Abstract

VILLARROEL, Daniel et al. Estimation and modeling of the spatial distribution of aerial plant biomass for Bolivia. Ecología en Bolivia [online]. 2022, vol.57, n.1, pp.5-18. ISSN 1605-2528.

In the present work we estimate the amount of aerial plant biomass (AGB) remaining in Bolivia, and we model its spatial distribution (pixel resolution: 100 m), evaluating the prediction efficiency and uncertainty. The modeling was performed by comparing data from field inventories (504 plots) and spectral environmental variables [selected under correlation thresholds (r≥0.8) and collinearity (VIF≥10)] using the Random Forest automated learning method. The evaluation of the model was performed using the Pearson correlation coefficient. The AGB model reached a high level of prediction (R2= 0.789; RMSE= 39.4 t ha-1), with values between 2.9 and 354.5 t ha-1, being significantly similar to the one quantified in the field (1.9 to 375.4 t ha-1). The total AGB estimated for Bolivia was 9.77 Gt, of which 90.1% is concentrated in the departments of Santa Cruz, Beni, Pando and La Paz. The largest amount of AGB (4.33 Gt) is stored on 15.6% of the national territory, in areas that concentrate between 200-300 t h-1. The prediction uncertainty was >50% in areas that store between 0-10 t ha-1; and <20% in areas with densities >50 t ha-1. The 20.7% of the BA is protected in the 22 national protected areas, which are gradually being deforested. The AGB model is a source of information that will make it possible to carry out more precise estimates and inventories of GHG emissions, and to evaluate the impact of deforestation and conservation in Bolivia.

Keywords : Carbon; Physiognomies; Plots; Protected areas; Random Forest; Vegetation cover.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )