SBIR-STTR Award

Common in the US, crown fires are the most dangerous type of wildfire. They travel rapidly and are more difficult to suppress than surface fires. There is a need to successfully model crown fire behavior due to the resource damage they cause. Modeling crown fire behavior is crucial for fire management activities. Fuel distribution is vital for determining fire behavior. To characterize fuels for wildfire risk, several crown fuel layers are required: canopy cover, stand height, crown base height, and crown bulk density. There is a current lack of validated techniques for producing these data layers and an inability of fire managers to take advantage of the data. The outcome of our research will translate ultimately into a better understanding of fire behavior, enhanced fire suppression efforts, increased safety for fire crews, and will ultimately reduce danger to human safety and the costs associated with wildfires. A major threat to national forests is uncontrolled wildland fires that affect natural resources (timber), forage for livestock and wildlife, outdoor recreation, and a vector to sequester carbon. Healthy forests provide habitats for various plant and animal species and protect soil quality, prevent soil erosion, and improve water quality. This proposed research will provide a tool for the USDA so they can provide research information and technologies to better manage forest ecosystems. It will allow USDA and other federal and commercial agencies to calculate forest biomass and potential fuels, as well as providing reproducible inputs into FARSITE. OBJECTIVES: HyPerspectives will address a serious obstacle to wildfire management by providing a software solution for the creation of crown fuel layers for the FARSITE fire simulator. By doing so, we will connect recent advances in Light Detection and Ranging (LiDAR) technology with an end-user community who is in need of LiDAR derived fuel data products. States, such as California and Wyoming, with high wildfire risks have initiated collection of statewide LiDAR coverage, while North Carolina has recently completed its statewide coverage. Other states have initiated plans for collection of statewide coverage. Thus the amount of LiDAR data accessible to the forestry/wildfire community is large, and continuously increasing. HyPerspectives will draw upon its successful experience in LiDAR software development, wildfire fuels research, and extensive ground-truth data to validate methods for extracting fire fuel parameters from LiDAR datasets. HyPerspectives will expand its current Extract LiDAR Features-Base (ELF-Base) software and design a new software module, ELF-Forest, which will allow users to calculate stand height, crown base height, crown bulk density, canopy cover, tree stem maps, slope, and aspect models for ingestion into the FARSITE model. Currently there are other software programs for deriving some forestry parameters, but none to date are comprehensive and simple for end-users to implement. During the creation of ELF-Forests HyPerspectives will create tools for the extraction of canopy bulk density and crown base height measurements from the LiDAR data that are FARSITE ready and allow for the increased accuracy of crown fire simulations. The ELF-Forest package will hold a strong potential for commercialization among foresters and wildfire managers. HyPerspectives has developed and will continue development of these processing tools within the ENVI/IDL (Environment for Visualizing Imagery/Interactive Data Language) program environment. The proposed work will also have many additional applications such as monitoring agricultural resources, combating eco-terrorism, monitoring energy sources, and measuring sequestered carbon in vegetation. APPROACH: The overriding goal of this project is to assess the applicability of several types and resolutions of LiDAR data for quantifying canopy parameters for use within the forestry and wildfire management communities. To accomplish this task we will adapt methods from the published literature to create input layers for the FARSITE modeling program and datasets useful for forest characterization. We will use our extensive ground truth data to validate each parameter and assess their potential for automated extraction. Our approach will address four critical areas: 1) Define the optimal preprocessing steps for LiDAR DTM and DFM creation with respect to quantifying wildfire risk factors. 2) Assess the data resolution and number of returns necessary for accurate crown base height calculations. 3) Validate both the empirical and theoretical methods for estimations of fire fuel parameters. 4) Define the requirements of LiDAR processing for end-users, and assess the potential for each fire fuel and forest characterization parameter for automation

The USDA is the managing agency for 192.5 million acres, much of which is national forest. A major threat to national forests is uncontrolled wildland fires. These fires are a serious threat to goods and services provided by national forests, including natural resources (timber), forage for livestock and wildlife, outdoor recreation, and a vector to sequester carbon. Healthy forests provide habitats for various plant and animal species and protect soil quality, prevent soil erosion, and improve water quality. This project will allow the USDA to better manage forest ecosystems through innovations in technology. It will allow the USDA and other federal and commercial agencies to calculate forest biomass and potential fuels, as well as provide reproducible inputs into the federally-mandated FARSITE fire software application. The end result will produce a toolkit in which 80% of our processing methodology is automated and 20% is customizable. Customization services will be sold to end-users in the fire management, timber, and insurance industries in conjunction with our web-based visualization application FIRESCAPET. Our browser-based mapping application fuses and displays data from real-time sensor networks at active fires with numerous remote-sensing imagery and weather forecast products. The services developed under this project will directly fill a significant deficiency in the firefighting community??-the lack of accurate, relevant, and accessible data for ingestion into fire spread models. OBJECTIVES: HyPerspectives partnered with Anasphere, Inc. (a successful small business in Bozeman, Montana) and Cisco Systems, Inc. (a world leader in networking technology) to enhance our FIRESCAPE application to offer not only advanced remote-sensing data products and customized reports, but on-site, real-time weather data, GPS tracking, and full data transfer and communications networks (including audio and video) deployable to active fires. With the added outcomes expected from the USDA Phase II effort, FIRESCAPE will ultimately provide end-users access to a complete team of expert analysts and engineers to gather, merge, and analyze fire-related data products through satellite communications networking. Our experts will then consolidate and simplify all the available data into custom, real-time data reports with geospatial context and deliver it to end-users to expedite high-level decision making, which can save valuable assets and lives. The FIRESCAPE application is ideally suited for delivering LiDAR-derived fire fuel maps (FARSITE ready) which will be created through this USDA SBIR project. The platform also provides a channel by which end-users can have access to pre- and post-fire remotely sensed data products (e.g., estimates of recoverable lumber, erosion estimates, forest inventories, etc.). Currently, the remote sensing products available through the application are based on MODIS imagery. With the success of this USDA project, we will be able to offer higher accuracy products to the fire management community. By offering these map products to end-users in the field utilizing Ciscos secure, integrated database management and internetworking solutions for data transfer and delivery, we are streamlining the decision process for end-users. The fire fuel layers created from HyPerspectives LiDAR datasets and processing methods offer an accurate, high-resolution solution for forest fire simulation modeling, such as within FARSITE. These data layers, which include stand height, canopy cover, crown base height, crown bulk density, and terrain models, will provide inputs to a tool that is federally mandated for use within the U.S. Forest Service to forecast the spread of wildfires. Our high-resolution data layers will lead to enhanced software simulations that can reduce firefighting risks and costs and improve wildfire control. A reduction of the damage that fires inflict on rural communities and increases in fire fighter safety are high-level returns expected from this research investment. APPROACH: In the Phase II research plan we describe several approaches for segmenting forested areas by CBH and waveform analysis to treat canopies with different shapes individually. These segmentation approaches show potential for higher-accuracy CBD determinations than currently available, increased portability and reproducibility, and are amenable to automation. Each approach will be tested with more data from Yellowstone and several new locations during Phase II. To develop models that are truly portable to forests across a wide range of ecosystem conditions, more data must be collected to validate the relationships between LiDAR metrics and true biomass values. HyPerspectives will meet the following six technical objectives during Phase II of the project: 1) Optimize successful Phase I methodologies to increase the efficiency of our LiDAR data processing 2) Validate LiDAR-derived canopy bulk density with more field data 3) Investigate the utility of LiDAR data to create maps of coarse woody debris 4) Test the portability of HyPerspectives predictive models 5) Develop routines for data fusion with optical imagery for increased utility 6) Integrate the LiDAR-derived fire fuel maps with FIRESCAPE Fire managers, in particular, do not have the budget or experienced staff to justify the purchase of a software tool for their own operation. As a result, we have shifted our focus from producing a LiDAR software program to providing a service where HyPerspectives processes and distributes LiDAR data products. The final software toolkit will not be a shrink-wrapped style package, but rather a series of automated steps for use by HyPerspectives analysts