NDVI as a tool for measuring impact of climate variability upon vegetation.

Land-atmosphere interactive processes are useful to understand impacts of climate year by year variability and to highlight possible trends, since the status of the natural vegetation cover is strongly controlled by climate factors. The so-called NDVI (Normalized Difference Vegetation Index), derived from the red and the near infrared channels of NOAA satellite, is a reliable indicator applicable to the analysis of photosynthetic biomass variations in vegetated areas. NDVI images, derived on a monthly basis by maximum composite value technique, can become a useful tool to monitor the dynamics of vegetation and to determine the maximum level of vegetation greenness observed over every year. Year by year variability of precipitation is likely to have a significant impact on the greenness of vegetation cover, since rainy seasons are expected to stimulate a much richer plants development than drier ones. As it can be seen in figure 1, NDVI varies greatly between mid-spring, when it reaches its maximum values, and late summer, when it drops to a minimum. Moreover production of biomass due to climatic conditions can be rather different from year to year. Over such areas precipitation and water content at ARPAS stations were compared against annual maximum NDVI index from 1998 to 2008, focusing on the period November-May. Precipitation for the selected areas was measured with the network of ground stations of ARPAS. Evapotranspiration was estimated by means of Hargreaves-Samani method applied to data from the above stations. Finally, estimation of the soil moisture content was carried out by means of a daily time step simplified water balance model. Despite the low resolution of NDVI images, the maximum value of each year responds quite well to variability of precipitation and soil water content. Different NDVI responses was observed in relation to the various land cover classes of CORINE data. Because of the low image resolution and of the complex spatial patterns of vegetated area under investigation, this analysis was performed using the second level of the CORINE hierarchical classification. However the test areas (table 1) are too wide to actually have a homogeneous land cover, as a consequences, correlations (table 2) are high since they all feel the effects of the low seasonal vegetations growing between different woodlands. In order to highlight that, five secondary test areas with homogeneous land cover were examined. Considering they are mainly evergreen they respond much less to year to year variability (correlation drops to lower values). However scrub can also be affected by the specific months in which precipitation falls, therefore November-May is probably a too long period to look for correlations.