| Title | Effect of the calibration of AVHRR data on the Normalized Difference Vegetation Index and compositing |
| |
| Author | D'Iorio, M; Cihlar, J; Morasse, C R |
| Source | Canadian Journal of Remote Sensing vol. 17, no. 3, 1991 p. 251-262 |
| Year | 1991 |
| Alt Series | Earth Sciences Sector, Contribution Series 20041242 |
| Document | serial |
| Lang. | English |
| Media | paper; on-line; digital |
| File format | pdf |
| Abstract | The calibration of NOAA data from Digital Signal Level (DSL) to apparent radiance is a crucial step in the processing of these data before compositing. Maximum Normalized Difference Vegetation Index
(NDVI) composites are used for vegetation monitoring of the Canadian landmass. The composite NDVI values display a smooth transition regionally within a given land cover type; however, the corresponding channel 1 and 2 values often show a bimodal
distribution, rendering a speckled appearance to the resulting composite images. In the majority of cases, the brighter pixels originate from high viewing angles. Furthermore, the sun-sensor geometry accentuates observable problems in early to
mid-summer images acquired from NOAA mid-afternoon ascending passes because of a "hot spot" being created, where shadows are minimized. Using a digital land cover map of Canada, it was found that, while the speckle problem is apparent over all types
of land cover, it is more pronounced in agricultural areas. The radiometric calibration of DSL from channels 1 and 2 increases relatively the NDVI values of near-nadir pixels. As a result, maximum NDVI composites derived from calibrated data are
substantially improved when more near-nadir pixels are selected. Nevertheless, residual speckle remains in composite channel 1 and 2 values. The reasons for this effect and the potential strategies for overcoming it are discussed in this paper.
|
| GEOSCAN ID | 218044 |
|
|