Disseminating Fire Weather/Fire Danger Forecasts through a Web GIS Andrew Wilson Riverside Fire Lab USDA Forest Service
Hawaii Fire Danger System Supporters Hawaii Department of Forestry & Wildlife Pacific Disaster Center Fire Consortia For Advanced Modeling of Meteorology and Smoke (FCAMMS) USDA Forest Service Riverside Fire Lab Maui High Performance Computing Center National Weather Service Environmental Systems Research Institute
Clients: Users US Army HI Department of Forestry And Wildlife National Park Service County/City Fire Departments Civil Defense
Hawaii Fire Danger Project 1970 s tabular data Point Forecasts Stations weren t where resource concerns were Many hands, many steps Renewed interest with increased accessibility
First Output (1974) FOR HAWAII CIVIL DEFENSE HAWAII FIRE DANGER INDEXES FOR 00Z 29 JAN 74 STATION T RH FF PPP FFM IC SC ERC OI BI FLI KOKEE (161) 57 93 2 0 23.4 0 1 0 0 0 0 BARK SANDS (162) 78 57 6.04 8.2 37 7 9 37 2 1 LIHUE (165) 78 65 8.31 31.0 0 0 0 0 0 0 WAIALEE (169) 77 69 4 0 9.9 28 3 8 28 1 0 WHEELER (170) 78 54 8.07 7.6 39 7 10 39 3 1 HONOLULU (182) 75 64 8 0 31.0 0 0 0 0 0 0 FORECAST FOR HAWAII FIRE DANGER INDEXES FOR HAWAII CIVIL DEFENSE (EFFECTIVE 24 HOURS FROM 00Z 29 JAN 74) STATION T RH FF PPP FFM IC SC ERC OI BI FLI KOKEE (161) 60 82 5 0 11.5 19 5 7 19 2 0 LIHUE (165) 78 65 10 0 9.2 31 9 9 31 3 1 WAIALEE (169) 77 73 10 0 10.2 26 8 8 26 2 1 HONOLULU (182) 77 65 10 0 9.2 31 11 9 31 3 1 MOLOKAI AP (186) 78 69 10 0 12.8 14 7 6 14 2 0
Hawaii Fire Danger Project 1970 s tabular data Renewed interest with increased accessibility High resolution weather forecasts Open forecast models Inexpensive, powerful computers Internet mapping
Models: Forecast Sources RSM, MM5, WRF Scalable and spatial Mosaics: National Digital Forecast Database (NDFD) Various sources: human eye, uncertain provenance (source model, resolution)
Mesoscale model MM5 Steps down from global NCEP model to 12, 6, 3, or even? kilometers Accessible software / hardware Increase spatial resolution = Increase temporal resolution Increase in processing time Physics scaling questions
First Level MM5 Domain 27 km
Second Level MM5 Domain 9km
Third Level MM5 Domain 3km
National Digital Forecast Database (NDFD) Downloads twice or more time daily with adjustments potentially hourly for eight forecast periods per day 3-hour forecast intervals Mosaic of different models Current output format: GRIB (surface) - thrifty XML (point) format fat for surfaces! Shape files
Fire Danger Elements Element Fuels Est. potential for: Use Burn Index (BI) f(type) Flame Length Staffing Levels Energy Release Component (ERC) f(type) Energy/area (BTU/ft 2 ) Type of suppression Ignition Component (IC) 1-hour Ease of starting Pre-positioning units Spread Component (SC) f(type) Rate of Spread Need for speed to suppress Keetch-Byram Drought Index (KBDI) No Upper Soil Dryness Seasonal (slow reaction) index
National Fire Danger Rating System Structure (cf. H. Reinbold) Calculated Input Output Site Description Fuel Model Slope Class Live Fuel Types Climate Class Average Annual Precipitation KBDI (88) Latitude Drought Fuel Ignition Component IC 1300 LST Observation Relative Humidity Temperature Cloudiness Wind Speed Fuel Stick Moisture 24-Hour Observations Precipitation Max/Min RH Amount Min Temp Maximum Temperature Precipitation Duration Season Code & Greenness Factor (88) 1-hr FM 10-hr FM 100-hr FM 1000-hr FM Contribution of dead FM to SC Spread Component SC Contribution of dead FM to ERC Burning Index BI Carryover Fuel Moistures (FM) 100-Hour (88) 1000-Hour Periodic Measurements Live Woody FM Live FM Energy Release Component ERC
Fire Danger Input Fuels Mapping by WFAS, HI-DOFAW, PDC Mean Annual Precipitation Slope Weather! Temperature, humidity, wind, precipitation, cloud cover Temporal (24-48 forecast periods @ 1-3 hours) Dynamic (updates daily or more often)
Display Constraints Weather and fire danger surfaces = rasters Display, Visualization Transparency Speed Key elements sent to tables = points Software raster query problems Multiple elements, one query Flat table and views of query-able points
HFDRS Processing MS-SQL/sde IMS MM5 NDFD Ascii FLT N F D R S SDERaster SDE2Table Raster Point View image query IMS perl fortran python SQL java/js/xml
Step One Data to Calculator Get data from meteorology sources and formats to NFDRS calculator degrib converts Grib files to Floating-point flat files (FLT) GDAL project and clip NFDRS Calculations on pixel-by-pixel basis perl scripts convert, check, massage data
HFDRS Processing MS-SQL/sde IMS MM5 NDFD Ascii FLT N F D R S SDERaster SDE2Table Raster Point View image query IMS perl fortran python SQL java/js/xml
Step Two Results to Map Service Fire Danger & Weather forecasts are FLT (float) files Rasters Convert FLT to BSQ SDERASTER Points/tables/views Convert to dbf SDETABLE Python, perl, C
HFDRS Processing MS-SQL/sde IMS MM5 NDFD Ascii FLT N F D R S SDERaster SDE2Table Raster Point View image query IMS perl fortran python SQL java/js/xml
Step Three - Display IMS and other applications Philosophies: Keep it simple Make temporal aspect manageable Java, JavaScript, XML, HTML,.
HFDRS Display Activities Animation Raster Point View image query IMS Images/ Point Queries Fire Danger Forecast Web Pages (interactive maps) Hydrograms RRA Observations Validation
http://fireweb.pdc.org
Current Environment NDFD NWS FTP site MM5 MHPCC - Super Computer UH (Dell Cluster), UCSB, DRI Fire Web ArcSDE 8.3, SQLServer 2000, ArcIMS 4.0.1 Apache/Tomcat java, javascript Python 2.3, perl 1.7, degrib 1.7,. Single Processor 2.8 Ghz Intel PC
Scalability Current prototype system Multiple instances: Hawaii (PDC-Maui), Southern, Northern California Multiple tiers: Separate crunching, database and web services Multiple processes for each service Central Services multiple sites mutliple tiers
Support User Community: Hawaii Department of Forestry & Wildlife Big Island Wildfire Coordinating Group On-line E-mail (direct links and to forums) Development Web Log Cooperators: Pacific Disaster Center DOFAW USFS MHPCC
Future Continue to respond to user needs Adapt to advances in GIS technologies: Raster handling Temporal display and storage Open source, NFDRS-in-a-box Geospatial Explorers Discover differences between reality and whitepapers (scaling, portability)
http://fireweb.pdc.org Andrew Wilson Resource Information Specialist Riverside Fire Lab aewilson@fs.fed.us