Dergi makalesi Açık Erişim

DERIVATION OF PM10 LEVELS USING OBRA ON LANDSAT 5 TM IMAGES: A CASE STUDY IN IZMIR, TURKEY

2015    Ozelkan, Emre; Karaman, Muhittin; Mostamandy, Suleiman; Avci, Z. Damla Uca; Toros, Huseyin

Remote sensing is a convenient and promising technique in air quality management. It can be used to determine local and global spatial distribution of Particulate Matter (PM10) data. Determination of PM10 data using multispectral satellite images' reflectance values was the main aim of this study. For this purpose, 19.02.2010 and 26.03.2011 dated Landsat 5TM multispectral satellite images were used. The PM10 data of Izmir, a metropolitan city in Turkey, was also gathered from in situ air quality measurement stations. The highest, lowest and average PM10 values measured were 172, 92.5 and 124.4 mu gm(-3) on 19.02. 2010, and 64.5, 31 and 48.7 mu gm(-3) on 26.03. 2011, respectively. Difference images, that were obtained by subtraction of atmospheric corrected images from top of atmosphere reflectance (TOA) calculated images, were used in the study. Optimum Band Ratio (OBRA) method was used to produce all possible band ratio (BR) indices of the difference images. Additionally, the filter sizes from 1x1 (direct value of a station) to 11x11 were applied to generate difference images for determining the suitable filter size correlated ones with in-situ PM10 data. The results were correlated with PM10 values. The BR of 1650 nm centered Mid-IR 1 (band 5) and 2315 nm centered Mid-lR 2 (band 7) with 3x3 filter size was found to be suitable to determine PM10 data with remote sensing. The highest correlation coefficients between PM10 and the values of [band 5 (B5)/band 7 (B7)] and B71B5 ratio indices were obtained as r=0.86 for 19.02. 2010 and 26.03. 2011, respectively. The results show that the B5/B7 and B71B5 ratio values of Landsat 5TM were more con-elated and appropriate than other BRs to determine PM 10. Additionally, the simulation of an online coupled regional meteorology-chemistry Enviro-HIRLAM model was run for the two dates. Model 45 h forecast results, being compatible with remote sensing results, indicated that the PM10 content in the region is higher on 19.02. 2010.