Monitoring Vegetation Regeneration after Wildfire

Slides:



Advertisements
Similar presentations
Lab for Remote Sensing Hydrology and Spatial Modeling Dept. of Bioenvironmental Systems Engineering, NTU Satellite Remote Sensing for Land-Use/Land-Cover.
Advertisements

(Multi-model crop yield estimates)
From Ignition to Spread Wildland Fire Forecasting and Color Maps Managing fire on populated forest landscapes October , 2013 Banff International.
Satellite data products to support climate modelling: Phenology & Snow Cover Kristin Böttcher, Sari Metsämäki, Olli-Pekka Mattila, Mikko Kervinen, Mika.
Application of Earth Observation data in Agricultural drought monitoring Farai Kuri University of Zimbabwe.
USDA FOREST SERVICE Technology & Development Program National Soil Quality Monitoring Field Guide.
Mapping Burn Severity. Burned Area Reflectance Classification (BARC)
USDA Forest Service, Remote Sensing Applications Center, FSWeb: WWW: LANCE User Working Group.
1 USDA Forest Service Hurricane Preparedness Data Sets Developed Southern Wildland Fire Risk Assessment GIS data sets (infrastructure data) for potential.
Remote Sensing of Fire Severity in Interior Alaska (some of the time) Dave Verbyla Dept. of Forest Sciences, UAF Dave Verbyla.
Prevention - Containment - Safety Unmanned Aerial System Support to the Pikes Peak Wildfire Protection Partners.
Remote Sensing – Fire Weather Product Presentation
Current applications of GIS in Wildfire. Remote Sensing Used for a wide array of projects – Satellites Vegetation greeness used to calculate post fire.
USDA Forest Service Remote Sensing Applications Center Forest Inventory and Analysis FIA, Remote Sensing and REDD Remote Sensing Applications and the Forest.
USDA Forest Service Remote Sensing Applications Center Operational Remote Sensing Support Programs USDA Imagery Planning and Coordination December 2-4,
DROUGHT MONITORING THROUGH THE USE OF MODIS SATELLITE Amy Anderson, Curt Johnson, Dave Prevedel, & Russ Reading.
Accuracy Assessment. 2 Because it is not practical to test every pixel in the classification image, a representative sample of reference points in the.
Use of Remote Sensing Data for Delineation of Wildland Fire Effects
Updating Erosion Hazard Ratings in a Post-fire Assessment A GIS Tool for Soil Scientists.
USDA Forest Service Remote Sensing Applications Center Forest Inventory and Analysis New Technology How is FIA integrating new technological developments.
USDA Forest Service Remote Sensing Applications Center Operational Satellite-Based Remote Sensing Fire Support Programs WRAP/FETS Project Meeting.
MODIS: Moderate-resolution Imaging Spectroradiometer National-Scale Remote Sensing Imagery for Natural Resource Applications Mark Finco Remote Sensing.
Fire Severity and Bark Beetle Infestation in California Elizabeth Pascale, Courtney Lewis, and Rebecca Fisher.
The role of remote sensing in Climate Change Mitigation and Adaptation.
Geographic Techniques for Teachers GCU 674. Today’s Challenges Local, National, Global Environmental, Social, Political, Economic … What is done to help.
PREFER 1 st Annual Review Meeting, 5-6 Dec 2013, Milano-Italy PREFER WP 3.2 Information support to Recovery/Reconstruction Task 7 Damage Severity Map PREFER.
Abstract: Dryland river basins frequently support both irrigated agriculture and riparian vegetation and remote sensing methods are needed to monitor.
Analysis of Forest Fire Disturbance in the Western US Using Landsat Time Series Images: Haley Wicklein Advisors: Jim Collatz and Jeff Masek,
Translation to the New TCO Panel Beverly Law Prof. Global Change Forest Science Science Chair, AmeriFlux Network Oregon State University.
USDA Forest Service, Remote Sensing Applications Center, FSWeb: WWW: USFS Citation Aircraft.
PREFER 1 st Annual Review Meeting, 5-6 Dec 2013, Milano-Italy PREFER WP 3.2, Task 7.
Fuel Treatment Making a Difference Steve Harbert Tim Rich Burn Boss Refresher February Burn Boss Refresher 2009.
USDA Forest Service, Remote Sensing Applications Center, FSWeb: WWW: AMS Update TFRSAC Fall.
Mapping Vegetation on the Copper River Delta 1 Lidar Strategy Pacific Northwest Region Using LiDAR to Inform NEPA April 23, 2015 BLM Office | Salem, OR.
Texture Metric Comparison of Manual Forest Stand Delineation and Image Segmentation Richard M. Warnick, Ken Brewer, Kevin Megown, Mark Finco, and Brian.
Honeysuckle in the Taconics
SATELLITE AND AERIAL IMAGE DATA, MOBILE COMPUTING, GIS, AND GPS FOR INTEGRATED CROP MANAGEMENT (ICM) Chuck O’Hara, Dan Reynolds, Roger King John Cartwright,
National Agriculture Imagery Program NAIP Informational Meeting USDA Forest Service Benefits from NAIP Cooperation July 19, 2006 Washington DC Bill Belton.
FRONT RANGE CFLRP/SRLCC MONITORING UPDATE 1 st Year Spatial Heterogeneity Results.
Development of Vegetation Indices as Economic Thresholds for Control of Defoliating Insects of Soybean James BoardVijay MakaRandy PriceDina KnightMatthew.
Pollutant Emissions from Large Wildfires in the Western United States Shawn P. Urbanski, Matt C. Reeves, W. M. Hao US Forest Service Rocky Mountain Research.
What is GIS ? A method to visualize, manipulate, analyze, and display spatial data “Smart Maps” linking a database to the map.
LANDSCAPE SCALE MONITORING OF FOREST TREATMENTS AND DISTURBANCE S.E. Sesnie 1,2, A.D. Olsson 1, B.G. Dickson 1, A. Leonard 3, V. Stein Foster 3 and C.
Updated Cover Type Map of Cloquet Forestry Center For Continuous Forest Inventory.
Earth Data Analysis Center at The University of New Mexico Geospatial Knowledge to Support New Mexico EDAC.
U.S. Department of the Interior U.S. Geological Survey October 22, 2015 EROS Fire Science Understanding a Changing Earth.
USDA Forest Service, Remote Sensing Applications Center, FSWeb: WWW: National Geospatial Fire.
S-244 Field Observer & Fire Effects Monitor Fire Effects Monitoring Lesson 2: Methods for Measuring Fire Effects Photo: Marshel Moy.
Optimization Techniques for Natural Resources SEFS 540 / ESRM 490 B Lecture 1 (3/30/2016)
Developing the Vegetation Drought Response Index (VegDRI): Monitoring Vegetation Stress from a Local to National Scale Brian Wardlow National Drought Mitigation.
Fire Regime Condition Class: - Using the FRCC Mapping Tool -
USDA Forest Service, Remote Sensing Applications Center, FSWeb: WWW: USFS AMS Operations Spring.
Mapping Forest Burn Severity Using Non Anniversary Date Satellite Images By: Blake Cobb Renewable Resources Department with Dr. Ramesh Sivanpillai Department.
USDA Forest Service, Remote Sensing Applications Center, FSWeb: WWW: Rapid Disturbance Assessment.
SLM Intervention Impact Assessment Using Remotely Sensed Data: Impact Assessment of the Tana-Beles Integrated Water Resources Development Project WB Land.
Term Project Presentation
VEGA-GEOGLAM Web-based GIS for crop monitoring and decision support in agriculture Evgeniya Elkina, Russian Space Research Institute The GEO-XIII Plenary.
Cold Springs Fire Project
Jesse Minor University of Arizona School of Geography and Development
An Introduction to VegDRI
Shannon Clemens GIS for Water Resources December 2, 2008
Orin J. Hutchinson1 with Dr. Ramesh Sivanpillai2
Meng Lu and Edzer Pebesma
Evaluating Land-Use Classification Methodology Using Landsat Imagery
The GISCO task force “Remote Sensing for Statistics”
Predicting the hydrologic and water quality implications of climate and land use change in forested catchments Dennis P. Lettenmaier Department of Civil.
AGEOS CB ACTIVITIES AGEOS By Aboubakar Mambimba Ndjoungui
PREDICTING THE WEATHER
The Pagami Creek Wildfire
World Bank Land and Poverty Conference March 2018
Presentation transcript:

Monitoring Vegetation Regeneration after Wildfire Jess Clark USFS Remote Sensing Applications Center In cooperation with: Marc Stamer (San Bernardino NF), Kevin Cooper (Los Padres NF), Carolyn Napper (San Dimas T&D), Terri Hogue (UCLA) USDA Forest Service, Remote Sensing Applications Center, FSWeb: http://fsweb.rsac.fs.fed.us WWW: http://www.fs.fed.us/eng/rsac/

Need for Post-fire Monitoring Wildfire Effects BAER Assessments and Treatments Monitoring Requirements Who, how often, for how long, who pays? Values at Risk This slide provides the conceptual framework for this project. After wildfires occur, the severity of the fire is mapped and delivered to BAER teams. They make professional judgments of treatment needs based on the severity map and potential values at risk. After the BAER team leaves, the local forest (district ranger) is responsible for vegetation monitoring; however, the BAER team is still allowed to monitor the treatments they implemented and can use BAER funds to re-apply treatments for up to three years if the need persists. At some point, district rangers get pressure from the public to remove invasive treatments (like the k-rail shown in the picture above) but rangers are unsure whether the risk has fallen sufficiently where they can confidently allow treatment removal. USDA Forest Service, Remote Sensing Applications Center, http://fsweb.rsac.fs.fed.us Photo credit: Robert Leeper

Severity mapping (NBR / dNBR) Monitoring (NDVI / EVI) Role of Remote Sensing Severity mapping (NBR / dNBR) Monitoring (NDVI / EVI) Predictive Modeling (Regression) Decision Support Tools USDA Forest Service, Remote Sensing Applications Center, http://fsweb.rsac.fs.fed.us

Role of Remote Sensing Vegetation Indices NDVI, EVI, NBR All are image derivatives that take advantage of the spectral richness in Landsat to discern differences in plant vigor and soil moisture content. USDA Forest Service, Remote Sensing Applications Center, http://fsweb.rsac.fs.fed.us

Snapshot in time (NBR / dNBR) Severity Mapping Snapshot in time (NBR / dNBR) e.g., BAER, RAVG, MTBS Classes / protocols well defined Using remote sensing technology is well defined and understood for mapping post-fire effects (e.g., severity). BAER, RAVG, and MTBS are three examples of programs national in scope that do this. People understand the value and limitations in each of these programs and (hopefully) use them appropriately. USDA Forest Service, Remote Sensing Applications Center, http://fsweb.rsac.fs.fed.us

NDVI / EVI for monitoring over time Trends Analysis Current compared to pre-fire condition This first example shows a possible method, which is to track the NDVI trend of zones / polygons over time. This approach will allow someone to see whether certain zones have recovered to pre-fire greenness condition. In this example, you can see an immediate green-up in some areas by early May the following year and then another drastic green-up during May-June 2011. Some zones, however, exhibit a much slower recovery. USDA Forest Service, Remote Sensing Applications Center, http://fsweb.rsac.fs.fed.us

NDVI / EVI for monitoring over time Hybrid Static Cover Layer Pixel values represent actual cover values This approach is a bit of a hybrid approach. This approach is a composite of all images available during a growing season with the maximum NDVI value chosen and displayed. This allows us to account for vegetation that greens in May but browns in August. While this is the approach we have carried through, it isn’t the panacea. Maybe an area was relatively barren before the fire occurred; this layer does not account for change. USDA Forest Service, Remote Sensing Applications Center, http://fsweb.rsac.fs.fed.us

Project Objectives Assess effectiveness of remote sensing to monitor vegetation regeneration Methods: Field data collection Remote sensing based observations Correlation analysis / predictive modeling Application USDA Forest Service, Remote Sensing Applications Center, http://fsweb.rsac.fs.fed.us

Locations Six fires: Old (2003), Am. River Complex (2008), La Brea (2009), Station (2009), Bull (2010), Canyon (2010) Am. River Complex All fires were in California due to cooperating partners: San Bernardino NF, Los Padres NF, San Dimas T&D, and UCLA. Bull / Canyon La Brea Station Old USDA Forest Service, Remote Sensing Applications Center, http://fsweb.rsac.fs.fed.us

Methods – Field Data Collection Pole-mast photography “Plot” = area of homogeneous ground condition Between 4 and 10 photos per plot Photos interpreted later Landsat pixels represent 900 sq. meters of ground on the earth, so we tried to focus on areas that would encompass at least 2 Landsat pixels. Anything less homogeneous would be adversely affected by spatial error in the Landsat imagery as well as GPS locations. USDA Forest Service, Remote Sensing Applications Center, http://fsweb.rsac.fs.fed.us

Methods – Photo Interpretation Pole-mast photography Each photo interpreted cover vs. no-cover Stats summarized by plot (4 to 10 photos) We interpreted 600 points per plot and we calculated that we only “needed” 400 to properly meet statistics assumptions. To achieve a 95% confidence interval, we need to interpret at least 400 points per plot, regardless of the number of photos acquired in the plot. 95% Confidence Interval = 2 * sqrt (p(1-p)/n) 95% CI = +/- 0.05; p = 0.5 (worst case scenario; equal chance of the ground exhibiting cover or bare ground) Solved for n; n = 400 We decided to collect 600 points so we could ignore or throw out points that intersected anomalous things in the photos (humans, monopod, etc.) Cover is defined as living plant material, those that either have sprouted since the fire or those that did not die in the initial fire. One major caveat here is in plots that have existing canopies – high (living) canopy cover but little understory. The monopod setup does not get above the canopy, so we’re poorly suited to deal with that. USDA Forest Service, Remote Sensing Applications Center, http://fsweb.rsac.fs.fed.us

Methods – Satellite Imagery Imagery collected pre- and post-fire NDVI / EVI creation Pixel values summarized by plot areas I used normal Landsat delivery methods (USGS-GloVis) as well as a new approach (Google Earth Engine). Google has the entire Landsat data record spinning on their servers and allow for “clip-zip-ship” functionality as well as cloud-based computing and generation of a number of indices. I downloaded image chips around the fires and their accompanying vegetation indices. I then created zonal stats based on plot locations (all pixels that intersected the plot) and summarized then in tabular and graphical form. USDA Forest Service, Remote Sensing Applications Center, http://fsweb.rsac.fs.fed.us

NDVI and EVI both showed relatively high correlation to ground cover Results NDVI and EVI both showed relatively high correlation to ground cover Leads to application of thresholds for thematic output Both EVI and NDVI performed fairly well in terms of the big picture. The NBR performed very poorly and was not included in further analysis. Armed with these equations, I was able to apply the algorithm to new fires that had no field data and make a map of predicted ground cover. % Ground Cover % Ground Cover EVI Value NDVI Value USDA Forest Service, Remote Sensing Applications Center, http://fsweb.rsac.fs.fed.us

Discussion and Limitations Less than perfect field data collection Clumping of fires, not values Some historic (recovered) fires and some current (still black) fires Muddy results in critical data range Poor linear function in the 0.2 – 0.35 NDVI data range Even with 6 different fires, we could minor clumping of fires rather than values; for example, Bull/Canyon tended to be the lower end of the scale (newer fire) while the Old Fire represented much of the higher end of the scale. Also, the correlation in what I would consider the critical data range (40% cover) is poor. Overall, the correlation is strong, but locally, there are plenty of weaknesses. USDA Forest Service, Remote Sensing Applications Center, http://fsweb.rsac.fs.fed.us

Application for New Fires Time series imagery for new fires Horseshoe 2, Monument, Schultz NDVI and EVI cover map Available for evaluation Schultz Phoenix Again, I used Google Earth Engine to gather new Landsat imagery (all cloud-free imagery acquired after the fire) and stacked the maximum NDVI / EVI value into a single composite. Tucson Horseshoe 2 Monument USDA Forest Service, Remote Sensing Applications Center, http://fsweb.rsac.fs.fed.us

Schultz (2010) – Applied Results The Schultz Fire burned outside Flagstaff during 2010. These are still preliminary results with no proper field validation. USDA Forest Service, Remote Sensing Applications Center, http://fsweb.rsac.fs.fed.us

Horseshoe 2 (2011) – Applied Results Horseshoe 2 burned south of Phoenix last year. This fire had considerable green-up in portions. USDA Forest Service, Remote Sensing Applications Center, http://fsweb.rsac.fs.fed.us

Monument (2011) – Applied Results Monument burned last year and is the fire we’re going to visit tomorrow. I will have a few print outs to do a quick qualitative assessment while we’re in the field. USDA Forest Service, Remote Sensing Applications Center, http://fsweb.rsac.fs.fed.us

Tool for resource managers / line officers Decision Support Tool Tool for resource managers / line officers When has the risk sufficiently lessened? This project includes the creation of a decision support tool that will help resource managers better answer the question, “Can we open the Graham Harbor Campground yet?” The graphic shows the mountains along Lake Chelan in Washington. The Pot Peak Fire (2004) burned 47,000+ acres but the Graham Harbor Campground was saved. However, the entire hillslope above the campground was burned fairly severely. USDA Forest Service, Remote Sensing Applications Center, http://fsweb.rsac.fs.fed.us

Post-fire Watershed Planning Decision Support Process Decision Support Tool Post-fire Watershed Planning Decision Support Process Define critical values Define AOI Acquire imagery and VI Summarize VI by AOI Probability of damage Identify risk More ESR work needed? After the predictive modeling is performed and validated, it is used as an input to a decision support tool. The district ranger or other land manager will use the flowchart shown above to evaluate whether watersheds need additional attention or whether their recovery has sufficiently reduced the risk of future damage (to life and property). USDA Forest Service, Remote Sensing Applications Center, http://fsweb.rsac.fs.fed.us

Jess Clark jesstclark@fs.fed.us 801-975-3769 Comments / Questions? Jess Clark jesstclark@fs.fed.us 801-975-3769 USDA Forest Service, Remote Sensing Applications Center, FSWeb: http://fsweb.rsac.fs.fed.us WWW: http://www.fs.fed.us/eng/rsac/