Sensitivity Analysis and Uncertain quantification for the land model CLM4

In this study, we introduce an uncertainty quantification (UQ) software framework for sensitivity analyses and uncertainty quantification on selected parameters of the Community Land Model (CLM) that are related to soil hydrology, surface and subsurface runoff. Simulations are performed at selected watersheds and flux tower sites spanning a range of hydrologic and climatic regimes in the U.S., with the focus of studying the effect of soil and run-off generation on CLM simulated hydrologic cycle. The efforts include identification of significant CLM model parameters, determination of their baseline values and physical ranges, and simulations for each combination of sampled parameters. To deal with the computational difficulties and to explore the multi-dimensional parameter space, we first employ effective sensitivity analysis methods to study the CLM model sensitivity. Secondly we use efficient sampling approaches to reduce the number of forward calculations while exploring the parameter space and quantify the input uncertainty. We will demonstrate the proposed sensitivity analysis and uncertainty quantification framework comparing model sensitivity in two different hydrologic and climatic regimes.