Aboveground biomass and Leaf Area Index (LAI) mapping for Niassa Reserve, northern Mozambique

Estimations of biomass are critical in miombo woodlands because they represent the primary source of goods and services for over 80% of the population in southern Africa. This study was carried out in Niassa Reserve, northern Mozambique. The main objectives were first to estimate woody biomass and Leaf Area Index (LAI) using remotely sensed data [RADARSAT (Cband, lambda = 5.7cm)] and Landsat ETM+ derived Normalized Difference Vegetation Index (NDVI) and Simple Ratio (SR) calibrated by field measurements and, second to determine, at both landscape and plot scales, the environmental controls (precipitation, woody cover density, fire and elephants) of biomass and LAI. A landcover map (72% overall accuracy) was derived from the June 2004 ETM+ mosaic. Field biomass and LAI were correlated with RADARSAT backscatter (r(biomass) = 0.65, r(LAI) = 0.57, p < 0.0001) from July 2004, NDVI (r(biomass) = 0.30, r(LAI) = 0.35; p < 0.0001) and SR (r(biomass) = 0.36, r(LAI) = 0.40, p < 0.0001). A jackknife stepwise regression technique was used to develop the best predictive models for biomass (biomass = -5.19 + 0.074 * radarsat + 1.56 * SR, r(2) = 0.55) and LAI (LAI = -0.66 + 0.01 * radarsat + 0.22 * SR, r(2) = 0.45). Biomass and LAI maps were produced with an estimated peak of 18 kg m(-2) and 2.80 m(2) m(-2), respectively. On the landscapescale, both biomass and LAI were strongly determined by mean annual precipitation (F = 13.91, p = 0.0002). On the plot spatial scale, woody biomass was significantly determined by fire frequency, and LAI by vegetation type.