Italian Experiments

Influence of Topography on Vegetation Indexes of Forested Ecosystems

rondonia
Courtesy of NASA/JPL

CRAST - Centro Ricerca Analisi Spaziale e Telerilevamento
FacoltÓ di Agraria, UniversitÓ Cattolica del Sacro Cuore
Research group coorinator:  Dott. Agr. Massimo Vincini
row
crast Landscape morphology has a direct influence on water movements, soil physical and chemical properties and on the productivity of the vegetation cover. Many studies have demonstrated the relationship between landforms and productivity of phytocoenosis or cultures and it should be expected that this relationship can be proved for remote sensing vegetation indexes.

Remote sensing of rugged terrain is hampered by topographic effects on spectral signatures. Hence, the effect of topography on backwards radiance should be taken into account before considering the influence of topography on vegetation. The "topographic effect" is easily observed in this composite BGR of Landsat TM bands 3, 4 and 5 of the Valboreca site (one of the two sites of the project in the northern Apennines): slopes facing the sun are brighter whereas slopes facing the opposite direction are darker as a consequence of the different fluxes of incident sunlight over a given area.

Fig. 1 Fig. 2

TM data topographic dependence can be evaluated by calculating the correlation coefficients between the reflectance of a given band and the cosine of solar illumination angle (i) - the angle regulating the topographic effect, between the solar vector and the vector normal to the surface. The Figure shows a correlation coefficient’s value of 0.76 between TM band 4 and the cosine of solar illumination angle for the mixed deciduous forest area of the Valboreca site, thus indicating that about 58% of the variance of NIR reflectance in this rugged forested area is explained by the topographic effect. Several proposed algorithms use Digital Elevation Models (DEMs) as the basis for topographic correction. The following Figure, reporting the same data after the application of the non-lambertian Minnaert model shows the effectiveness of the model at removing the topographic effect:

Fig. 3 Fig. 4

In spite of the reported effectiveness topographic normalization methods are used infrequently, probably as a consequence of the lack of high-resolution DEMs. This constraint may be removed upon completion of the Shuttle Radar Topography Mission that will provide a huge SIR-C/X-SAR interferometric DEM data-set.

The synergistic use of optical remote sensing data and Digital Elevation Models at the same spatial resolution can be extremely helpful for the evaluation of the influence of topography on productivity and therefore on vegetation indexes of forested ecosystems. A better comprehension of this relationship would benefit the inventory of forest resources as well as the evaluation of potential land productivity in areas of relief and of limited cartographic coverage. The project proposed by CRAST (Centro Ricerca Analisi Spaziale e Telerilevamento- UniversitÓ Cattolica del sacro Cuore) will use SRTM DEM data for the Minnaert topographic normalization of Landsat TM data of two forested areas in the northern Apennines (mixed deciduous forest and beech forest). Forest cover information such as structural/floristic composition will be collected for the beech test area by GPS survey. NDVI and other vegetation indexes derived from the corrected TM data, will be used as dependent variables in multiple linear regressions against landscape morphology parameters, obtained from the SRTM DEM, that can affect vegetation productivity. Among these parameters, elevation, slope, landsurface curvature and catchment area will be used as independent variables.