Impact of Climate Changes on Vegetation: a time series trend analysis

The intra-annual and inter-annual dynamics of development of the vegetation cover are tightly linked to water availability and influenced - in the short period - by the water input produced by rainfalls and in the long period by the reserves cumulated in the aquifers.

The modelling of the vegetation dynamic's’ trend is therefore an indicator of water availability that, if assessed in a relatively long time series, gives the estimation of the climatic changes impact on ecosystems.

The assessment of the status of vegetation can be carried out through direct interventions in field and by means of remotely sensed images. Among the various satellites providing a continuous flow of information in space and time, the NOAA-AVHRR sensor has represented a benchmark for the studies concerning vegetation monitoring carried out by means of NDVI (Normalized Difference Vegetation Index).

This index provides indirect indications on the vegetation condition such as the Leaf Area Index (LAI) and the Fraction of Photosynthetically Active Radiation (FPAR), (Prince 1991, Los 1998).

The decadal frequency of images, added to the data available in the twenty-years-old archive make NOAA-AVHRR imagery a precious tool for the evaluation of the status and development of the phenomena concerning a certain area. The level of resolution (8x8 km) is, moreover, appropriate for the implementation of analyses at the regional scale.

Since the ’80s IBIMET-CNR researches have been developing innovative techniques and methodologies for the analysis of remote-sensing imagery, with application at different scales - local to regional - in the framework of a number of programmes concerning food security and natural resources management.

In particular, in the framework of the current research programmes on the Sahelian and Mediterranean areas, regarding the monitoring of the processes of desertification and development of products aimed at preventing and managing food crises, a study was carried out aimed at targeting and defining the vegetation index trends, in order to identify the areas characterised by remarkable dynamics of degradation of the vegetation cover (hot spots).

This result is the starting point for the implementation of further studies aimed at identifying the causes and determinant elements of the evolution process affecting the environmental components, in order to evaluate the vulnerability context of the different systems, as a support for the management of emergency and definition of specific planning actions in the medium and long period.

After the first assessment of results, the Sahelian region shows a prominent increase in vegetation values in the course of the historical series. In particular, this increase is situated in the land stripe that is characterised as Semi-Arid from the climatic point of view (specifically at the border between Mali and Mauritania, north of Burkina Faso, at the Mali border, and in the agro-pastoral band of Chad), which, moreover, represents the limit of transition between the agricultural and the pastoral parts of the region. Some other remarkable trends have been observed in the Ferlo (Senegal) pastoral region, in the Azawagh area, and in the Keita and Bouza areas in Niger.

The analysis has also pointed out some “hot spot” areas characterised by a decrease in the vegetation response. These areas, which are indeed extremely limited in terms of surface with respect to the context of the analysis, are situated in particular in the Segou (Mali) region and in central/northern Nigeria.

As regards the Mediterranean region a generally positive vegetation trend is recorded, with some exceptions in plain areas such as the Guadalquivir area in Spain, the plain of the Po river, and the plains of southern France.

Methodology

· Gathering of NOAA-AVHRR NDVI decadal data - series 1982-2000
· Image geo-referencing,
· Application of procedures for the elimination of outliers and smoothing of profiles (de-clouding),
· Statistical analysis and Pearson correlation maps for each decade and related significancy,
· Statistical analysis and linear regression maps for each decade,
· Synthesis maps indicating, for each decade and for each 8X8Km grid point, the type of growth (if positive or negative) and intensity of the phenomenon,
· Maps of NDVI maximum value decade shifting .