How LiDAR data will be used in the LANDFIRE Remap

[Megan Sebasky]

Hi all,

I've heard a lot of interest, especially through the recent Northeast LANDFIRE priorities survey, in understanding how LiDAR data will be incorporated in the LANDFIRE Remap. Below is what I have learned so far, and may be subject to change.

Current: When LiDAR data became available at increasingly larger extents, the LANDFIRE team embarked on research to determine how they could use this data to improve LANDFIRE mapping efforts. For those unfamiliar, LiDAR is a new method of remote sensing that can capture very detailed information on ground elevation as well as vegetation structure (see representations here and here). This data has the potential to refine many of the datasets in the LANDFIRE suite: percent canopy cover, canopy height, canopy base height, and even understory vegetation metrics. However, due to its high cost, LiDAR data is not available nationally and therefore by definition cannot be a direct input to LANDFIRE mapping. Also, the strength of the research on using LiDAR for different metrics is variable: canopy height and canopy cover are well understood and supported at broad scales, while more complex metrics are not fully understood in different vegetation types across the country. Therefore, the mappers proposed to use LiDAR data only for vegetation height and cover (EVC and EVH) and only as training data in their remote sensing analysis. Where airborne lidar data are available (assembled by the 98 mapping tiles covering the conterminous US), canopy height and cover metrics are generated from the lidar point clouds, gridded at 30 m.  A sample of these pixels are then selected to use as supplemental training data for developing EVC and EVH models.  These models also rely extensively on plot-based training data as well as Landsat imagery and DEM and derivative data layers.

Future: LANDFIRE mapping will be able to incorporate more information from LiDAR data as it becomes more widespread in terms of distribution and knowledge. There is another USGS mapping program, the 3D Elevation Program (3DEP), which is leveraging LiDAR data, mainly in terms of topography but also maintaining awareness of the detailed vegetation structure data that can be calculated. 3DEP is fostering the collection of lidar data, which may be used in future LANDFIRE mapping.

 

Below is a map showing many (but not all) locations where LiDAR exists and can be accessed by the LANDFIRE team, and highlights the one area (mapping tile) in our region that does not have any LiDAR data.

[Megan Sebasky]

Update: LANDFIRE released more information on how LiDAR is being used in the Remap as well as metrics on how well it has improved structure data in the first two prototype areas. Here is the text (below) and image (attached) from their January postcard (available here: http://mailchi.mp/tnc/january-2018-postcard-remap-interview-helpdesk-data-call?e=b8bf47d388):

LANDFIRE Remap: Integrating lidar   LF Remap, an innovative vegetation and fuels mapping effort to produce current base maps of the LF product suite, has focused efforts towards advancing LF mapping methodologies spanning several topical areas. These areas include LF Reference Database (LFRDB), Satellite Image Compositing, Lifeform modeling, Existing Vegetation Type (EVT) modeling, and Vegetation Structure modeling.   For Remap, LF is amending the Existing Vegetation Height (EVH) and Existing Vegetation Cover (EVC) legends to represent continuous percent cover and height to represent the landscape structure characteristics and variability at a finer thematic resolution, on which fire fuel modeling is greatly dependent. Continuous structure products are possible by enhancing reference data through incorporating lidar data in combination with the LFRDB. Although there are tens of thousands of LFRDB plots across the United States (U.S.), structure data gaps remain in several regions. Incorporating lidar observations will increase reference data and reduce vegetation structure data gaps. LF is aware that lidar data are not available everywhere and is building a modeling process that attempts to mitigate this issue. What LF found was that incorporating lidar data in the two prototype areas (Grand Canyon and Northwest) increased the amount of EVC reference data by 310% in the Grand Canyon area and by 79% in the Northwest area. Further results of LF Remap prototyping in the two study areas confirmed that incorporating lidar-derived plots increases reference data considerably, resulting in a more comprehensive reference database that better represents the continuous nature of vegetation structure characteristics than using reference plots alone. Including lidar reference plots has shown higher correlations with validation plots for both EVC and EVH, indicating the inclusion of lidar reference data increases vegetation structure model accuracies. For more information, please review the poster: LANDFIRE Remap: Integrating lidar for Improving Vegetation Structure Mapping

[nelan...@gmail.com]

Update:

The New York Department of Environmental Conservation submitted LiDAR data to LANDFIRE to fill the one hole where LF was completely missing data, shown in the map on the first post of this thread. The data has been consolidated, sent across the country, and is now in the hands of the LANDFIRE mapping team in Sioux Falls, South Dakota. This is NOT an easy feat, and I thank April Marinov and her colleagues at the NY DEC for making this happen. Here is a map of the coverage of LiDAR submitted: