2m, green line), 2 m specific humidity (SH2m, orange line) and precipitation (PRCP, blue line) in the ridge regression model24. Each dot denotes the estimated AGMT using daily input. The green, orange and blue bars on the right denote one standard deviation of estimated daily AGMT using T2m, SH2m and PRCP during the historical period (that is, 1850–1950), respectively. The black error bars denote the 2.5th–97.5th percentiles of the daily estimated AGMT in 1850–1950. b, Time series of the ratio of the annually averaged AGMT to the AGMT of the upper limit of test statistics (that is, 97.5th percentile of the daily estimated AGMT in 1850–1950). The first year that the ratio exceeds 1 for each case is indicated.
