Landsat ETM+, Terra MODIS and NOAA AVHRR: Issues of scale and inter-dependency regarding land parameters.
The first hindrance found in having high spatial resolution satellite images like Landsat is that information is not available very often (i.e. 16 days) while satellite of bigger pixel sizes are revisiting the same part of the earth very often. To overcome this, the technique of combining data from more than one satellite, to take advantage of high temporal and high spatial resolution, is commonly used. Moreover, a difference in the output signal of the three sensors is expected since each sensor is capturing the reflected energy in varying band widths. In Table 1 are listed the band widths of the red and near-infrared bands of the three sensors and their radiometric resolution. Both ETM+ and MODIS red and near-infrared spectral bands are narrower than those of AVHRR, causing different response from vegetation, which in turn alters the computed vegetation index. The narrower width in the near-infrared part of the spectrum elimi nates the effect of the water absorption region (around 950 nm and 1100 nm),
and also renders the red band more sensitive to chlorophyll absorption (van Leeuwen et al., 1999). However small these effects, can give significant differences in vegetation response of dense canopies and multiple leaf layers (Hoffer, 1978).
Table 1: Band widths and radiometric resolution of red and near-infrared bands.
|
| sensor | red (nm) | near-infrared (nm) | radiometric resolution |
|
| ETM+ | 630-690 | 750 -900 | 0-255 |
| MODIS | 620-670 | 841 -876 | 0-4095 |
| AVHRR | 580-680 | 725 -1100 | 0-1023 |
|
Vegetation indices have become a common tool in monitoring vegetation change, for land cover classification, for calculation of leaf area index and biomass production. They provide a spatially distributed indication of chlorophyll appearance, easily calculated from satellite spectral bands, which is their main advantage.
Primary objective of this paper is to assess the differences of land parameters calculated through various sensors. A key land parameter under consideration is the vegetation index NDVI, and the examined sensors are two widely used and a recently available one. Another objective is to attribute the differences to spectral, technical and scale factors, and check the possibility of substitution of one with another.
1.1 Study Area
Situated in the Hubei Province, Central China, the Zhanghe Irrigation District is a sub-basin north of the Yangtze Changjiang River. The net irrigated area reported is approximately of 160,000 ha, providing a large proportion of Hubei Province grain production. Rice production is widespread in the irrigation system, and the recent decline in water availability to agricultural purposes has not decrease much the global rice production due to a proportional increase of efficiency of water use by the farmers (Dong et al., 2001). The undulating terrain is constraining the size and shape of the fields, and their discrimination is a difficult task, even with higher resolution satellites.
