Analysis time series of vegetation indices using Landsat-8 satellite imagery

This study aims to apply time series analysis vegetation indices using Landsat 8 satellite to separate evergreen and seasonal plants of the Abu-Ghraib area and assess the NDVI, and NDVSI indces to detect vegetation. The proposed classifying method is based on determining the temporal score of the vegetation index and then applying the minimum and maximum criteria for the classification decision. The scenes of the area for different seasons over 4 years were classified using vegetation indices to classify the land cover into two classes vegetation or non-vegetation areas. Two kinds of vegetation indices were applied to make the final classification using a k-NN algorithm to decide the region into three classes (evergreen, seasonal plants, or non-vegetation). The confusion matrix was used to calculate the metrics (MSE & MAE) and assess the accuracy of the classification. The results showed that the k-NN algorithm is capable to classify satellite data into three vegetation indicators with high accuracy when determining the optimum threshold of vegetation index. The NDVSI index can be considered as a better indicator for studying vegetation cover in dry and humid areas and areas with low vegetation cover.