An Evaluation of the Capability of IRS-P6 Data for Monitoring Pollarding Forest Areas of Northern Zagros (Case Study: Kurdistan, Pollarded Forests of Baneh)

Abstract

To evaluate the capability of IRS-P6 LISS-III data to be employed for monitoring the pollarding forest areas in Northern Zagros, some parts of pollarded forests located around Baneh city were selected as a case study area. The pollarding area was determined as the ground truth in a 3-year alternation period using a global positioning system (GPS). Radiometric and geometric corrections were applied to the image and then the data pre-processed, using 2 methods of Spectral Rationing and Principal Component Analysis (PCA). Likewise, multi-spectral bands were fused with IRS-1C PAN image, using the Intensity–Hue–Saturation Method (IHS). The obtained results were combined with the original bands. The separability of classes was studied using Bhuttacharrya Distance Criteria. The resulting data was classified using Maximum Likelihood Algorithm. Then the classified image was compared with ground truth on a pixel by pixel basis. In order to determine the classification accuracy, four parameters encompassing Overall Accuracy, Kappa Coefficient, Producer Accuracy, and User Accuracy were used. The results showed that most of the classes were completely separated from Northern Koor class. The highest overall accuracy was 70% and a Kappa Coefficient of 60% obtained through a five-class classification of the bands combination PCA (4, 2, 3) -1, 4, 1. In this classification the resulted User accuracy and Producer accuracy were more than 50% for all classes expect for southern Khert. Results of the study revealed the high capability of the abovementioned image and methods to separate the pollarding areas and to prepare the map of the area.