Modelação e caracterização do comportamento do fogo em diferentes níveis de cobertura florestal no posto administrativo de Combomune

Abstract

The present work aims to create, adjust models and characterize the behaviour of fire, in a mopane ecosystem located in administrative post of Combomune. This research made a direct mapping using LANDSAT 8 images (Operational Land Imager), relating the NDVI to the site- adjusted Torres canopy using coverage model through the least squares. Since mapping, three (3) levels of canopy coverage have resulted. Fifteen (15) plots of 5*10 m were allocated at each canopy cover level, where prescribed fires were conducted, data on dependent, independent variables (fuel models) of the propagation rate (R), and length (Fl). Knowing the influence of climate on fire behaviour, independent climate variables of the region coming from the Chokwè meteorological station were also used. The fire flame intensity models (I) had fire foci as an dependent variable. The I special analysis based on the fire foci MODIS MCD14ML product. The results showed a good fit of the Torres model with the parameters in Combomune (R 2 = 0,6479; RMSE = 0,56707) and with the NDVI (R 2 = 0,6826; RMSE = 21,246). The models of R and F1 present a very satisfactory performance, with R 2 Ajust varying from 0,4255 to 0,9978, while the models of I do not present a good relationship with their independent variables. The R and the F1 present a different behaviour for different levels of canopy coverage. The average flame R was 0,17 m/s, between a confidence interval of [0,04158 to 0,29841 m/s] (α=5%) and the average Fl, was a confidence of 1,23 m within a confidence interval between [0,234 to 2,226 m] (α = 5%). The average flame I was 39,316 KW with a confidence interval between [- 3,152 to 81,785 KW]. All fire behaviours parameters showed lower values in areas of dense forest [90-100%] of coverage for presenting a higher, frequency of trees with DAP greater than 7,6, which do not contribute to the propagation of surface fires