Sensitivity Exploration of SimSphere Land Surface Model Towards Its Use for Operational Products Development from Earth Observation Data

The use of Earth Observation (EO) data combined with land surface process models is at present being explored to assist in better understanding the natural processes of the Earth as well as how the different components of the Earth system interplay. However, before applying any modelling approach in performing any kind of analysis or operation, a variety of validatory tests needs to be executed to evaluate the adequacy of the developed “model” in terms of its ability to reproduce the desired mechanisms with the necessary reality. Sensitivity analysis (SA) is an integral and important validatory check of a computer simulation model or modelling approach before it is used in performing any kind of analysis operation. The present study builds on previous works conducted by the authors in which a sophisticated, cutting edge SA method adopting Bayesian theory has been implemented on a land surface process model called SimSphere with the aim of further extending our understanding of its structure and of establishing its coherence. This land surface model has been widely used as an educational tool in different Universities across the world, as a stand-alone tool and synergistically with EO data in deriving key parameters characterising land surface processes. SimSphere use is currently under investigation by two Space Agencies for deriving spatio-temporal estimates of energy fluxes and soil surface moisture from a technique in which the model is used synergistically with Earth Observation data. The GSA method employed here provided a further insight into the model’s architectural structure, and allowed us to determine which model input parameters and parameter interactions exert a significant influence on the selected model outputs, and which are inconsequential. Analysis of the SA results indicated that only a small fraction of the model input parameters have an appreciable influence on the examined target quantities. Results, however, did suggest the presence of highly complex interactions structure within SimSphere, which drove a considerable fraction of the variance of the variables simulated by the model. The main findings are discussed in the context of the future model use including its synergy with EO data for deriving the operational development of key land surface parameters from space.