Sensitivity analysis on spatial models: a new approach

Linda Lilburne 1, Debora Gatelli 2 and Stefano Tarantola 2
1 Landcare Research
Private Bag 69, Lincoln, Canterbury 8152, New Zealand
Tel.: +64 3 325 6700; Fax +64 3 325 2418
lilburnel@landcareresearch.co.nz
2 Econometric and Applied Statistics Unit, Joint Research Centre of the European Commission
Via E. Fermi 1, 21020 – Ispra,  Italy
Tel.: +39 0332 789 928; Fax: +39 0332 785 733
debora.gatelli@jrc.it; stefano.tarantola@jrc.it

Abstract
Sensitivity analysis involves determining the contribution of individual input factors to uncertainty in model predictions. The most commonly used approach when doing a sensitivity analysis on spatial models is using Monte Carlo simulation. There are a number of techniques for calculating sensitivity indices from the Monte Carlo simulations, some more effective or efficient than others. These techniques are summarised along with their limitations. A new technique for undertaking a spatial sensitivity analysis based on the Sobol' method is proposed and tested. This method is global, variance-based, and model-free. The technique is illustrated with two simple test models.

Keywords: sensitivity analysis, uncertainty analysis, simulation, Monte Carlo

In: Caetano, M. and Painho, M. (eds). Proceedings of the 7th International Symposium on Spatial Accuracy Assessment in Natural Resources and Environmental Sciences, 5 – 7 July 2006, Lisboa, Instituto Geográfico Português

AttachmentSize
Lilburne2006accuracy.pdf686.59 KB