The emergence of a new generation of high-resolution remote sensing systems in recent years could potentially allow for more accurate and efficient estimation of crown fire behavior variables. With spatial resolutions often falling in the sub-meter range, the spatial data provided by these sensors can support more detailed measurement of the forest canopy structure, which in turn can increase the accuracy and effectiveness of existing fire behavior models. In particular, the ability of active infrared (laser) and microwave (radar) airborne sensors to acquire direct, three-dimensional measurements of canopy structure can significantly improve estimation of the quantity and distribution of crown fuels (crown bulk density, canopy base height, stand height). However, there is a need for the development of analytical techniques and algorithms to automatically and efficiently extract the required information from the enormous quantity of data provided by these high-resolution remote sensing systems, as well as to assess their utility and cost-effectiveness for the application of fire behavior modeling. The objective of this project is to carry out an extensive investigation of the utility of A) active infrared (LIDAR) sensor data and B) active microwave (IFSAR) sensor data for the estimation of crown fire fuel density, type and condition. This project will complement and support JFSP –funded fuels mapping and characterization efforts currently underway at the national, state, and local levels.

LIDAR (LIght Detection And Ranging) is an operationally mature remote sensing technology that can provide detailed geometric (XYZ position of reflecting surface) measurement of the forest canopy and the ground surface. In forested areas, LIDAR pulses bounce back from foliage and branches composing the canopy. LIDAR can therefore provide detailed information relating to the vertical distribution of canopy biomass -- measurements that can provide direct spatial data inputs to wildland fire behavior models such as FARSITE. In particular, analysis of the spatial distribution of LIDAR returns within a specified grid cell area could yield estimates of numerous spatial variables critical to fire behavior modeling – ground elevation, slope, aspect, height to base of live crown, canopy height, and crown bulk density.

We propose to investigate the potential utility of small footprint LIDAR data for providing information relating to type, condition, quantity and spatial distribution of crown fuels. The costs of LIDAR acquisition are steadily declining, indicating that this technology should provide a cost effective alternative to field measurement of crown fire behavior variables. It is expected that this data could be available at a cost of < $1/acre, making it competitive with existing remote sensing techniques and significantly less expensive than field-based approaches. As an optical sensor, LIDAR does have limitations as a source for spatial data acquired over extensive areas. LIDAR does not have the capability to penetrate cloud cover and therefore its availability is weather-dependent, an important consideration for real time data acquisition under adverse atmospheric conditions. In contrast, sensors operating in the microwave range of the electromagnetic spectrum, such as radar, do not have this limitation.

There is also tremendous potential for the use of IFSAR (InterFerometric Synthetic Aperture Radar) to estimate fire behavior variables over large areas at relatively low cost and with an all-weather capability. It is believed that this information derived from interferometric radar data could be directly related to the crown fire behavior variables that are important as inputs to current fire behavior models. In addition, backscatter intensity in radar images is related to foliage moisture, another variable that is of great importance in fire modeling.

We propose to relate both the three-dimensional geometry and the radar backscatter intensity information obtained from interferometric SAR data to the physical characteristics of the forest canopy vegetation. Advanced statistical inferential techniques can be utilized to estimate the parameters pertaining to different vegetation layers, including depth and density.

We expect to develop an estimate of the crown bulk density for a given cell size directly from the interferometric radar data, as well other spatial inputs to the FARSITE fire behavior model. The results of this radar analysis will be compared to the estimates of crown fire behavior variables obtained by optical sensors (LIDAR) and traditional field-based methods.

The cost of IFSAR data acquisition over extensive areas will likely be significantly lower than that of other active remote sensing technologies, including LIDAR. IFSAR can be acquired from a jet aircraft platform flying at relatively high altitudes (6000 – 12000 meters), enabling the acquisition of data over large areas in a short period of time (3600 – 7200 km² /hr), day or night, and through cloud cover. IFSAR, therefore, has a distinct advantage over optical remote sensing technologies if real time data acquisition is needed under adverse atmospheric conditions such as the presence of clouds or smoke.

Estimation of crown fire behavior variables has traditionally been based upon ground measurements. However, field measurement techniques are not standardized and are not consistently applied across agency boundaries, which can complicate and frustrate efforts to model fire behavior at the landscape scale. There is a need for a standardized, efficient approach to measuring crown fire fuels in forest stands exhibiting a wide variety of structural characteristics. In addition, this information from crown fuel measurements collected in the field needs to be integrated with information derived from remotely sensed data in a statistically robust approach to maximize the accuracy of fire behavior modeling.

Estimates of crown base height and canopy height can be obtained from routine stand inventory data. Aspect and elevation can be derived from a digital terrain model of the area, while estimates of crown bulk density can be generated from a tree list using a stand growth-and-yield model such as FVS.

We propose to utilize the tremendous amount of ground truth inventory data that is currently available to us (see Study Areas and Datasets section of this proposal) and to collect additional field data in order to both 1) validate remotely sensed estimates of crown characteristics, and 2) develop an integrated approach to measurement of crown fuels incorporating both field measurements and remotely sensed data.

In addition, we will utilize the large number of topographic survey measurements, acquired throughout both the Capitol Forest and Fort Lewis study areas, to assess the absolute accuracy of biomass spatial distributions that are estimated from remotely sensed data.

Additional field measurements will be carried out on all study areas, and will be based upon the field-based techniques developed by the crown fuel measurement studies currently underway at the USFS Rocky Mountain Research Station Fire Sciences Lab (Reinhardt, 2001) and the USFS Pacific Southwest Region AMS Enterprise Team (Fites-Kaufman, 2001).

This remote sensing project was funded in part by the Joint Fire Sciences Program. Additional funding has been provided by USDA Forest Service Pacific Northwest Research Station, US Department of Defense, Washington State Department of Natural Resources, and the Precision Forestry Cooperative at the University of Washington.

We propose to carry out an extensive investigation of the utility of A) active infrared (LIDAR) sensor data, B) active microwave (IFSAR) sensor data, and C) field-based techniques for the estimation of crown fire fuel density, type and condition. This project will complement and support JFSP –funded fuels mapping and characterization efforts currently underway at the national, state, and local levels.

• LIDAR data sets from both helicopter and fixed wing platforms with varying densities (high-density helicopter data acquired for Capitol Forest, 1998 & 1999; medium-density fixed-wing data acquired over Capitol Forest, 2000; high-density fixed-wing data acquired over Capitol Forest, 2003)
  LIDAR canopy model    |    LIDAR DTM
  
  
• Multifrequency, polarimetric IFSAR data (X-band and P-band) acquired in September, 2002 over Capitol Forest study area
  IFSAR canopy model    |    IFSAR DTM
  
  
• Large scale (1:3000) and medium scale (1:7000, 1:8000, 1:12000) aerial photography with a variety of emulsions (normal color, color infrared, B&W) and orthophotos.
  
  
• 350 surveyed ground topography points under forest canopy within Capitol Forest study area
  
  
• High quality photogrammetric mapping
  
  
• 125 1/5^th acre inventory plots in Capitol Forest with measurements of tree height, height to base of crown, diameter at breast height (DBH), species, crown class, condition. Stem locations were measured on a selection of these plots.
  
  

The major deliverable of this project is a thorough analysis of the capabilities of the LIDAR and IFSAR technologies for estimation of crown fire behavior variables. Results of these analyses will be reported in student theses, technical presentations, and peer-reviewed publications (listed below). In addition, a prototype software system will be developed to assist geographic information specialists in the processing of LIDAR and IFSAR data and the generation of canopy fuel maps.

Publications:   

Andersen, H.-E., W.W.Carson, R.J. McGaughey, S.E. Reutebuch, and G.F. Schreuder. 2003. Estimating crown fire behavior variables using airborne laser scanner data. (in preparation)

Andersen, H.-E., W.W.Carson, R.J. McGaughey, S.E. Reutebuch, and G.F. Schreuder. 2003. Estimating crown fire behavior variables in a Pacific Northwest conifer forest using multifrequency polarimetric InSAR (PolInSAR) (in preparation)

Andersen, H.-E.,  J. Foster, and S. Reutebuch. 2003. Estimating forest structure parameters within Fort Lewis Military Reservation using airborne laser scanner (LIDAR) data. Proceedings of the Second International Precision Forestry Symposium, USA, June 16-18, 2003. (to appear)

Reutebuch, S.,  H.-E. Andersen, K. Ahmed, T. Curtis. 2003. Evaluation of laser light detection and ranging (LIDAR) measurements in a forested area. In: Curtis, R., D. Marshall, and D. DeBell, eds., Silvicultural options for young-growth Douglas-fir forests: The Capitol Forest Study—Establishment and First Results. General Technical Report PNW-XXX. Portland, OR. U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station. (in press)

Andersen, H.-E.,  R.J. McGaughey, W. Carson,  S. Reutebuch, B. Mercer, and J. Allan. 2003. A comparison of forest canopy models derived from LIDAR and INSAR data in a Pacific Northwest conifer forest. International Archives of Photogrammetry and Remote Sensing, Dresden, Germany, 2003, Vol. XXXIV, Part 3 / W13. pp. 211-217.  

Reutebuch, S., R. J. McGaughey, H.-E. Andersen, and W. Carson. 2003. Accuracy of a high-resolution LIDAR-based terrain model under a conifer forest canopy. Canadian Journal of Remote Sensing 29(5): 1-9.

Andersen, H.-E.. 2003. Estimation of critical forest structure metrics through the spatial analysis of airborne laser scanner data. Ph.D. dissertation. University of Washington, Seattle, WA.

Andersen, H.-E., S. Reutebuch, and G. Schreuder. 2002. Bayesian object recognition for the analysis of complex forest scenes in airborne laser scanner data. International Archives of Photogrammetry and Remote Sensing, Graz, Austria, 2002, Vol. XXXIV, Part 3A, pp. 35-41.

Presentations:

Mercer, B., J. Allan, C. Hoffman, M. Schwaebisch, S. Reutebuch, W. Carson, and H.-E. Andersen. 2003. P-Band Polarimetric InSAR (PolInSAR) for extraction of ground DTM beneath forest canopy. International Geoscience and Remote Sensing Symposium, Toulouse, France, July 21-25, 2003

Mercer, B., J. Allan, N. Glass, C. Hoffman, M. Schwaebisch, S. Reutebuch, W. Carson, and H.-E. Andersen. 2003. Extraction of Bare-Earth DEMs Beneath Forest Canopy Using P-Band Polarimetric InSAR (PolInSAR). Advanced SAR Workshop, Saint-Hubert, Quebec, Canada, June 25-27, 2003.

Andersen, H.-E.,  R.J. McGaughey, W. Carson, S. Reutebuch, B. Mercer, and J. Allan. 2003. An evaluation of canopy height models obtained from LIDAR and InSAR data in a Pacific Northwest conifer forest. ISPRS Workshop on Three-dimensional Mapping from InSAR and LIDAR. Portland, Oregon, June 17-19, 2003.

Mercer, B.,  J. Allan, N. Glass, S. Reutebuch, W. Carson, and H.-E. Andersen.  2003. Extraction of ground DEMs beneath forest canopy using P-band polarimeteric InSAR (PolInSAR). ISPRS Workshop on Three-dimensional Mapping from InSAR and LIDAR.  Portland, Oregon, June 17-19, 2003.

Andersen, H.-E.,  R.J. McGaughey, W. Carson,  S. Reutebuch, B. Mercer, and J. Allan. 2003. An evaluation of canopy height models obtained from LIDAR and InSAR data in a Pacific Northwest conifer forest. Second International Precision Forestry Symposium, Seattle, USA, June 16-18, 2003.

Andersen, H.-E. Estimation of critical forest structure metrics through the spatial analysis of airborne laser scanner data. 2003. Olympic Natural Resources Center Winter 2003 GIS Conference, Forks, WA, February 27-28, 2003.

Andersen, H.-E. The use of airborne laser scanner data (LIDAR) for forest measurement applications. 2002. Western Mensurationists Conference, Leavenworth, WA, June 23-25, 2002.

Means, J.E., K.Winterberger, H.-E. Andersen, and D. Marshall. 2002. Estimating and mapping forest structural characteristics with small-footprint LIDAR. Workshop on Three-dimensional Analysis of Forest Structure and Terrain using LIDAR Technology, Victoria, BC, Canada, March 14-15, 2002

Accomplishments:   

• All field plots have been established at Capitol State Forest and Mission Creek study areas
  
  
• Field plots have been precisely georeferenced using differential GPS and laser surveying equipment
  
  
• High-density LIDAR data have been collected at Capitol State Forest (2003) and Mission Creek (2004) study sites
  
  
• Multifrequency (X-band/P-band) high-resolution IFSAR data have been collected at Capitol State Forest (2002)
  
  
• LIDAR/IFSAR data visualization software (FUSION) is currently in development
  
  
• Preliminary results indicated high correlations between LIDAR-based metrics and field-measured plot-level fuel estimates (canopy height: R2=0.98; canopy base height: R2=0.87; canopy bulk density: R2=0.85; canopy fuel weight: R2=0.87 )
  
  
• Preliminary results also indicate high correlation between IFSAR metrics and field-based plot-level fuel estimates (canopy height: R2=0.88; canopy base height: R2=0.87; canopy bulk density: R2=0.78; canopy fuel weight: R2=0.80 )
  
  
• Relationships developed can be used to generate spatially-explicit fuel maps in GIS format

Application and Technology Transfer:   

This research will provide a methodology to process raw, high-resolution, remotely-sensed data into information relating to the spatial distribution of canopy fuels. It is expected that the development of a methodology for generating accurate, high-resolution canopy fuel maps will be valuable for a variety of applications, including planning and monitoring for fuel treatments, fire behavior modeling, and risk assessment. The methodology and results developed in this project will be thoroughly documented through publication of several peer-reviewed papers. These methods and products will continue to be transferred to the fire science and forest research community through workshops, seminars, and public presentations. Processing and visualization software is currently under development and should be available by the completion of the project.

Pictures:   

• “LIDAR_data_treemodels” is a graphic showing the location of field-measured trees superimposed on the LIDAR measurements within a 1/5th acre plot within the Capitol State Forest study area (credit: Robert McGaughey, USDA Forest Service PNW Research Station)
  
  
• “IFSAR_data_treemodels” is a graphic showing the location of field-measured trees superimposed on the IFSAR measurements within the same 1/5th acre plot within Capitol State Forest. (credit: Robert McGaughey, USDA Forest Service PNW Research Station)
  
  
• “lidar_can_grd” is a graphic showing the LIDAR-derived canopy surface model over the LIDAR-derived digital terrain model for the Capitol Forest study area (credit: Ward Carson, USDA Forest Service PNW Research Station)
  
  
• “ifsar_can_grd” is a graphic showing the IFSAR (X-band) canopy surface model over the IFSAR (P-band) digital terrain model for the Capitol Forest study area (credit: Ward Carson, USDA Forest Service PNW Research Station).