USDA Forest Service, Remote Sensing Applications Center,

Transcription

1 Deriving the BARC from Satellite Imagery Demonstration

2 Deriving the BARC from Satellite Imagery French Fire 2004 BARC Dataset Rodeo Chediski Fire 2002 Landsat 7 ETM + Imagery Step 1: Pre-processing Step 2: Modeling Step 3: Post-processing and Delivery

3 1: Pre-processing - Import Raw data come in LPGS format from the USGS Using ERDAS Imagine, we import the imagery into *.img format

4 1: Pre-processing Atmospheric Correction Accounting for sun angle and other atmospheric variables, we perform a top-of-atmospheric reflectance correction on Landsat imagery.

5 1: Pre-processing Terrain Correct Not necessary for Landsat imagery

6 1: Pre-processing - Subset We create the BARC on a square subset

7 2: Modeling Normalized Burn Ratio (NBR) NBR = (NIR Mid IR) / (NIR + Mid IR) Differenced Normalized Burn Ratio (dnbr) dnbr = Pre NBR Post NBR

8 3: Post-processing - Reproject We do all our processing in Albers Conical Equal Area projection. We then reproject the delivered products to the projection needed by the end user.

9 3: Post-processing Coloring We color all the classes as an initial classification for end users Green = Unburned / Very Low Aquamarine = Low Yellow = Moderate Red = High

10 3: Delivery FTP Fires with Forest Service BAER teams are posted to the FSWEB FTP site. Interagency BAER teams can find their data on our web site ( Included on the FTP site are preand postfire image subsets as well as the BARC datasets. Each dataset is in raster format.

11 3: Delivery s are sent out as soon as data is posted to FTP site. Note the cell phone number. We are available on weekends and holidays via the cell phone.

12 Editing the BARC to Create a Map of BURN SEVERITY

13 ALWAYS REMEMBER YOUR KEY OBJECTIVE: To quickly develop an accurate map of: Soil burn severity (soil scientist) Vegetation burn severity (forester) Watershed response (hydrologist) For use in emergency assessment analyses of: erosion potential flood/runoff potential debris flow potential forest mortality Threats to: Life Property Cultural and Natural Resources

14 BAER Team Objective Develop a Rehabilitation Plan Within 10 Days Inventory T&E species habitat affected Evaluate artifacts and cultural resources Predict runoff, flooding, threats to water quality Determine erosion potential, threats to soil productivity Prepare timber salvage plans and estimate reforestation needs

15 The BARC is NOT a Burn Severity Map! BARC = Burned Area Reflectance Classification (absolute biomass removal as indicator) The BARC BECOMES a Burn Severity Map (e.g., soil or veg) only AFTER field verification and edits are completed

16 The BARC does give us a HUGE boost in quickly developing an accurate Burn Severity Map Viveash Fire Santa Fe NF 2000

17 Field and aerial verification are still needed! Fly with image, not BARC (less bias) Make notes directly on image (H, M, L, M+, H-, etc) Compare to BARC back at BAER Den Compare to ground notes Adjust BARC as needed

18 BARC4 vs. BARC256 BARC4 Already lumped into four classes (no need to reclassify) Cannot be systematically adjusted Works best in forested environments Does not always accurately depict burn severity (certain vegetation communities) BARC256 Up to 256 classes Color-coded into four colors Easy to systematically adjust Allows for field adjustments to thresholds in a variety of environments Areas of very rocky soils, stark geologic contacts, and large exposed rock areas can still skew classes

19 Example of Adjusted BARC256 PIRU FIRE

20 DEMO BARC Pre-processing

21 Pre-processing the BARC to Create a Map of BURN SEVERITY Exercise

22 DEMO Systematic Editing (Threshold Tweaking)

23 Systematic Editing of the BARC to Create a Map of BURN SEVERITY Exercise

24 DEMO Locational Editing (Problem area fixes)

25 Locational Editing of the BARC to Create a Map of BURN SEVERITY Exercise

26 The End