Analysis and assessment of mountain pastures by integration of multispectral and ancillary data

The paper shows some operational techniques for the characterisation and mapping of pastoral resources in mountain areas. The utility of multispectral images was evaluated to produce thematic cartography of the studied pastures and to determine their pastoral value, a useful parameter for management purposes. Data collected by the multispectral ASTER sensor, with a 15 m spatial resolution in the VISNIR, 30 m in the SWIR, and 90 m in TIR, and the MODIS sensor, with a daily temporal step and a 250 m spatial resolution, were used. The images are classified using a maximum likelihood algorithm, while multivariate locally calibrated regression is applied to extend the pastoral value over the whole area. Results show a low capability of the satellite images in discriminating the different pasture vegetation types, whereas the assessment of pastoral value produces more promising achievements (r = 0.647;, RMSE = 6.99).