Use of spatial extrapolation as a strategy for estimating forest resources in non-inventoried areas

Abstract

In Mexico, the authorization of a forest use is subject to having information collected in the field, through a forest inventory.
However, forests are often located in difficult access conditions. This implies that their evaluation takes a long time and is
expensive, however, the current availability of remote sensors has supported the development of strategies that are not only cheaper, but also accurate. One of these is based on the extrapolation perspective, through which estimates of areas with low sampling intensity can be derived, based on information from areas where data with normal sampling intensities have been obtained. However, both areas must have homogeneous conditions in terms of species, densities, structures, topographic conditions, etc. According to the above, the objective of this work was to exemplify the extrapolation process carried out in temperate forests of the state of Chihuahua, Mexico. Through a traditional forest inventory, information was obtained from a region called La Nopalera, with which a model was developed, through regression and correlation trees (CART), for which the following variables were considered a) Field: density, possibility of use (m³ / ha) and crown diameter; b) From remote sensors: spectral bands of Landsat images and aerial photographs. The model was implemented in another forest region called Chocachi, for which information derived from a low intensity forest inventory was used. As a result, it was possible to estimate the possibility of exploitation in the latter region. The field values of the Nopalera region did not show a 1 to 1 relationship, in reference to the Chocachi, since the extrapolation estimates underestimated the values obtained in the field. However, based on a correlation equation, it was possible to establish the value of 2.4 as a correction factor.