Description of E3.1 parameters and ranges for the perturbed experiments conducted as part of this study. The topography input is not technically a parameter, but used as an index that mapped to specific boundary condition files. Parameters are unitless unless otherweise specified.

Scatter plots of global annual mean top-of-atmosphere (TOA) absorbed shortwave radiation (swabstoa_ave) as a function of the 52 perturbed parameters and resolved topography (TOPO). X-axis values for the topography "parameter" correspond to the n-sigma value of the envelope scheme, where "0.0" corresponds to the CTRL boundary condition.

Kernel density estimation plots of simulated top-of-atmosphere (TOA) net shortwave (NET SW TOA) and net longwave (NET LW TOA) radiation and shortwave and longwave cloud radiative effect (SW CRE and LW CRE, respectively) for E3.1 PPE members grouped by the imposed topographic forcing. 

Higher pattern correlation values correspond to better agreement between 2-D spatial distribution of simulated and observed rainfall. Unlike the distributions of the calculated mean biases, the pattern correlation distributions that increasing orographic forcing only leads to continuous improvements in simulated precipitation in western North America, whereas the elevated topographic boundary conditions result in lower model skill in simulations of regional precipitation in the "Amazon" and "India" regions and globally.