Meteorological data
10 years meteorological records of Ganges River basin were gathered from different organizations of respective countries. From collected meteorological data, average monthly temperature and average yearly rainfall were calculated for each meteorological station. These data were interpolated incorporating elevation, representing iso heightal lines. These lines were further interpolated to 30 arc sec cells. Similarly, temperature data were interpolated and converted to 30 arc sec cell data.
El
Fig 1: methodology of analysis
Kira's humidity index and Kira's warmth index were calculated using monthly average precipitation and monthly average rainfall data(Chunying, 1999). These indices are given by following equations.
Humidity Index,
Warmth Index,
P: yearly average precipitation(mm)
Tm : average monthly temperature(C)
Results and Discussion
Studying multi-temporal spectral patterns of each of the 25 clustes, 4 preliminary classes were defined. The classes were (i) snow & ice, (ii) bare land and semi bare land, (iii) crop land and (iv) forest.
Among the 25 clusters, 3 classes were found to be generated with pixels arising from scan error during the data acquisition but the total number of pixels in these classes were less .06% of total pixels. These clasess were excluded from further analysis. The remaining 22 classes were identified as forest(3 classes), semibare land (5 classes), bare land (2 classes), snow &ice (1) and crop land (11 classes). The following figure gives monthly variation NDVI of preliminary classes except for three months June, July, and September.
Fig 2: Spectral patterns of different vegetation classes based on Iso-data clustering
The Classes were defined using classification scheme developed by International Geographic Information System Examination Committee for Global Mapping to have conformity with other studies. 11 preliminary classes of crop land in the study area were regrouped into two classes, single crop and double crop based on temporal variance of NDVI.
According to Trwartha&Horn, 1980, approximate altitudinal limit of alpine vegetation on a tropical mountain is above 4000 m elevation. This altitudinal information was used in reclassifying two bare land (2 classes) semi bare land classes (5 classes) into two alpine bare/semi bare land classes and two bare/semi bare land classes.
