Landsat ETM+, Terra MODIS and NOAA AVHRR: Issues of scale and inter-dependency regarding land parameters.
Thomas Alexandridis
Ph.D. student in remote sensing and GIS applications (Aristotle Univ. of Thessaloniki)
13 Milona Street, Thessaloniki 54636, Greece
Tel: (30)- 310 -998778
E-mail: thalex@agro.auth.gr
Greece
Yann Chemin
PhD student in remote sensing applications (AIT-STAR)
206/7 Kamathawatte Road, Rajagiriya, Sri Lanka
Tel: (94)- 1-886053
E-mail: ychemin@yahoo.com
France
Abstract
Space observation has enjoyed increasing interest from various sides of resources management over the last decades. Within the past few years, the cost of data acquisition and processing has dropped, and a large amount of satellite data is already available on the Internet, free of any charge. This plethora of data has the advantage of bringing directly usable information to a wider range of users, both to scientists and to a larger, less specialized public. Whatever the case, the initial choice of the dataset depends on various parameters and special attention should be given to this. Under these conditions and with such an ungrudging supply of satellite data, the question arises: which is ‘better’ for a given application? But also, what does ‘better’ mean?
In this paper, three sensors used for the monitoring of vegetation are evaluated: NOAA AVHRR, Terra MODIS and Landsat ETM+. Although the issues of scale are the primary focus, the differences in calculating land parameters using imagery from these sensors are attributed to spectral, technical and scale factors. Multi-resolution analysis is performed on the examined datasets in order to identify the optimum scale of observation of vegetation in the study area. Moreover, the possibility of substitution of one with another is evaluated through checking the correlation coefficients between the datasets. Although it is believed that higher spatial resolution gives more accurate results in measurements made with remote sensing data, there is an indication that highest variability is explained by the vegetation index image of MODIS sensor, which leads to the conclusion that this may be the optimum sensor for this application. The level of similarity of the vegetation index pattern calculated from the three
sensors is not uniform, indicating that different sensors depict different characteristics of the ground. Valuable comments are made in the discussion regarding the usefulness of moderate resolution satellites.
1. Introduction
Remote sensing of land parameters has gained a lot in recognition since the high spatial resolution satellites for monitoring vegetation were launched in the late 70s. Since that time, research evolved in the applications of remote sensing for land management, towards developing methods to use satellites of various spatial resolution in monitoring of environment. Often reserved to scientists, because of cost and heavy calibration physics and processing, satellite imagery has become recently more affordable (in some cases even free). More important, a large amount of data preparation is now commonly performed before delivery. This makes satellite information more accessible to users, that require now only knowledge at the application level.
Available from the end of 1988, the 1.1 Km spatial resolution NOAA AVHRR satellite is the combined land and climate purpose sensor that is now the most widely used in medium to large area investigation. Its high spatial resolution correspondent is the Landsat series. Launched in 1999, Landsat 7 ETM+ is having 15- 30m visible and 60m thermal sensing pixel size. Earlier Landsat missions are not providing the comparative pre-processing level of NOAA AVHRR and Landsat 7 ETM+, making the transfer of its use to end-users a complex task. More recently (2000), MODIS sensor (on -board of Terra satellite) is roviding information from 250m to 1Km spatial level. Attractive advantage of MODIS is that a large amount of effort has been put into the development of directly usable pro ducts (information) after the processing of the data from the MODIS Science Team.
