Testing an Approach to Improving Fire Fuel Mapping by Modeling Fuel Structure and Types Based on Combined Satellite Imagery and Field Data - A Joint Fire Science Program Funded Research Effort

Project Summary:

This project was designed to study mapping techniques and prototype methodologies for meaningful vegetation parameters that can be used as input variables for large area fire fuels mapping and characterization. We recognize that remote sensing is ineffective for mapping fire fuels in a direct manner. However, satellite remote sensing may be effectively used to map/model natural vegetation types and structure classes that are essential elements of fire fuels characteristics (types, loadings, and distribution). There are research issues that need to be addressed to accomplish the objectives, including effective algorithms to map distribution of several key structure variables and integration of field data with remote sensing imagery.

Status Report:

• Mapping algorithms such as K Nearest Neighbor, Co-kriging, decision tree, and regression techniques have been researched in terms of literature review, technical evaluations, and preliminary results.
• K-NN algorithm has been developed using IDL and tested on two study areas (Alaska and Delaware River Basin)
• Preliminary results (vegetation types, age, basal area) have been completed showing good results.
• Conference papers are being drafted