Alejandro Gonzalez-Calvo

Synergy of GIS and Remote Sensing Data in Forest Fire Danger Modeling

Forest fires constitute an important problem for the environment degradation. In this paper, we propose a Dynamic Fire Risk Index (DFRI) that takes into account different static and dynamic factors of risk for fire occurrence. Variables like insolation hours, vegetation cover, altitude, slope, proximity to main roads, and fire statistics have been used to develop a Static Fire Risk Index (SFRI) using a logistic regression model. Using satellite data to derive water stress of forest, a new dynamic index is defined weighting the static index with the actual value of water stress indicators. This methodology has been previously tested for some fires in the Canary Islands (Spain), and, in this case, we prove its usefulness using both NOAA-AVHRR and Terra-MODIS sensors data. As test sites, two different fires that took place in September 2005 on La Palma Island and August 2007 on Tenerife Island (Canary Islands, Spain) have been considered in order to validate the suitability of these tools for a regional scale application, in an area where multiple microclimates are present mainly due to its steep orography and the trade winds.

Pixel and object-based classification approaches for mapping forest fuel types in Tenerife Island from ASTER data

Four classification algorithms have been assessed and compared with mapped forest fuel types from Terra-ASTER sensor images in a representative area of Tenerife Island (Canary Islands, Spain). A BEHAVE fuel-type map from 2002, together with field data also obtained in 2002 during the Third Spanish National Forest Inventory, was used as reference data. The BEHAVE fuel types of the reference dataset were first converted into the Fire Behaviour Fuel Types described by Scott and Burgan, taking into account the vegetation of the study area. Then, three pixel-based algorithms (Maximum Likelihood, Neural Network and Support Vector Machine) and an Object-Based Image Analysis were applied to classify the Scott and Burgan fire behaviour fuel types from an ASTER image from 3 March 2003. The performance of the algorithms tested was assessed and compared in terms of quantity disagreement and allocation disagreement. Within the pixel-based classifications, the best results were obtained from the Support Vector Machine algorithm, which showed an overall accuracy of 83%; 14% of disagreement was due to allocation and 3% to quantity disagreement. The Object-Based Image Analysis approach produced the most accurate maps, with an overall accuracy of 95%; 4% disagreement was due to allocation and 1% to quantity disagreement. The object-based classification achieved thus an overall accuracy of 12% above the best results obtained for the pixel-based algorithms tested. The incorporation of context information to the object-based classification allowed better identification of fuel types with similar spectral behaviour.

Satellite image based methods for fuels maps updating

Regular updating of fuels maps is important for forest fire management. Nevertheless complex and time consuming field work is usually necessary for this purpose, which prevents a more frequent update. That is why the assessment of the usefulness of satellite data and the development of remote sensing techniques that enable the automatic updating of these maps, is of vital interest. In this work, we have tested the use of the spectral bands of OLI (Operational Land Imager) sensor on board Landsat 8 satellite, for updating the fuels map of El Hierro Island (Spain). From previously digitized map, a set of 200 reference plots for different fuel types was created. A 50% of the plots were randomly used as a training set and the rest were considered for validation. Six supervised and 2 unsupervised classification methods were applied, considering two levels of detail. A first level with only 5 classes (Meadow, Brushwood, Undergrowth canopy cover >50%, Undergrowth canopy cover <15%, and Xeric formations), and the second one containing 19 fuel types. The level 1 classification methods yielded an overall accuracy ranging from 44% for Parellelepided to an 84% for Maximun Likelihood. Meanwhile, level 2 results showed at best, an unacceptable overall accuracy of 34%, which prevents the use of this data for such a detailed characterization. Anyway it has been demonstrated that in some conditions, images of medium spatial resolution, like Landsat 8-OLI, could be a valid tool for an automatic upgrade of fuels maps, minimizing costs and complementing traditional methodologies.