2.3 Sensitivity analysis for GALDIT parameters
The sensitivity analysis of GALDIT parameters was used to verify the consistency of the analysis results in the process of the vulnerability assessment (Saidi et al., 2013). To minimize the error and uncertainty of the vulnerability index, Babiker et al. (2005) and Saidi et al. (2011) adopted and evaluated appropriate parameters by performing two methods of sensitivity analysis, that is, the single parameter sensitivity analysis developed by Napolitano and Fabbri (1996) and the map removal sensitivity analysis developed by Lodwik et al. (1990). Since the development of the theoretical weights of GALDIT parameters in 2001, the modification of ranges and weights of vulnerability parameters of the GALDIT analysis method was continuously attempted (Kazakis et al., 2018; Gorgij and Moghaddam, 2016). Among these attempts, the sensitivity analysis of the parameters helps determine the weights effectively by comparing with theoretical weights in assessing the vulnerability with the score and weight of each index (Gontara et al., 2016; Mahrez et al., 2018). The existing GALDIT has determined parameters using averages for a long time. When calculating the monthly SWI vulnerability index, the error and uncertainty of the weights of the parameters must be minimized.
The single-parameter sensitivity analysis compares the “effective” or “real” weight with the theoretical weight for the area in which the index is assessed. The effective weight W (%) for the assessment area of the GIS is calculated as follows (Babiker et al., 2005):