Coastal wetland monitoring using optical and SAR time series

KARSZENBAUM, H., KANDUS, P., PRATOLONGO, P., STUMFT, R. Y MARTINEZ, J. M.

Miami, USA

Conferencia; 7th International Conference on Remote Sensing for Marine and Coastal Environments; 2002

National Oceanic and Atmospheric Administration (NOAA)

In this paper we used 15 LANDSAT 7 +ETM and 14 ERS-2 SAR images acquired during the 1999-2001 period. Images were registered and rectified to UTM coordinate system. Digital counts in LANDSAT images were transformed to reflectance and atmospheric factors affecting path radiance were removed. NDVI, infrared indexes were calculated. SAR data were calibrated, co-registered, and temporally filtered to reduce the radiometric uncertainty due to speckle. During fieldwork plant community composition, stand density and plant biomass data were recorded. The Normalized Vegetation Index (NDVI) and coefficient of backscattering were calculated. Training samples of about 5 pixels each from 10 different cortadera marsh sites and 20 junco marsh areas were extracted and analysed. Seasonal changes in day length were modelled trough a sine function. For cortadera marsh mean annual NDVI, and Sigma signal curves were obtained fitting the Julian day-ordered series to the sine function attained. Results indicate that optical and SAR temporal patterns of junco communities are complex, and difficult to interpret. In the upstream portion of the delta, this is due to fire dynamics and, in the recently formed bars of the De la Plata River, because both optical and radar signals are mainly conditioned by the effect of tides. Results obtained on cortadera marshes indicate that temporal trends in both, optical and radar signals show similar behavior reflecting green/death biomass temporal patterns and having a stable performance in terms of optical and microwave response. Interannual variability observed in both, optical and SAR time series, for cortadera marsh suggests a greening delay that is related to differences in both mean temperature and rain patterns. This work shows that optical and radar data could become useful tools for wetland health monitoring and for determining the influence of environmental factors in wetland vegetation growth patterns. Furthermore, once the normal pattern has been modelled, deviations from normal could be determined, and explained.