Geo-statistical analysis for pattern evaluation of fire ignition points in rural areas: the adaptive kernel density estimation approach and its implementation in the territory of the Gargano National Park
Published 2005-02-28
Keywords
- kernel density,
- bandwidth,
- fire patterns,
- ignition points,
- fire occurrence
Copyright (c) 2013 Italian Journal of Forest and Mountain Environments
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
Abstract
Fire managers need to study fire history in terms of frequency and spatial distribution in order to predict and realistically estimate fire risk. The exact location and the «spot» nature of the fire ignition points should be converted into continuous
variables so that they could be integrated with other area data types. To better understand the gravity of the problem and its spatial frequency, the relative density should be an interpretable unit value (i.e. number of ignition points/ km2). In this paper we express the fire ignition points density as a continuous variable considering their space variability distribution by means of the adaptive kernel interpolation approach.
The study-case is an area of 1500 km2 in the Gargano peninsula, due to its high frequency of fire occurrence. The relative density has been estimated considering 583 fire ignition points over a six years period (1997-2002). A calibration method is proposed in order to set up the geo-statistical parameters useful to create a continuous density surface where the local overestimation and underestimation is minimised.
The density map allows to analyse the fire problems in the Gargano National Park. The localisation of the hot spots points out some important considerations, useful for fire management and for fire assessment.