The relation was obtained comparing several NDVI profiles derived from Landsat 7 ETM + imagery with in situ data relative to (semi-) natural vegetation (de Jong et al., 1998). C-factor values range between 0 and 1 ( Fig. 5). The following steps were performed in order to represent CD 2665 space-variant and time-variant vegetation change with better accuracy:1)Water body masking. This step was carried out by masking the area characterised by negative values of NDVI in both seasons.2)Pixel-based NDVI calculation using the following expression:equation(14)NDVI=max(NDVIwNDVIs).NDVI=maxNDVIwNDVIs.The use of the NDVI maximum is reasonable considering that both images show several areas characterised by a negative value of NDVI. For the winter image this is due to the shadowing effect of orography. Negative NDVI pixels in the summer image are the consequence of a significant atmospherically absorption due to gasses that affect the NIR band.3)Estimation of C corresponding to NDVI using Eq. (13). De Jong\'s equation gives negative values of C for NDVI higher than 0.534. The areas with negative values of C correspond to forest land cover. According to Borrelli et al. (2013), for these pixels a C-factor equal to 0.003 was assigned.