Biradar, CM, Thenkabail, PS, Islam, MA, Anputhas, M, Tharme, R, Vithanage, J, Alankara, R, Gunasinghe, S (2007). Establishing the best spectral bands and timing of imagery for land use-land cover (LULC) class separability using Landsat ETM+ and Terra MODIS data. CANADIAN JOURNAL OF REMOTE SENSING, 33(5), 431-444.

Abstract
The main goals of this study were to (i) establish Landsat enhanced thematic mapper plus (ETM+) and moderate resolution imaging spectroradiometer (MODIS) spectral bands best suited for land use - land cover (LULC) class separability, and (ii) study the role of the timing of imagery best suited for LULC class mapping. The study was carried out in the lower portion of the Uda Walawe River basin of southern Sri Lanka. The expansion of irrigated agriculture in this basin has resulted in several distinct changes in the LULC classes and their distribution. The area is dominated by agriculture, plantations, chena (slash and burn) lands with various types of natural vegetation such as degraded forests and scrubland, and wetlands with recently developed irrigation canals and tanks. The results showed that the two shortwave-infrared (SWIR) bands of Landsat ETM+ (bands centered at 1.650 and 2.220 mu m) and MODIS (2.130 and 1.640 mu m) and the thermal band (11.450 mu m) of Landsat ETM+ were most sensitive in separating an overwhelming proportion of the 15 LULC classes studied. However, other bands, though not as powerful as thermal or SWIR bands, by themselves, often play a vital role in separating certain specific LULC classes that are not easily separable by thermal and (or) SWIR bands. The MODIS monthly time series showed that the timing of the imagery was crucial in the separability of LULC classes. An overwhelming proportion of the classes were separated from one another using the data for the two wettest months (November and December) and the driest month (July). All 15 LULC classes were separable using the three wettest months (November, December, and January) and the two driest months (June and July).

DOI:

ISSN:
1712-7971