FIRERISK contains the following attributes: VALUE, the identifier for each unique combination; COUNT, the number of 30 m cells occupied by each combination; the seven input variables PVT, potential vegetation type; COVERTYPE, CANOPY, and SIZECLASS; SLOPE, ASPECT, and ELECLASS; and twelve derived variables FUELMOD, fuelmodel; SPREAD and SPREADCLASS, rate of spread index and class; FLAME and FLAMECLASS, flame length index and class; FIRELINE and FIRELINECLASS, fireline intensity index and class. HFREGCH and HFREGSB, Colin Hardy and Steve Barrett historical fire regime models; CFREGCH and CFREGSB, Colin Hardy and Steve Barrett current fire regime models; and CONDCLASS, level of departure from Steve Barrett historical fire regime.
FIRERISK was assembled from existing datasets (see project descriptions below), and includes parts of 23 Landsat Thematic Mapper (TM) scenes acquired in 1991-1995. To highlight primary differences among mapping projects, the three project outputs -- SILC-1, CICP, and EMONT -- are summarized below.
USFS REGION 1 SILC-1 DATA. First major classification project undertaken by Wildlife Spatial Analysis Lab, completed in June 1996. Eighteen TM scenes in northern Idaho and western Montana were classified according to cover type, forest and shrub canopy cover, and forest and shrub size class. Riparian cover types were separately classified and stored in a different database; these were not classified according to canopy and size class. Minimum mapping unit (MMU) was 2 ha (5 ac) for upland types, 30 m for riparian. Average number of polygons per scene database (excluding riparian) was 290,000. For more information, please see: Redmond, R.L, Z. Ma, T.P. Tady, J.C. Winne, et al. 1996. Mapping existing vegetation and land cover across western Montana and northern Idaho. Final report, contract #53-0343-4-000012, submitted to USDA Forest Service, Northern Regional Office, Missoula, MT, 12 June 1996.
USFS REGION 4 CICP DATA. Completed in August 1997. Six TM scenes in central Idaho were classified according to cover type, canopy cover for forest and shrub types, and size class for conifer forest types. Cover types were generally classified according to the scheme used for SILC-1, although a few types were added or their codes slightly modified. Riparian cover types were maintained in the same database as all other attributes, and so the number of polygons per database nearly doubled. MMU was 2 ha for upland types, and 0.4 ha (>4 cells) for riparian. For more information, please see: Redmond, R.L., T.P. Tady, F.B. Fisher, M. Thornton, J.C. Winne. 1997. Landsat vegetation mapping of the Southwest and Central Idaho Ecogroups. Final report, contract #53-0261-6-25, submitted to USDA Forest Service, Boise National Forest, Boise, ID, 14 November 1997.
EASTERN MONTANA (EMONT, or SILC-2) DATA. Completed in November 1997. Twenty TM scenes in central and eastern Montana were classified according to cover type and, for forest types only, canopy cover. Size class was not classified for either trees or shrubs. As in central Idaho, a few cover types were added to or modified from the existing classification scheme. Also as in central Idaho, riparian cover types were integrated in the "main" database, and the number of polygons per database is roughly double that of SILC-1. MMU was 2 ha for upland types, and 0.4 ha for riparian. EMONT is often called SILC-2.
Outputs from these three projects were revisited for Montana Gap Analysis (MT-GAP), a statewide biodiversity assessment that hinged on construction of a land cover layer. (See Redmond, R.L., M.M. Hart, J.C. Winne, W.A. Williams, P.C. Thornton, Z. Ma, C.M. Tobalske, M.M. Thornton, K.P. McLaughlin, T.P. Tady, F.B. Fisher, S.W. Running. 1998. The Montana Gap Analysis Project: final report. Unpublished report. Montana Cooperative Wildlife Research Unit, The University of Montana, Missoula. xiii + 136 pp. + appendices.) Most significantly, land cover data were recoded from nearly 100 to just over 50 types; the MT-GAP classification scheme is reflected in this product. To make data as consistent and as useful as possible, scene classifications were updated slightly in other ways as well; again, those updates are reflected here.
TM SCENES BY SOURCE PROJECT
SILC-1: p38r29 (path 38, row 29), p39r27, p39r28, p39r29, p40r27, p40r28, p41r26, p41r27, p42r26, p42r27, p42r28, p43r26, and p43r27 (n = 13 of 23 scenes).
CICP: p40r29, p41r28, p41r29, and p42r29 (n = 4 of 23 scenes).
EMONT: p37r27, p37r28, p37r29, p38r27, p38r28, p40r26 (n = 6 of 23 scenes).
GENERAL LIMITATIONS
The following are general limitations of the database; more specific limitations are presented below and in pertinent sections of the metadata that also accompany the grid.
1. These data are derived from remote sensing and modeling to enable general assessments about fuel characteristics and potential fire conditions over large geographic areas. Any decisions based on the data should be supported by field validation and involve more detailed modeling and analyses, especially at scales finer than 1:100,000.
2. Although the resolution of the data is 30 m2 , their expected accuracy does not warrant analysis of areas or sites this small. More appropriate analysis areas would include subwatersheds (6th field hydrologic units) or linear zones, such as urban wildland interfaces, that extend over thousands of hectares.
3. The data products and model outputs represent a snapshot in time corresponding most closely to the early 1990s. For example, the dates of the satellite imagery from which the land cover inputs were derived ranged between 1991 and 1995.
Scale - The database was produced with an intended application at the mid-scale level, that is, across geographic areas from several hundred thousand to millions of hectares in size. The data provide a coarse-filter approach to analysis, meaning that not every occurrence of every type is mapped, only larger, more generalized distributions of certain types (be they vegetation, slope classes, fuel models, or their derivatives). The data are also based on the USGS 1:100,000 scale of mapping in both detail and precision.
When determining whether to apply the data to a particular use, there are two primary questions: do you want to use the data as a map for the particular geographic area, or do you wish to use the data to provide context for a particular area? The distinction can be made with the following example: You could use the land cover data to determine the approximate amount of ponderosa pine forest occurring in a county or a subwatershed, or you could map ponderosa pine with aerialphotography to determine the exact amount. You then could use the data to determine the approximate percentage of all ponderosa pine in the Northern Region that occurs in the subwatershed, and thus gain a sense of how important its distribution is to maintaining ponderosa pine in the region.
Appropriate Uses - The above example illustrates two appropriate uses of the data; as a coarse map for a large area such as a subwatershed or county, and to provide context for finer-level maps. A general list of applications follows:
Regionwide fuels and fire planning
Regional habitat conservation planning
County wildfire management planning
Coarse-filter evaluation of potential fuel buildups in relation to private property, human infrastructure (e.g., utility or transportation corridors, public water supplies, etc) roadless/wilderness areas, etc.
Determining management responsibilities for fuel loads and associated wildfire risk to facilitate cooperative management and planning to reduce both.
Environmental impact assessment and prioritization for post-fire rehabilitation efforts.
Estimation of potential economic impacts from loss of natural resources to recent wildfires.
Educating land managers, public policy makers, and citizens about general landscape conditions.
Inappropriate Uses - It is far easier to identify appropriate than inappropriate uses of GIS data. Generally speaking, the distinction between appropriate and inappropriate use is related to the resolution of the data, size of geographic area being analyzed, and precision of the answer required for the question; when these determinants cease to be compatible, the use becomes less appropriate. Examples of clearly inappropriate uses might include:
Use of the data to map small areas (less than thousands of hectares), typically requiring mapping resolution at 1:24,000 scale and using aerial photographs or ground surveys.
Combining these data with other data finer than 1:100,000 scale to produce new hybrid maps that will be used for fine-scale analyses.
Generating specific areal measurements from the data finer than the nearest hectare.
Establishing exact boundaries for post-fire rehabilitation or treatment.
Establishing definite occurrence or non-occurrence of any feature for an exact geographic area (for land cover, the percent accuracy will provide a measure of probability).
Estimating health or condition of single or small groups of cells.
Establishing a measure of accuracy of any other data by comparison with these data.
Altering the data in any way and redistributing them as a UM data product.
Using the data without acquiring and reviewing the metadata and readme file.
For rough measures of accuracy for land cover, canopy, and size class, please see project reports, especially for Montana Gap Analysis and Central Idaho (citations in Abstract above). Because much of the project area is covered by MT-GAP data, a summary of accuracy for MT-GAP land cover is provided below.
Thematic accuracy of the MT-GAP land cover map was assessed using a bootstrap method which did not require the collection of an independent set of reference data. Cover type classification accuracies were estimated for 45 types; these averaged 61.4%, and ranged from 4.4% for Western Hemlock to 93.2% for Missouri Breaks. Interpolation of the mean error estimates at each ground reference point allowed us to map the land cover accuracy across the state. Estimated mean accuracy exceeded 80% in the southwest corner (Beaverhead and Madison Counties) and in the western portion of the Highline in Glacier, Toole, and Pondera Counties; lower estimated accuracies were associated with some of the insular mountain ranges in central Montana from Gallatin County north through Cascade and Judith Basin Counties. For more information on accuracy of vegetation attributes, please refer to the MT-GAP project's final report. NOTE: land cover type values were changed using a recode table provided by Jeff Jones and Doug Berglund of the Flathead National Forest. No accuracy assessment was done on the land cover type recoding.
Slope, aspect, and eleclass attributes were mapped from a project-wide digital elevation model (DEM) assembled from USGS 7.5 minute DEMs. Accuracy of the project-wide DEM was not assessed, but is assumed to fall within the typical range of the source data. "The accuracy of a DEM is dependent upon the level of detail of the source and the grid spacing used to sample that source. The primary limiting factor for the level of detail of the source is the scale of the source materials. The proper selection of grid spacing determines the level of content that may be extracted from a given source during digitization." (Taken from generic metadata for 7.5 minute DEMs, http://edcwww.cr.usgs.gov/nsdi/html/dem75/dem75) For more information on DEM accuracy, see:
USGS National Mapping Program Standards, http://mapping.usgs.gov/standards/
Standards for Digital Elevation Models, ftp://www-nmd.usgs.gov/pub/dem_html/standards_dem.html
7.5-Minute Digital Elevation Model Data, http://edcwww.cr.usgs.gov/glis/hyper/guide/7_min_dem
Not all of the inputs were available for the entire project area; for example, some data are missing for potential vegetation (PVT) and human population density, especially in central Idaho. Missing data are flagged accordingly.
Although cover type, canopy, and size class attributes are logically consistent, it should be noted that:
1) Because SILC-1 riparian data were mapped in a separate database, then pasted back into the COVERTYPE grid for this product, riparian cells have inherited CANOPY/SIZECLASS values from the underlying SILC-1 polygons they replace. Although the resulting COVERTYPE/SIZECLASS/CANOPY combinations have been made consistent by riparian lifeform, and thus are logical (e.g., canopy class 2 and size class 3 assigned to riparian broadleaf forest), the two attributes still come from separate classifications with different polygon structures.
2) SIZECLASS was not mapped within the EMONT area. Because of the importance of that attribute, for the two TM scenes (P38/R27 and P38/R28) that were classified for both EMONT and SILC-1, we used COVERTYPE from the more recent EMONT project (with improved classification techniques), but SIZECLASS for forest types from the SILC-1 project. As with riparian above, this means the two attributes came from separate classifications with different polygon structures. Hence, we took steps to minimize the effects of those differences (see processing steps).
Once 7.5 minute DEMs were acquired at 30 m resolution, the necessary steps were taken to convert them to ArcInfo grid format. Individual grids were then merged to tiles corresponding to USGS 1:100,000 scale quadrangles (30 x 60 minute blocks), and projected to Albers Equal Area Conic Projection (parameters as listed below, except that the latitude of origin was set at 44.25 degrees). The 100k tiles were then assembled into a 30 m project-area grid.
Albini, F.A. 1976. Estimating wildfire behavior and effects. USDA Forest Service Gen. Tech. Rep. INT-30, Intermountain Forest & Range Exp. Station, Ogden, Utah, 84401.
Anderson, H.E., 1982. Aids to determining fuel models for estimating fire behavior. USDA Forest Service Gen. Tech. Rep. INT-122, Intermountain Forest & Range Exp. Station, Ogden, Utah, 84401.
Andrews, P.L. and C.H. Chase. 1989. BEHAVE: Fire behavior prediction and fuel modeling system -- BURN Subsystem, Part 2. USDA Forest Service Gen. Tech. Rep. INT-260, Intermountain Forest & Range Exp. Station, Ogden, Utah, 84401.
Barrett, S. W. 2000. Modeling historical and current fire regimes for the northern Rocky Mountains. Unpubl., Draft contract report (PO: 43-0385-0-0052), USDA Forest Service, Kalispell, MT. 39 pp.
Hardy, C.H., J.P. Menakis, J.K. Brown, and D.L. Bunnel. 1998. Mapping historic fire regimes for the western Unitied States: Integrating remote sensing and biophysical data. In pp. 288-300, Proc. of the 7th Forest Service Remote Sensing Applications Conf., 1998 April 6-9; Nassau Bay, TX, Bethesda MD.
Jones, J., Brewer, K., Enstrom, G., and J. Caratti. 1997. Documentation of the modeling of potential vegetation settings and vegetation response units using topographic variables. USDA Forest Service, Northern Region, electronic publication.
In ArcEdit, created a composite boundary for the FIRERISK project area based on boundaries for SILC-1 data and for PVT data.
Buffered the composite boundary by 1 km, then comverted to 30 m grid format. The boundary grid was used in later processing steps to ensure a consistent number of rows and columns, as well as a consistent pixel increment, in future grid outputs.
SIZECLASS was not mapped for EMONT scenes, but is important for defining fuel models and other anticipated applications. Thus, within the EMONT area, for the two TM scenes (P38/R27 and P38/R28) that were classified for both EMONT and SILC-1, we used COVERTYPE from the more recent EMONT project (with improved classification techniques), but SIZECLASS for forest types from the SILC-1 project. In order to minimize scale differences between the two projects and prevent polygon fragmentation, SIZECLASS was calculated for each EMONT polygon using a ZONALMAJORITY filter of a merged SILC-1 SIZECLASS grid. This was considered a reasonable step because: 1) although the polygon structure differs between the two classifications, the same imagery date was used; and 2) due to the more detailed polygon structure of the EMONT classifications, we presumed that EMONT polygons were generally unlikely to encompass multiple SILC-1 polygons of different SIZECLASS.
Codes differed between grids. A common coding system was developed based on the PVTEAST codes, with new codes added for those PVTWEST types that did not occur within PVTEAST. PVTWEST was recoded as follows:
PVTWEST Code Combined Code Description 1100 27 Urban 2100 12 Agriculture 3100 15 Dry grass 5000 28 Water 7300 19 Rock 7500 37 Barren 8100 14 Alpine grass 1 30 ABRG1 2 31 ABGR2/PSME1/PSME2 3 32 ABLA1/ABLA2 4 33 PIAL/LALY 5 34 PICEA 6 9 PIPO 7 35 TSHE/THPL 8 36 TSME
After PVTWEST was recoded, it was combined with PVTEAST to form a single PVT grid. In areas of overlap between the input grids, the PVTEAST grid was used because it had been most recently updated. The output grid had 30 m resolution.
On 8/23/00, PVT codes were examined by Ken Brewer and Jeff Jones, who recommended three changes:
1) Collapsing codes 6 and 34, picea_east and picea_west, into code 6;
2) Combining codes 4 and 5, laly and pial, with code 33, pial/laly; and
3) Using the 30 m digital elevation model to split code 32, abla1/abla2, so that areas at or below 1830 m are coded 1, and areas above 1830 m are coded 2.
Changes were made, and a new 30 m grid was created.
PVT recodes: If COVERTYPE = 1100 and PVT was not 27 then assign PVT = 27 If COVERTYPE = 2010 and PVT was not 12 then assign PVT = 12 If COVERTYPE = 2020 and PVT was not 12 then assign PVT = 12 If COVERTYPE = 3100 and PVT = 0 then assign PVT = 16 If COVERTYPE = 3110 and PVT = 0 then assign PVT = 16 If COVERTYPE = 3130 and PVT = 0 then assign PVT = 16 If COVERTYPE = 3150 and PVT = 0 then assign PVT = 16 If COVERTYPE = 3170 and PVT = 0 then assign PVT = 16 If COVERTYPE = 3180 and PVT = 0 then assign PVT = 16 If COVERTYPE = 3300 and PVT was not 20 then assign PVT = 20 If COVERTYPE = 3309 and PVT was not 20 then assign PVT = 20 If COVERTYPE = 3310 and PVT was not 20 then assign PVT = 20 If COVERTYPE = 3350 and PVT was not 20 then assign PVT = 20 If COVERTYPE = 3510 and PVT was not 20 then assign PVT = 20 If COVERTYPE = 4140 and PVT = 29 then assign PVT = 10 If COVERTYPE = 4203 and PVT = 29 then assign PVT = 11 If COVERTYPE = 4205 and PVT = 29 then assign PVT = 8 If COVERTYPE = 4212 and PVT = 29 then assign PVT = 11 If COVERTYPE = 4214 and PVT = 29 then assign PVT = 8 If COVERTYPE = 4223 and PVT = 29 then assign PVT = 11 If COVERTYPE = 4290 and PVT = 29 then assign PVT = 11 If COVERTYPE = 4300 and PVT = 29 then assign PVT = 11 If COVERTYPE = 5000 and PVT was not 28 then assign PVT = 28 If COVERTYPE = 6110 and PVT = 1/2/14/33 then assign PVT = 40 If COVERTYPE = 6110 and PVT = 7/8/9/11/15/16/17/18/20/21/23/24/25/29/30 then assign PVT = 38 If COVERTYPE = 6110 and PVT = 6/10/22/31/35 then assign PVT = 39 If COVERTYPE = 6120 and PVT was not 29 then assign PVT = 29 If COVERTYPE = 6130 and PVT = 1/2/14/33 then assign PVT = 40 If COVERTYPE = 6110 and PVT = 7/8/9/11/15/16/17/18/20/21/23/24/25/29/30 then assign PVT = 38 If COVERTYPE = 6110 and PVT = 6/10/22/31/35 then assign PVT = 39 If COVERTYPE = 6200 and PVT was not 18 then assign PVT = 18 If COVERTYPE = 6300 and PVT was not 25 then assign PVT = 25 If COVERTYPE = 6400 and PVT was not 25 then assign PVT = 25 If COVERTYPE = 7300 and PVT was not 19 then assign PVT = 19 If COVERTYPE = 7500 and PVT was not 37 then assign PVT = 37 If COVERTYPE = 7600 and PVT = 0/13/16 then assign PVT = 19 If COVERTYPE = 7800 and PVT was not 37 then assign PVT = 37 If COVERTYPE = 8100 and PVT = 0/1/2/16/17/18/19/33 then assign PVT = 14 If COVERTYPE = 9100 and PVT = 6/11 then assign PVT = 14
There were 370 different COVERTYPE recode rules. Please see COVERTYPE_PVT_RECODE source for information about recode table.
Non-combustible types:
COVERTYPE Fuel Model 1100 Urban 99 5000 Water 99 7300 Rock 99 7500 Mines, Quarries, Gravel Pits 99 8100 Alpine Tundra/Meadows 99 9100 Snowfields or ice 99
Grassland types:
COVERTYPE Fuel Model 2010 Agriculture lands - Dry 1 2020 Agriculture lands - Irrigated 1 3100 Upland Grasslands 1 3110 Altered Herbaceous 1 3130 Very Low Cover Grasslands 1 3150 Low/Moderate Cover Grasslands 1 3170 Moderate/High Cover Grasslands 1 3180 Montane Parklands & Subalp Md 1 6200 Gramminoid and Forb Riparian 1
Shrubland types:
COVERTYPE Fuel Model 3200 Mixed Mesic Shrublands and SIZECLASS = 0 and CANOPY = 1/2 1 and SIZECLASS = 0 and CANOPY = 3/4 5 and SIZECLASS = 5 and CANOPY = 1/2/3 1 and SIZECLASS = 6/7 5 3300 Mixed Xeric Shrubs 5 3309 Silver Sage 5 3310 Salt-Desert Shrub/Dry Salt Flat 5 3350 Sagebrush 5 3510 Mesic Shrub-Grassland Assoc 5 3520 Xeric Shrub-Grassland Assoc 5 6300 Shrub Riparian 5 6400 Mixed Riparian 5
Broadleaf forest types:
COVERTYPE Fuel Model 4140 Mixed Broadleaf Forest and SIZECLASS = 0 8 and SIZECLASS = 1/2 5 and SIZECLASS = 3/4 8 4300 Mixed Broadleaf & Conifer Forest and SIZECLASS = 0 8 and SIZECLASS = 1 5 and SIZECLASS = 2/3/4 8 6120 Broadleaf Riparian and SIZECLASS = 0 8 and SIZECLASS = 1/2 5 and SIZECLASS = 3/4 8 6130 Mixed Broadleaf & Conifer Riparian and SIZECLASS = 0 8 and SIZECLASS = 1/2 5 and SIZECLASS = 3/4 8
Conifer forest types:
COVERTYPE Fuel Model 4000 Low Density Xeric Forest 1 4200 Single Conifer Species Forest and SIZECLASS = 0 8 and SIZECLASS = 1 5 and SIZECLASS = 2 8 and SIZECLASS = 3/4 10 4203 Lodgepole Pine and SIZECLASS = 0 8 and SIZECLASS = 1 5 and SIZECLASS = 2 8 and SIZECLASS = 3/4 10 4205 Limber Pine and CANOPY = 1/2 1 and CANOPY = 3/4 and SIZECLASS = 0 8 and CANOPY = 3/4 and SIZECLASS = 1 5 and CANOPY = 3/4 and SIZECLASS = 2 8 and CANOPY = 3/4 and SIZECLASS = 3 8 and CANOPY = 3/4 and SIZECLASS = 4 10 4206 Ponderosa Pine and CANOPY = 1/2 1 and CANOPY = 3/4 and SIZECLASS = 0 9 and CANOPY = 3/4 and SIZECLASS = 1 5 and CANOPY = 3/4 and SIZECLASS = 2/3 9 and CANOPY = 3/4 and SIZECLASS = 4 10 4207 Grand Fir and SIZECLASS = 0 8 and SIZECLASS = 1 5 and SIZECLASS = 2 8 and SIZECLASS = 3/4 10 4210 Western Red Cedar and SIZECLASS = 0 8 and SIZECLASS = 1 5 and SIZECLASS = 2 8 and SIZECLASS = 3 8 and SIZECLASS = 4 10 4211 Western Hemlock and SIZECLASS = 0 8 and SIZECLASS = 1 5 and SIZECLASS = 2 8 and SIZECLASS = 3 8 and SIZECLASS = 4 10 4212 Douglas-fir and SIZECLASS = 1 5 and PVT = 0 and SIZECLASS = 0 8 PVT = 2/8/14/15/19/33/36/37 and SIZECLASS = 0 8 and SIZECLASS = 2 2 and SIZECLASS = 3 8 and SIZECLASS = 4 8 PVT = 1/6/7/9/10/11/12/27/28/29/30/31/35 and SIZECLASS = 0 8 and SIZECLASS = 2/3 8 and SIZECLASS = 4 8 4214 Rocky Mountain Juniper and CANOPY = 1/2 1 and CANOPY = 3/4 5 4215 Western Larch and SIZECLASS = 0 8 and SIZECLASS = 1 5 and SIZECLASS = 2 8 and SIZECLASS = 3 8 and SIZECLASS = 4 10 4216 Utah Juniper and CANOPY = 1/2 1 and CANOPY = 3/4 5 4223 Douglas-fir/Lodgepole Pine and SIZECLASS = 1 5 and PVT = 0 and SIZECLASS = 0 8 PVT = 2/8/14/15/19/33/36/37 and SIZECLASS = 0 8 and SIZECLASS = 2 2 and SIZECLASS = 3 8 and SIZECLASS = 4 8 PVT = 1/6/7/9/10/11/12/27/28/29/30/31/35 and SIZECLASS = 0 8 and SIZECLASS = 2/3 8 and SIZECLASS = 4 8 4260 Mixed Whitebark Pine Forest and CANOPY = 1/2 1 and CANOPY = 3/4 and SIZECLASS = 0 8 and CANOPY = 3/4 and SIZECLASS = 1 5 and CANOPY = 3/4 and SIZECLASS = 2 8 and CANOPY = 3/4 and SIZECLASS = 3 8 and CANOPY = 3/4 and SIZECLASS = 4 10 4270 Mixed Subalpine Forest and SIZECLASS = 1 5 and PVT = 0 and SIZECLASS = 0 10 PVT = 2/8/9/12/14/15/19/27/28/37 and SIZECLASS = 0/2 8 and SIZECLASS = 3/4 8 PVT = 1/6/7/10/11/30/31/35 and SIZECLASS = 0/2/3/4 10 4280 Mixed Mesic Forest and SIZECLASS = 0 8 and SIZECLASS = 1 5 and SIZECLASS = 2 8 and SIZECLASS = 3 8 and SIZECLASS = 4 10 4290 Mixed Xeric Forest and SIZECLASS = 1 5 and PVT = 0 and SIZECLASS = 0 8 PVT = 2/8/14/15/19/33/36/37 and SIZECLASS = 0 8 and SIZECLASS = 2 2 and SIZECLASS = 3 8 and SIZECLASS = 4 8 PVT = 1/6/7/9/10/11/12/27/28/29/30/31/35 and SIZECLASS = 0 8 and SIZECLASS = 2/3 8 and SIZECLASS = 4 8 6110 Conifer Riparian and SIZECLASS = 0 8 and SIZECLASS = 1 5 and SIZECLASS = 2 8 and SIZECLASS = 3 8 and SIZECLASS = 4 10
Other lifeform classes:
COVERTYPE Fuel Model 4400 Standing Burnt Forest 5 7600 Badlands 1 7604 Missouri Breaks 1 7800 Mixed Barren Sites 5
The two remaining COVERTYPES, Cloud and Cloud Shadow, could not be assigned fuel models because they offered insufficient data.
COVERTYPE Fuel Model 9800 Cloud -1 9900 Cloud shadow -1
In all, six new attributes were created. Attribute values were obtained from BEHAVE (v.4.4,; Burn Subsystem, FIRE1W program, Direct Module) for 84 combinations of FUELMOD (n = 7), SLOPE (n = 4), and ASPECT (n = 3; SW, NE, and all others including Flat). Moisture and wind inputs were held constant and derived as described below. Once BEHAVE runs were complete, the outputs were assembled into lookup tables for each variable, fire rate of spread, flame length, and fireline intensity. An ArcInfo AML was then used to populate those attributes (SPREAD, FLAME, and FIRELINE) for each unique combination in the FIRERISK grid. In the Tables module, values for those attributes were then roughly split into thirds to create the new attributes SPREADCLASS, FLAMECLASS, and FIRELINECLASS. See attribute definitions for further details.
BEHAVE FIRE1 Direct Inputs
The solar energy inputs were based on the following aspect classes and were related to fuel moisture ratings for 1 hr, 10 hr, 100 hr, live herbaceous (LH), and live woody (LW). Input values for these five fuel moisture ratings are summarized in the table below. They are based on the average of the lowest, highest and mean values recorded on 8/23/00 and 8/24/00 at stations within zones across the Northern Region. Averages of the lowest fuel moisture ratings were applied to the "high" solar energy sites (SW aspect); averages of the mean readings for each fuel moisture rating were applied to moderate solar energy sites (F, N, E, SE, S, W, NW aspects); and averages of the highest fuel moisture readings were applied to low solar energy sites (NE aspect).
Solar Fuel Moisture Energy Aspect(s) 1 hr 10 hr 100 hr LH LW
High SW 4 5 6 30 58 Mod F/N/E/SE/S/W/NW 6 7 7 40 63 Low NE 8 9 11 60 82
All BEHAVE FIRE1 runs assumed a constant upslope wind of 15 mi/hr at 20 feet above ground. The wind speed at mid-flame height was calculated for each Fuel Model by applying the following adjustments (see Andrews and Chase 1989):
Fuel Model Adjustment Factor Mid-flame Wind Speed (mi/hr) 1 .4 6 2 .3 4.5 5 .4 6 6 .4 6 8 .2 3 9 .2 3 10 .1 1.5
Not assigned to cover type 4400, Standing Burnt Forest, for any project.
May be particularly problematic for riparian cover types within the extent of SILC-1 data; use with caution. Because SILC-1 riparian data were mapped in a separate database, then pasted back into the COVERTYPE grid for this product, riparian cells have inherited CANOPY values from the underlying SILC-1 polygons they replace. Although the resulting COVERTYPE/CANOPY combinations have been made consistent by riparian lifeform, and thus are logical (e.g., canopy class 2 assigned to riparian broadleaf forest), the two attributes still come from separate classifications with different polygon structures. Riparian polygons may represent distinct inclusions within much larger SILC-1 polygons, meaning that CANOPY may not accurately represent a given riparian site.
One of seven input variables used to define unique combinations (i.e., one of seven grids input to the COMBINE function).
Not assigned to cover types 4000, Low Density Xeric Forest, and 4400, Standing Burnt Forest, for any project.
May be particularly problematic for riparian cover types within the extent of SILC-1 data; use with caution. Because SILC-1 riparian data were mapped in a separate database, then pasted back into the COVERTYPE grid for this product, riparian cells have inherited SIZECLASS values from the underlying SILC-1 polygons they replace. Although the resulting COVERTYPE/SIZECLASS combinations have been made consistent by riparian lifeform, and thus are logical (e.g., size class 3 assigned to riparian broadleaf forest), the two attributes still come from separate classifications with different polygon structures. Riparian polygons may represent distinct inclusions within much larger SILC-1 polygons, meaning that SIZECLASS may not accurately represent a given riparian site.
Again, SIZECLASS was not mapped for EMONT, but is important for defining fuel models and other anticipated applications. Thus, within the EMONT area, for the two TM scenes (P38/R27 and P38/R28) that had been classified for both EMONT and SILC-1, we used COVERTYPE from the more recent EMONT project (with improved classification techniques), but SIZECLASS for forest types from the SILC-1 project. See processing steps for details.
One of seven input variables used to define unique combinations (i.e., one of seven grids input to the COMBINE function).