College of Agriculture & Natural Resources

Permanent URI for this collection

https://ir.knust.edu.gh/handle/123456789/13

Browse

Browsing College of Agriculture & Natural Resources by Subject "Cover and management (C) factor,"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    Estimation of USLE’s C-factor using vegetation indices (VIs) for soil erosion modelling in Lake Bosumtwi Basin, Ghana.
    (2012-12-02) Loh, Seyram Kofi
    Water erosion is one of the most challenging environmental problems in Bosumtwi basin; this is due particularly to the rough terrain and human activities on the landscape. The Universal Soil Loss Equation (USLE) is a commonly used model in estimating soil loss by water. The USLE’s cover and management factor (also known as C - factor) represents the combined effects of plant, soil cover and management on erosion. Therefore, this variable keeps changing as soil surface cover keeps changing. The use of remote sensing vegetation index to estimate C - factor proved to be reliable, consistent and useful when considering inaccessible and large regions. The normalized difference vegetation index (NDVI), is the most used algorithm in estimating the C - factor despite the availability of other indices believed to overcome some inherent problems associated with the use of NDVI. This study compares the NDVI to the EVI, in their capability to map land cover types, to be applied in C-factor estimation. Plots of the divergence statistics (derived from ISODATA clustering algorithm in ERDAS Imagine) against class numbers helped determine the optimal numbers the Hypertemporal vegetation index’s (VI’s) images can be classified into. The classification map outputs were reclassified based on the similarities of the profiles generated from the mean annual VI values. The resultant map of the supervised classification was compared to already classified map (produced from a single-date ASTER and data collected from the field) for legends assigning and validation. The overall accuracies of the two VI maps are both high (80 % or EVI and 70% for NDVI) and the kappa statistics of 0.7 and 0.5 for NDVI and EVI respectively. Forty-six (46) 2010 Hypertemporal MODIS EVI were applied in developing an annual C-factor model. In validating the model output; a C-map developed by traditional method – based on EVI classified map and literature assigned C – values - was compared to the C-factor model developed using Exponential Function (E. F) approach. The C-factor model showed an overall accuracy of 76% and 0.6 kappa value.