Evaluation of the best remotely-sensed based indices to detect vegetation cover in Khalkhal County

قربانی, اردوان and میرزایی موسی وند, امیر and اسمعلی, اباذر (2010) Evaluation of the best remotely-sensed based indices to detect vegetation cover in Khalkhal County. Masters thesis, University of Mohaghegh Ardabili.

[img] Text
MAINREPORT_1390-2-25_final_NDVI_Pazhohesh.pdf

Download (1MB)
Official URL: http://uma.ac.ir/

Abstract

This study was conducted to evaluate and calibrate about 30 remotely sensed based indices in the Khalkhal County using correlation analysis and full caliabration process for the NDVI to use these indices for vegetation / land cover mapping. By considering seasonality and phenological patterns three sensor images including 1) Landsat TM 1987/09/30, 2) Landsat ETM+ 2002/07/29 and 3) IRS P6LISSIII 2008/05/26 were selected. Image preprocessing including geometric and radiometric correction and topographic normalization were conducted before image interpretation. The NDVI map was derived and correlated with field collected data and the accuracy of the map was assessed. In the second stage, 30 indices were analysed using correlation analyses. Finally, field collected data were classified and analysed versus Indian sensor as the same procedure of the second stage. Results show that, although there is evidiance that, the NDVI is performing logically, however the valus of the produced map is far from the real field situation. Moreover, the results of accuracy ssessment for the NDVI show no relaiable results. Results of correlation analysis for individual land covers show also no significant relationships. To sum up, using the NDVI in rutin maner in the Khalkhal County and the same areas has no relaiable and robust results in vegetation mapping. The results of correlation analysis for 30 indices using correlation analysis show that, totally 11 indecies such as NDVI, SAVI, TVI, MSI, MND, PD321, PD322, PD312, IPVI, IRON OXIDE and MIR show no significant relationships with different landcovers. Ten indices including TNDVI, RDI, PD311, NIR, RA, IR1, MSR, MINI, IR2 and MIRV1only show significant relationships with one land cover and one sensor. Finally, indices such as DVI, VNIR1, MIRV2, VI3, PVI, VI2, VI1, VNIR2 and NLI show significant relationships with more than one land cover and sensor. By considering the results of this section the VNIR2 and NLI are recognized as the best indices for the study area. The results of a pilot study by classified land covers were no considerable. Results of the three used sensors showed that the IRS is the best sensor, TM is in the second stage and ETM+ has the lowest capability in detecting land covers in the study area. It is recommended that these results should be repeated with more sensors, images and data. In producing the maps using the indices, it is recommended that initially images should be segmented based on different land cover types and then they should be used for land cover mapping.

Item Type: Thesis (Masters)
Persian Title: ارزیابی مناسب¬ترین شاخص¬های سنجش از دور در ارزیابی پوشش گیاهی منطقه خلخال
Persian Abstract: -
Supervisor:
SupervisorE-mail
-, -UNSPECIFIED
Advisor:
AdvisorE-mail
-, -UNSPECIFIED
Subjects: Faculty of Agricultural Sciences & Natural Resources > Department of Water Engineering
Divisions > Faculty of Agricultural Sciences & Natural Resources > Department of Water Engineering
Divisions: Subjects > Faculty of Agricultural Sciences & Natural Resources > Department of Water Engineering
Faculty of Agricultural Sciences & Natural Resources > Department of Water Engineering
Date Deposited: 09 Jul 2019 08:19
Last Modified: 09 Jul 2019 08:19
URI: http://repository.uma.ac.ir/id/eprint/7615

Actions (login required)

View Item View Item