The usefulness of coarse resolution satellite sensor data for identification of biomes in Kenya

Sammanfattning: Vegetation phenology (seasonal rhythm) is closely related to seasonal dynamics in the lower atmosphere and is therefore an important element in global climatic models and vegetation monitoring. Remote sensing is the primary means by which we can observe the dynamic characters of the earth's biosphere. Description and mapping of the natural vegetation, and its interpretation in terms of soil, climatic and other information, are important in land use planning, particularly in areas where detailed information on soil and climate is insufficient, as is often the case in East Africa. The aim of this study is to investigate the usefulness of coarse resolution (1 km) data from the NOAA AVHRR sensor for the identification and separation of principal vegetation communities (biomes) and to define seasonal rhythms for these communities. The Advanced Very High Resolution Radiometer (AVHRR) sensor is carried by a series of meteorological satellites operated by the National Oceanic and Atmospheric Administration (NOAA). The primary advantage of the AVHRR is the frequent temporal (daily) coverage over large areas, which allows a better opportunity to obtain cloud free coverage during important phenological stages. Normalised Difference Vegetation Index (NDVI) data derived from AVHRR offer a means of evaluating phenological characteristics, such as flowering and senescence. A total of 53 AVHRR NDVI composites covering Kenya between 1 April, 1992 and 21 September, 1993, were obtained from National Aeronautics and Space Administration (NASA). The results show that some essential biomes such as grassland and dry forest were possible to identify and separate. Phenological characteristics, such as beginning and end of growing season were only possible to extract for grassland and dry forest. Even though cloud detection and interpolation methods were applied on the composites, the influence of clouds degraded the image quality and it was therefore difficult to interpret the NDVI profiles. This study indicates the potential usefulness of 1-km NOAA data for monitoring vegetation phenological cycles and has demonstrated that atmospheric effects (e.g. clouds) must be understood quantitatively for specific inventory purposes.