Monitoring and Assessment of Rangeland in Semi Arid Areas of Sudan Using Remote Sensing Techniques North Kordofan State

Abstract

The study was conducted in North Kordofan State with the main objective was to investigate differences in rangeland sites and the use of remote sensing techniques for assessing and monitoring rangelands in North Kordofan State. Three rangeland sites were identified in the study area as gardud, sand sheet and sand dune. At each site a sampling area of 5 km x 5 km was marked and four transects of 500 m each were located in each site. Within each site 200 quadrates (I m x1m) were located systematically at 10 m intervals resulted in 50 quadrates per transect to assess vegetation measurements mainly plant cover, biomass production and plant density. The data collection has been carried out during the years 2010 and 2011. Ten soil samples were taken and analyzed for physical attributes and organic matter in each site to investigate .differences in soil types within the sites Remote sensing techniques were used to assess rangeland where MODIS/TERRA surface reflectance (MOD09Q1) at 250 m spatial resolution and MOD09A1 at 500 m spatial resolution was used for seven years from 2005 to 2011 in October and in the middle of February and April. Two vegetation indices were used, Normalized Difference Vegetation Index (NDVI) and the Soil-Adjusted Vegetation Index (SAVI). They were computed from MODIS imagery using ERDAS imagine and ARC/GIS software. The indices were compared for their values in assessing status of rangeland in the selected rangeland sites. Principal Component Analysis (PCA) was performed using TNT MIPS software using selected types of data (two vegetation indices, topographic data and vegetation surface reflectance within the three bands of MODIS data) to 4 reduce redundancy in the data set. A new multiband image were created for two periods 2005-2011 including the original image and the most useful spectral information to run image processing and automated classification. Color composite and unsupervised classifications were applied using the data components that showed high contribution in the .variance value showed by PCA According to the results there were remarkable differences between the three range sites with respect to plant cover and productivity, attributed to site characteristics, mainly soil type which dictated vegetation characteristics as affected by soil. The study also showed differences between the values of NDVI and SAVI along the three range sites which reflected the effect of the low plant cover and consequently soil reflectance. These differences are important assessment indicators for the variation in vegetation characteristics, while the variation among the sites was attributed to the differences in soil characteristics. The use of MODIS band 7 enabled to differentiate between gardud range site at one side and sand sheets and sand dune range sites at another side. This is mainly due to the remarkable difference in the soil characteristics of these two sites. The results also showed that the use of the principal component analysis (PCA) with the selected variables showed high difference, reflected in variance and eigen-values and hence can be used for differentiation. Application of color composite map for the three main variable of the PCA enabled differentiation between the three sites, making use of differentiation caused by the variation caused by soil and vegetation jointly. The use of these three variables to make a color composite image for two periods 2005/2011 enabled differentiation between three sites as major units for monitoring, while the use of unsupervised classification with these three variables cross checked with 5 ground points enabled identification of main vegetation groups within the .sites, which is a more refined reference for monitoring The study recommended that the use of MODIS data for rangeland monitoring provide that this data is available having wide coverage and .good temporal resolution 6