Hall-Beyer, MrykaDing, Junyan2012-09-132012-11-132012-09-132012Ding, J. (2012). Exploring the Relationship between Monthly Precipitation and EVI Vegetation Productivity Index of Serengeti Nation Park (Master's thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca. doi:10.11575/PRISM/26119http://hdl.handle.net/11023/206Precipitation has significant impact on the vegetation productivity in the Serengeti area of eastern Africa, an arid and semi-arid region. Previous studies indicate that the response of vegetation productivity to precipitation varies as the consequence of difference in total precipitation, soil properties, and vegetation type. In order to explore and explain the non-stationary relation between vegetation productivity and each of the mean and the variation of precipitation, I examine the correlation between vegetation and precipitation of Serengeti National Park using monthly precipitation and Moderate Resolution Imaging Spectroradiometer (MODIS) derived Enhanced Vegetation Index (EVI) images from 2001 to 2009. First, I compute the mean EVI and mean precipitation based on monthly images. To separate the mean and variation of precipitation, the mean precipitation is subtracted from each of the original precipitation images; the new series images represent the variation of precipitation. Then the new series of variation of precipitation is subjected to Fourier PCA (principal component analysis) analysis to generate a few (usually five) most representative PCA components. Among the PCA components, PC1 and PC3 are found not correlated with mean precipitation and thus are used together to represent the variation of precipitation of the entire period. To explore the non-stationary relation between EVI and precipitation, geographically weighted regression (GWR) models are used with mean precipitation and PC1 and PC3 as independent variables and mean EVI as dependent variable. Three GWR models are created with 1) mean precipitation alone, 2) PCA components, and 3) both mean precipitation and PCA components together, as independent variables. Finally, global linear least square models are used to detect how the correlations between mean EVI and precipitation are affected by soil WHC (water holding capacity) and N content, and vegetation type (independent variables). It is found that the highest correlation between mean EVI and mean precipitation occurs at north SNP (Serengeti National Park), and decreases towards both the west and east; the highest correlation between mean EVI and the variation of precipitation occurs in the north SNP and the south-east regions, and decreases towards the north and west. The correlation between mean EVI and both mean and variation of precipitation together are maximized at the west side of north SNP and decrease towards the south-east. The amount of precipitation and soil water holding capacity have a significant impact on the correlations between mean EVI and both mean precipitation and the variation of precipitation; soil nitrogen content has significant impact on the correlation between mean EVI and the variation of precipitation; only forest, woody savanna, and savanna are found to have significant impact on the correlation between mean EVI and the variation of precipitation.engUniversity of Calgary graduate students retain copyright ownership and moral rights for their thesis. You may use this material in any way that is permitted by the Copyright Act or through licensing that has been assigned to the document. For uses that are not allowable under copyright legislation or licensing, you are required to seek permission.Remote Sensingvegetation productivityprecipitationGeographically Weighted Regressionmaster thesis10.11575/PRISM/26119