Linear and segmented linear trend detection for vegetation cover using GIMMS normalized difference vegetation index data in semiarid regions of Nigeria

Abstract

Quantitative analysis of trends in vegetation cover, especially in Kogi state, Nigeria, where agriculture plays a major role in the region's economy, is very important for detecting long-term changes in the phenological behavior of vegetation over time. This study employs the use of normalized difference vegetation index (NDVI) [global inventory modeling and mapping studies 3g (GIMMS)] data from 1983 to 2011 with detailed methodological and statistical approach for analyzing trends within the NDVI time series for four selected locations in Kogi state. Based on the results of a comprehensive study of seasonalities in the time series, the original signals are decomposed. Different linear regression models are applied and compared. In order to detect structural changes over time a detailed breakpoint analysis is performed. The quality of linear modeling is evaluated by means of statistical analyses of the residuals. Standard deviations of the regressions are between 0.015 and 0.021 with R 2 of 0.22-0.64. Segmented linear regression modeling is performed for improvement and a decreasing standard deviation of 33%-40% (0.01-0.013) and R 2 up to 0.82 are obtained. The approach used in this study demonstrates the added value of long-term time series analyses of vegetation cover for the assessment of agricultural and rural development in the Guinea savannah region of Kogi state, Nigeria. Downloaded From: http://remotesensing.spiedigitallibrary.org/ on 08/06/2015 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx Osunmadewa, Wessollek, and Karrasch: Linear and segmented linear trend detection for vegetation. Downloaded From: http://remotesensing.spiedigitallibrary.org/ on 08/06/2015 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx Osunmadewa, Wessollek, and Karrasch: Linear and segmented linear trend detection for vegetation. Downloaded From: http://remotesensing.spiedigitallibrary.org/ on 08/06/2015 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx Osunmadewa, Wessollek, and Karrasch: Linear and segmented linear trend detection for vegetation. Downloaded From: http://remotesensing.spiedigitallibrary.org/ on 08/06/2015 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx Osunmadewa, Wessollek, and Karrasch: Linear and segmented linear trend detection for vegetation.

Tensor based statistical segmentation of green vegetation canopy images

“…These formulations are defined as the Normalized Difference Index (NDI), the Excess Green Index (ExG), the Excess Red Index (ExR) and the Color Index of Vegetation Extraction (CIVE). In threshold index based methods the intensity value of each pixel in gray scale images is classified by comparing with the one or more predetermined threshold value (Osunmadewa, Wessollek, & Karrasch, 2015;Hassanein, Lari, & El-Sheimy, 2018;Kaur & Kaur, 2014).…”

Spatio-temporal monitoring of vegetation phenology in the dry sub-humid region of Nigeria using time series of AVHRR NDVI and TAMSAT datasets

“…The dry sub-humid region of Nigeria have experience noticeable vegetation cover transition due to large scale land use and land cover transformation over the last few decades [11,12]. It is evidenced from the report of Global Forest Resource Assessment that the total forested area in West Africa as of 1990 was 91,589,000 hectares while it reduces to 81 979,000 hectares and 73 234,000 hectares in year 2000 and 2010 respectively [13].…”