Document Type : Scientific article
Authors
1
Ph.D. student of Watershed management, Faculty of Natural Resources, Urmia University, Urmia, I.R. Iran.
2
Associate Professor, Department of Rangeland and Watershed management, Faculty of Natural Resources, Urmia University, Urmia, I.R. Iran.
3
Associate Professor, Department of Forestry, Faculty of Natural Resources, Urmia University, Urmia, I.R. Iran.
4
Ph.D. student of Watershed management, Faculty of Natural Resources, Kashan University, Kashan, I.R. Iran.
Abstract
On May 31, 2003 the Scan Line Corrector (SLC) in the ETM+ instrument on Landsat 7 failed. Until now, a wide variety of gap-filling methods have been developed to recover missing pixels in the Landsat 7 SLC-off images. In present study, a newly-developed approach known as the Weighted Linear Regression (WLR) method was evaluated on the simulated SLC-off TM image acquired on 18 June 2011 when plants and vegetated lands are growing. The statistical measures of the RMSE (< 0.02), the Pearson correlation coefficient (R = 0.99), and the Nash-Sutcliffe (NSE = 0.91) showed that the WLR is highly capable of predicting missing pixels values. Based on a supervised image classification technique so-called the Maximum Likelihood (ML) applied on the predicted image, a land cover map for the desire region was generated. The accuracy assessment results consist of the Overall accuracy (OA= 89.7%), the Kappa coefficient (K= 0.85), the Allocation Disagreement (AD = 3.2), and the Quantitative Disagreement (QD = 6.9), revealed a high ability of the WLR for land cover mapping. Therefore, under the lack of Landsat TM imagery, the application of this method for recovering missing pixels is highly suggested to be useful for producing any required land cover map in Khoy region from Landsat ETM+ SLC-off images.
Keywords
)