Analysis of smoke and cloud impact on seasonal and interannual variations in normalized difference vegetation index in Amazon
Abstract
Normalized difference vegetation index (NDVI) derived from National Oceanic and Atmospheric Administration (NOAA)/Advanced Very High Resolution Radiometer (AVHRR) is a unique measurement of long-term variations in global vegetation dynamics. The NDVI data have been used for the detection of the seasonal and interannual variations in vegetation. However, as reported in several studies, NDVI decreases with the increase in clouds and/or smoke aerosol contaminated in the pixels. This study assesses the smoke and clouds effect on long-term Global Inventory Modeling and Mapping Studies (GIMMS) and Pathfinder AVHRR Land (PAL) NDVI data in Amazon. This knowledge will help developing the correction method in the tropics in the future. To assess the smoke and cloud effects on GIMMS and PAL, we used another satellite-derived data sets; NDVI derived from SPOT/VEGETATION (VGT) data and Aerosol Index (AI) derived from Total Ozone Mapping Spectrometer (TOMS). Since April 1998, VGT has measured the earth surface globally including in Amazon. The advantage of the VGT is that it has blue channel where the smoke and cloud can be easily detected. By analyzing the VGT NDVI and comparing with the AVHRR-based NDVI, we inferred smoke and cloud effect on the AVHRR-based NDVI. From the results of the VGT analysis, we found the large NDVI seasonality in South and Southeastern Amazon. In these areas, the NDVI gradually increased from April to July and decreased from August to October. However the sufficient NDVI data were not existed from August to November when the smoke and cloud pixels were masked using blue reflectance. Thus it is said that the smoke and clouds mainly cause the large decreases in NDVI between August and November and NDVI has little vegetation signature in these months. Also we examined the interannual variations in NDVI and smoke aerosol. Then the decrease in NDVI is well consistent with the increase in the increase in AI. Our results suggest that the months between April and July are the most reliable season to monitor the vegetation.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2004
- Bibcode:
- 2004AGUFM.B31B0208K
- Keywords:
-
- 1600 GLOBAL CHANGE (New category);
- 1640 Remote sensing