Tall Timbers Research Station partnered with the United States Forest Service (USFS) Southern Research Station and Fish and Wildlife Service (FWS) to collect high resolution data for a prescribed burn plot in order to create data that could be used by high resolution fire behavior models. Field measurements, terrestrial laser scanning (TLS), and airborne laser scanning (ALS) were collected for 3 plots in the Piedmont region. The runs presented here are for the 'Western Plot' in the Hitchiti domain. Ignitions were done as a combination of handlines in the exterior and drone air-dropped ignition spheres ('ping pong balls'). ALS/TLS measurements were used to generate tree characteristics and surface fuel representations by USFS employees.
Provided are Zarr arrays containing the bulk density over time for 2 different runs: the original moisture condition and an adjusted moisture environment where riparian areas were modified to represent the higher moisture content present. Riparian areas were designated as those under 125m in elevation. The simulation was run for 17,000 s, which is a little over 4.5 hours. The first time step then is the initial condition of the fuel. The arrays are structured as [ntimes,nz,ny,nx]. The 'ntimes' is not the total simulation time but the amount of time-steps that were outputted. For these simulations, that would be every 100 seconds. Ny and Nx will be set for 1237, 1165 for both runs, and Nz is set to 16 (vertical cells in the fuel grid).
Provided is also the generating text files for the run with the exception of the DAT arrays containg the fuel information. The run files are left here for completeness.
QUIC-Fire - Version: Jan2022
Please contact Daniel Rosales (dgiron@talltimbers.org) or Zachary Cope (zcope@talltimbers.org) for any questions,comments or concerns about the data.
MODEL PARAMETRIZATION
Model Parametrization - Fuels
Canopy and surface fuels were built using the methodology described in Linn et al. 2005. Canopy fuel was constructed using tree locations and attributes derived from the ALS data collected for the burn plot. The geographic locations of the trees were constructed using a tree detection algorithm designed to find treetops within the canopy height model (CHM). The crowns were delineated using Silva 2016 algorithm and converted into polygons. The following tree attributes were calculated for each polygon: crown height, height to live crown (htlc), and crown radius. The tree attributes from the crown polygon were joined with the tree locations using a spatial join for all intersecting features. Using the Linn et al. 2005 fuel building algorithm, each tree was converted into a three-dimensional axisymmetric shape bound to the top and bottom by two paraboloids to represent an idealized tree. Fine fuels were added to each tree shape, with fuel declining toward the center of the trunk and toward the bottom of the canopy. Fuel from the trees was subsequently split between voxels based on how it overlapped with the three-dimensional voxel array. Surface fuel was constructed by approximating the placement of litter and grass fuels beneath the canopy. The Linn et al. 2005 algorithm assumes that grass concentrations fall beneath the canopy and that litter load increases based on the amount of canopy above it. Clip plot data from the site was used to calculate the average fuel loads for grass and litter fuel. Average litter fuel loading was calculated by taking the average sum of the dead woody litter, pine needle, and conifer litter from each clip plot, and average grass fuel loading was calculated by taking the average the surface fuel values categorized as other, which consisted of grasses, forbs, vines, and conifer seedlings. The average litter and grass fuel loading values were determined to be 0.53 and 0.15 kg/m2, respectively. Fuel loading concentration for litter and grass placed by the Linn et al. 2005 fuel building algorithm where set to match the average surface fuel concentrations calculated from the field data.
Model Parametrization - Weather
The reconstructions used hourly wind data collected at the Brender Remote Automatic Weather Station (RAWS) to simulate the direction and speed of the ambient wind.
Model Parametrization - Ignitions
The Hitchiti experimental burns used a combination of drip torch ignition and aerial drone ignitions. The reconstruction used point ignitions to simulate the ignition patterns of the burn. To replicate the aerial ignition pattern, point ignitions were placed in 10m increments along the portions of the drone’s flight path that the drone was dropping balls. This resulted in a total of 1,514 aerials which was slightly under the roughly 1600 ignitions reported by the drone pilot. Times for the aerial ignitions were determined by taking the timestamp from the nearest point of the drone flight path shapefile to each ignition point. To reconstruct the drip torch ignitions, point ignitions were placed in 2m increments along the line shapefile for the drip torch ignitions. The times for the drip torch ignition were determined using the start and stop time attributes for each line. It was assumed that each line was ignited at a constant rate. The times of point ignitions along the line were set to occur in even time intervals, with the first ignition occurring at the reported start time and the last ignition at the reported stop time.
Model Parametrization - Fuel Breaks
The streams and roads were two relevant fuel breaks on the western burn plot. Line shapefiles for each of these features were used to replicate the fuel breaks within the QUIC-Fire fuel domain. Streams and roads were given an estimated width of 3 and 6m, respectively, and fuel was removed from cells within the fuel domain that overlapped with the buffered line features
See Jupyter Notebook demonstrating how to access the data (https://github.com/BurnPro3D/data-api-notebooks/blob/main/access-QuicFire-QF-Hitchiti-Piedmont-Prescribed-Fire-Ignition-With-Adjusted-Moisture-Scenarios-data.ipynb)