Combined Use of Active and Passive Remote Sensing for Mapping Distribution and Biomass of Coastal Mangroves

Abstract:

Remote sensing provides a potentially fast, cost-effective, and efficient tool for mapping and monitoring mangroves located in relatively inaccessible areas where field measurements are often difficult and expensive. In this study, we examined the utility of combining Landsat-8 (LDCM), ALOS-PALSAR, and SRTM satellite imagery for mapping mangrove species composition, its canopy height and biomass distribution in the Mimika District of Papua, Indonesia. Image segmentation of ALOS-PALSAR radar data were used to delineate mangrove areas, while flexible statistical expert-based classification of spectral signatures from Landsat-8 (LDCM) images were used to classify mangrove associations. The overall accuracy of mangrove mapping for the entire area was 94.38% with kappa coefficient of 0.94 when validated with field data and QuickBird image data with 2.44 m spatial resolution. Mangrove height and biomass were mapped using the SRTM-based elevation, which were calibrated with field-measured canopy height via regression models. There was a strong linear relationship between the SRTM data and field-measured vegetation height (r = 0.87 and adjusted R² = 0.76). A bootstrap simulation of 10,000 runs with replacement resulted in an error of 3.03 m (RMSE) and 2.33 m (MAE) for mean tree height over 30 m pixels. SRTM-derived canopy height and plot-level biomass from the 22 mangrove plots showed a strong non-linear relationship with an R²=0.75. Our results showed that mangrove standing biomass in the Mimika District varies from 70.32 Mg/ha to 511.80 Mg/ha with mean biomass error of 65.23 Mg/ha (RMSE) and 58.10 Mg/ha (MAE) over a pixel of 90 m. This study explored a set of reliable methodologies which can be applied for mapping and monitoring mangrove dynamics of large areas in Indonesia.