Text S2 Observational and remote-sensing derived dataset

We adopted gridded daily temperature from both the Berkeley Earth Surface Temperature Gridded Land Daily Temperature dataset (hereafter Berkeley Earth, http://berkeleyearth.org/data/) and the Climate Prediction Center Global Daily Temperature (hereafter CPC, https://psl.noaa.gov/data/gridded/data.cpc.globaltemp.html). The Berkeley Earth Gridded Land Daily Temperature dataset includes daily maximum and minimum temperature at a 1° × 1° latitude-longitude grid from 1880 to 2021. Uncertainties in gridded temperature records at a hemispheric scale are greater prior to 1950 due to decreasing data availability and spatial coverage of weather stations, so we calculated thermal-based indices from 1950 to 2014 to be consistent with the historical simulations. The CPC temperatures are 0.5° × 0.5° latitude-longitude resolution daily data calculated using the CPC archive of Global Telecommunication System daily reports and the Shepard algorithm. CPC covers 1979-2021 and we used 1979-2014 to compare against historical simulations.
To assess simulated LAI amplitude and seasonality, we also adopted the Global LAnd Surface Satellite LAI. This product provides eight-day composites of LAI, based on the Advanced Very High Resolution Radiometer (AVHRR) reflectance data. We obtained the 0.5° x 0.5° dataset and aggregated the raw GLASS LAI to a 1° x 1° latitude/longitude grid by averaging the four GLASS grid points in each model grid. We then fitted a smoothing spline to the time series of each grid point for each year and used it to interpolate the 8-day data to daily resolution. Finally, we calculated the day-of-year (DOY) time series of spring onset as indicated by different LAI thresholds.