Modelización de biomasa forestal aérea mediante técnicas deterministas y estocásticas

Abstract

By estimating forest biomass, it is possible to determine the amount of forest resources existing in a given territory, however, this is an expensive and time-consuming process. Therefore, the objective of the present study was to model the aerial tree biomass of a forest ecosystem located in the south central region of the state of San Luis Potosi. To define this process, we compared deterministic methods (weighted reverse distance) and stochastic methods ( kriging and cokriging ), with which the forest biomass was determined based on field and spectral data. Field data corresponded to 50 conglomerates of the National Forest Inventory, from which the biomass was calculated by using allometric equations. The spectral data (traditional NDVI -red and infrared bands of electromagnetic spectrum) were derived from a Landsat 5TM image of the year 2009. With the results of calculation of biomass and NDVI, semivariograms and cross variograms were tested with spherical, exponential and gaussian models to analyze which will result in the best fit. Subsequently, the exponential model derived from the cokriging technique was selected, based on which a square root value of the RMSE mean squared error of 32.01 Mg ha -1 was obtained. Finally, based on the selected model, a map was generated of the aerial biomass distribution, in which results that range from the 0.85 Mg ha -1 to 157 Mg ha -1 are presented.