Monitoring vegetation phenological stages using remote sensing data for pasture management in arid and semi-Arid lands of Kenya

The objective of this study was to monitor grassland phenological stages in selected arid and semi-arid lands (ASALs) of Kenya, so as to provide information useful in pasture management. Five years of Normalized Difference Vegetation Index (NDVI) data from the VEGETATION instrument onboard SPOT were extracted over three ASAL districts. Extraction points were based on a land cover map that showed the location of grassland in the three districts. Piecewise logistic functions were applied on the NDVI data in order to identify phenological stages. RFE data were then used to relate the identified stages to rainfall using lagged correlation. Curves of the correlated lagged rainfall and NDVI from the determined phenological stages were plotted to compare their temporal patterns. Spatial patterns of length of the growth period were also assessed. Interannual phenological stages appeared to follow a clear growth – senescence temporal pattern. Two growth periods were identified in all the districts, consistent with known cycles of different grass and browse species in the areas. Peak growth was seen to occur during the short rains in Kajiado district and during the long rains in Baringo district. Growth in the two seasons was almost the same in Garissa district. Phenological stages were significantly correlated to different lags of rainfall, with response to a longer lag observed during the March to June growth period. Patterns of lagged rainfall were also found to be similar to those of NDVI at the different stages. The length of both growth periods showed spatially coherent patterns that signified the distribution of different pasture species. Given these results, logistic functions were able to model grassland phenological stages in the ASALs.