A sensitivity analysis of the SimSphere SVAT model in the context of EO-based operational products development

Authors Organisations
Type Article
Original languageEnglish
Pages (from-to)166-179
Number of pages14
JournalEnvironmental Modelling and Software
Volume49
DOI
Publication statusPublished - Nov 2013
Links
Permanent link
View graph of relations
Citation formats

Abstract

Use of simulation process models often combined with Earth Observation (EO) data, has played a key role in extending our abilities to study land surface interaction processes and enhancing our understanding of how different components of the Earth system interplay. Use of these synergistic techniques aims to improve the estimates of key parameters characterising land surface interactions by combining the horizontal coverage and spectral resolution of remote sensing data with the vertical coverage and fine temporal continuity of simulation process models.

This study performs a Global Sensitivity Analysis (GSA) on the SimSphere land surface model aiming to further extend our understanding of the model structure and establish its coherence. It builds on previous works conducted on the model to which a sophisticated and cutting edge GSA meta-modelling method adopting Bayesian theory is implemented. Our first objective is to examine the effect of assuming uniform probability distribution function (PDFs) for the model inputs/outputs on the sensitivity of key quantities simulated by SimSphere. A further objective is to explore the sensitivity of new, previously unexplored variables simulated by the model, namely of the Daily Evaporative, Non-Evaporative Fractions and Radiometric Temperature.

The GSA conducted assuming uniform PDFs showed comparable results to previous studies in terms of identifying the most sensitive model inputs to each of the outputs considered. Yet, in absolute terms, the statistical parameters measuring the sensitivity of the model inputs were notably different. SA on the newly examined model outputs showed largely explainable results and allowed identification of the most responsive model inputs and interactions. In general, our results provided further evidence supporting the model coherence and correspondence to the behaviour of a natural system. The implications of the main findings are discussed in the framework of the model use either as a stand-alone tool or synergistically with EO data, particularly so towards the operational development of such products.

Keywords

  • energy fluxes, soil moisture, SimSphere, triangle, earth observation, sensitivity analysis, BACCO GEM-SA, Gaussian process emulators