Section 2.7 -Integrating high-resolution satellite images and topographic data for rangeland resources mapping in pastoral landscapes of eastern Africa
Description
Stack Landsat images for three different dates. Pixel-wise timeseries trend analysis with PVI and downloadable R- script. Large slope coefficients of linear regression analysis indicate highly deciduous surfaces with highly dry-out dynamics and land surfaces that stay greener for most part of the season show slow dry-out dynamics.
Files
Steps to reproduce
data images comprise Perpendicular vegetation indices (PVI) for three different dates that were stacked. The first PVI in the stack images is the one closest to the cumulative highest rainfall in a season and therefore represents when vegetation was very healthy. Other two PVI represents vegetation in senescence stages and therefore show brown vegetation. Pixel-wise linear regression model was fitted to the data images and its slope coefficient was extracted using downloadable R- script deposited at the GitHub https://github.com/mohamedshibia/Pastoral-pastures-descriptors . Slope coefficient of linear regression model was used as a proxy measure of vegetation dry-out dynamics. Large negative values indicate fast drying surface vegetation, such as grasses and highly deciduous bushes. Slow dry out dynamic is represent by smaller slope values. Slope metrics were to be combined with the first PVI to interpret individual vegetation types.