Revision as of 14:26, 22 September 2006; view current revision
←Older revision | Newer revision→
[edit]8.3. The use of Normalised Difference Vegetation Index (NDVI) images.
During periods of drought conditions, physiological changes within vegetation may become apparent. Satellite sensors are capable of discerning many such changes through spectral radiance measurements and manipulation of this information into vegetation indices, which are sensitive to the rate of plant growth as well as to the amount of growth. Such indices are also sensitive to the changes in vegetation affected by moisture stress. The visible and near infra-red (IR) bands on the satellite multispectral sensors allow monitoring of the greenness of vegetation. Stressed vegetation is less reflective in the near IR channel than non-stressed vegetation and also absorbs less energy in the visible band. Thus the discrimination between moisture stressed and normal crops in these wavelengths is most suitable for monitoring the impact of drought on vegetation. Aridity anomaly reports used by IMD do not indicate arid regions. They give an indication of the moisture stress in any region on the time scale of one or two weeks, and they are useful early warning indicators of agricultural drought (Das, 2000). The Normalized Difference Vegetation Index (NDVI) is defined by them as: NIR VIS NDVI NIR VIS + = - where NIR and VIS are measured radiation in near infra-red and visible (chlorophyll absorption) bands. The NDVI varies with the magnitude of green foliage (green leaf area index, green biomass, or percentage green foliage ground cover) brought about by phenological changes or environmental stresses. The temporal pattern of NDVI is useful in diagnosing vegetation conditions. The index is more positive the more dense and green the plant canopy, with NDVI values typically 0.1 – 0.6. Rock and bare ground have NDVI near zero, and clouds, water and snow have NDVI less than zero. Moisture stress in vegetation, resulting from, prolonged rainfall deficiency, is reflected by lower NDVI values. Such a decrease could also be caused by other stresses, such as pest/disease infestation, nutrient deficiency, or soil geochemical effects. Discrimination of moisture stress from other effects does not present a problem in coarse resolution data over large areal units, as neither pest/disease attack nor nutrient stress is selective in terms of area or crop type. Finally three more indices characterizing moisture (VCI), thermal (TCI) and vegetation health (VT) conditions were constructed following the principle of comparing a particular year’s NDVI and Brightness Temperature (BT) with the entire range of variation during extreme (favourable/unfavourable) conditions. Since the NDVI and BT interpret 26 extreme weather events in an opposite manner (for example, in case of drought, the NDVI is low and BT is high; conversely, in a non drought year, the NDVI is high while the BT is low), the expression for TCI was modified to reflect this opposite response of vegetation to temperature. The VCI and TCI were found as: ( ) 100 ( ) max min min NDVI NDVI VCI NDVI NDVI - = ´ - ( ) 100 ( ) max min max BT BT TCI BT BT - = ´ - where NDVI, NDVImax and NDVImin are the smoothed weekly NDVI, its multi-year absolute maximum, and minimum, respectively; BT, BTmax, and BTmin are similar values for BT. The VCI and TCI approximate the weather component in NDVI and BT values. They change from 0 to 100, reflecting variation in vegetation conditions from extremely poor to optimal. In drought years leading to yield reduction, VCI and TCI values drop below 35 (Kogan, 1997). This level was accepted as a criterion for drought detection. The VCI and TCI were also combined in one index (VT) to express their additive approximation of vegetation stress, as shown by equation 2 VT = (VCI + TCI ) With the development of the validation data set, some weights will be assigned to the
VCI and TCI indices.