Oregon Health & Air Quality Team Landsat 7 Tutorial NASA DEVELOP Spring 2013 Stephen Quinn (Team Lead) Zachary Toll (Team Lead) Amanda Gray Brittany Parsons Carrea Dye Marlene Lichty
Tutorial Review • GIS Resources and downloading • Creating Indices/Ratios (NDVI, NDMI, NBR, dNBR) • Creating a supervised classification • Creating a Fire Risk Map • Cartography (Map Design) • Creating a Fire Susceptibility Map
GIS Resources • The following are a few resources to use in order to begin selecting what imagery you want • EarthExplorerhttp://earthexplorer.usgs.gov/Landsatand more • GLOVIShttp://glovis.usgs.gov/Landsatand more • Reverb ECHO http://reverb.echo.nasa.gov/ MODIS, CALIPSO, and more • NOAA CLASS http://www.class.ncdc.noaa.gov/saa/products/catSearch VIIRS and more
Earth Explorer • For our project we used EarthExplorer to download our Landsat 7 imagery (You can use which ever satellite or website works best for your project). • EarthExplorer: • Under “Search Criteria” select coordinates and create a red polygon overlaying your AOI (Area Of Interest). To make it easier you can manually click each point on the map. • Ensure your AOI is as small as possible in order to get images that are centered above your AOI • Also select a date range for your images. • We did a pre and post fire map for 2012, so we chose dates as recently as we could after the fire, as well as a year before the fire Select Coordinates Select Date Range
Data Sets • Under Data Sets select “Landsat Archive” • Select Landsat 7 or “L7” by your dates, as well as “L4-5 TM” and “L1-5 MSS” • L4-5 TM is Landsat 4 and 5 Thematic Mapper sensor, while L1-L5 MSS is Landsat’s1-5 Multi Spectral Scanner
Results • Under “Results” you can select the “Show Footprint” to see where the image is, “Browse Overlay” to see the actual image overlaid, and “Download Options” if you like what you see. • You will need to create an account with the USGS to download imagery • If the data is not available, you will have to select the shopping cart and they will email you your imagery in about a day • ALWAYS ENSURE TO DOWNLOAD TIER 1 “Level 1 Product” DATA. Should be >200 MB. Footprint, Overlay, Download, And Shopping Cart
Saving Data • Ensure to save your data to a place where you can easily find and remember it, ideally the desktop
7-Zip File Manager • Search for the 7-Zip File Manager • In the file manager, locate your zipped image folder • Select the file and click the “Extract” button. • Do this step again to ensure all files are unzipped/extracted
Creating Indices/Ratios • Before you begin, you will have to add a toolbox • Right click in while space in the ArcToolbox and import the “DEVELOP_Tools” folder “DN to Reflectance Toolbox”
Creating Indices/Ratios • Find the Landsat DN to Reflectance toolbox which you uploaded and select the tool. Import all individual images you will use to create the Indices/Ratios • For NDVI, NDMI, NBR, and dNBR you will need images from bands 2, 3, 4, 5, and 7. • In “Metadata Text File” selected text file should end in “_MTL.txt” • Of course this step varies with your imagery and sensor you use. Try process of elimination if you have to. • In “Radiance or Reflectance” choose radiance • Save the new reflectance files in a new folder to organize your data. • Ideally name the new folder “ReflectanceImages” Find “Landsat DN to Reflectance tool”
Creating Indices/Ratios • Ensure your new reflectance images are loaded by selecting the “Add Data” button and finding them in their saved folder • Search for Raster Calculator in window search • Calculations for all Indices/Ratios: • (Float(B1-B2)/(B1+B2)) • For NDVI use B3 and B4 • For NDMI use B4 and B5 • For NBR use B4 and B7 • For dNBR use pre and post fire NBR’s • NEVER MANUALLY TYPE IN CALCULATION, use text symbols in calculator! (Loaded) reflectance images
Indices/Ratios • Ensure to save your new indices and ratios into folder to organize them well. • You can then load your NDVI’s, NDMI’s etc. into the scene using the “Add Data” button to look at them. • Voilá! After you complete all the Raster Calculations your Indices/Ratios are finally produced.
Supervised Classifications • Supervised classifications are created with a composite image of three different bands. For this example load up bands 2, 4, and 5 of the reflectance images from a single image. • Under the “Customize” tab click “Extensions” ensuring “Spatial Analyst” is checked and ready to use. • Once loaded, under the “Windows” tab click “Image Analysis” then click the composite button to create a composite image of your three bands. • Ensure Band 2 is shown as blue, Band 4 is shown as green, and Band 5 is shown as red. The image should look like the one below if you have the colors set properly. Composite image button
Supervised Classification • Beside the “Help” tab, right click in a blank space and click “Image Classification”. The tab should look similar to this: • Then click on training sample manager and begin creating polygons of your different classes. The more classes the better. Try to use more than 6 if you can, they should ideally be Water, Urban, Agriculture, Sediments, EvergreenForest, ShrubLand, Grassland etc. • You can also change the color, name and samples in the training sample manager. Once you select many polygons for one class (such as water) use the merge button to make one class for all water polygons. • Ensure to constantly save your Training Samples in case ArcMap crashes! • Once you have created many classes select “Classification” in the “Image Classification” toolbar and select “Interactive Supervised Classification”.
Creating a Fire Risk Map • We used five variables to create our fire risk map • Soil type, fuel cover, proximity to roads, slope, and wild-urban interface
Soil Types • To access Soil Survey Data, navigate to: http://soildatamart.nrcs.usda.gov/ • Click Select State. Click the state you wish to download, i.e. Oregon. • Click Select County. Click the county you wish to download. Click Select Survey Area. • On the next screen, click Download data. • Enter your e-mail address in the Please enter e-mail address window and click Submit Request. • While waiting for the data to be delivered, repeat Steps 2 through 5 to download multiple counties. • Once delivered, open the layers in ArcGIS 10 and reproject the layers using the Project tool into a common projection system. For this study, we used WGS 84 UTM Zone 10N. 8. Once reprojected, merge the layers into the Master_Soil layer by using the Merge tool. 9. You may only be able to merge points with points, lines with lines etc.
Soil Types • Each county has various soil types and soil names. Use the resources on the Soil Data Mart website in order to be able to reclassify the soil names into soil families • For example, the soil families that exist in Deschutes county are Alfisols, Entisols, Histosols, Inceptisols, Spodosols, and Ultisols. • Each polygon must be classified into a soil family. Open the attribute table of the Master_SoilLayer. Click Table Options from the top menu bar, then click Add Field. • Name the new field Soil_Code, change the Type to float Float. Click OK. • Select all the appropriate polygon for one soil family, for example Histosols. Then, click the Show Selected Records icon at the bottom of the screen. • Right-click the heading Soil_Code, then click Field Calculator. Insert the new code integer for each soil family. See table below. • For example, all Histosols should get the code 3. • All polygons that are not a soil type (for example, water) should be classified as 0. • Once all families have been coded, click Geoprocessingfrom the top menu bar, then click the Dissolve tool. • Insert Master_Soil as Input Features. • Redirect the Output Feature Class to your working space and name the dissolved • image as Master_SoilFam. • Check the box next to Soil_Code in the Dissolve_Field(s) window.
Soil Types • Once the new layer is dissolved, open the attribute table to verify that one polygon exists for each the Soil_Code value. • If completed, add a new field in the attribute table called Weight, and change the type to Float (as you did in Step 10). • Establish a weight between 0 to 100 for each soil class based on its combustibility. • For example, we gauged combustibility for each soil type based on historical ground fires and organic content. Refer to table below for example of weight values chosen. • All non-soil type categories remain the value of 0. • Open ArcToolbox, navigate to the following tool: Conversion Tools→ToRaster→Polygon to Raster. • Insert the Master_SoilFam into the Input Features. • Change the Value Field to Weight. • Navigate the Output Raster Dataset to your working folder and name the file Soil_Raster. • Input 30 into the Cellsize window. Leave all other defaults. Click OK. • Open ArcToolbox, navigate to the following tool: Spatial Analyst Tools→MapAlgebra→Raster Calculator. • The equation should read: Float(“Soil_Raster”*0.01) • Navigate the Output Raster to your working folder. • Name the final image Soil_Variable. The soil layer now has membership and will be used in the Fuzzy Overlay after the other layers have been completed.
Fuel Cover • To access LANDFIRE Data, navigate to: http://landfire.cr.usgs.gov/viewer/ • Click zone 10for the region in Oregon of the Pole Creek Fire. • Click the globe icon in the upper left-corner, and click Download Data. • In the Download Data window on the right, expand the LF2008 (Refresh – LF_110) category, followed by the Fuel category. • Check the box next to the layer us_110 40 Scott and Burgan Fire Behavior Fuel Model, i.e. the first option in the listing. • Click the Define Rectangular Download Area for Seamless Data and draw a box around the study area. • In the new window, click the Download button next to the layer titled us_110 40 Scott and Burgan Fire Behavior Fuel Model. • Open this new raster layer in ArcGIS 10, and project into WGS 84 UTM Zone 10N using Project Raster. Name the output rasteras FBFM40_10N. Make sure to download all necessary tiles to cover your study area. For this study, both Dare and Pender counties are within the same USGS tile.
Fuel Cover • Before the weights can be calculated, the layer needs to be scaled to the study area. • To access a County shapefile for North Carolina, navigate to: http://www.fhwa.dot.gov/planning/processes/tools/nhpn/ • Download the County Boundaries for the state. • Open this new layer in ArcGIS 10, andopen the Select by Attributes tool from the Selection tab in the menu bar. • Enter the expression: "NAME" = 'DARE' OR "NAME" = 'PENDER‘ • Click OK. • Right-click the county layer and navigate to Selection→Create Layer From Selected Features. • Reproject the new layer of the selected counties into WGS 84 UTM Zone 10N using the Project tool. Name the output: Cntys_10N. • Open the Polygon to Raster tool and insert the Cntys_10Nlayer as Input Feature. • Leave the defaults for the Cell assignment type and Priority Field. • Set the Value Field to FIPS. • Navigate the Output Raster Dataset to your working folder, and name the image: Cntys_Rast. • Set Cellsize to 30. • Click Environments. Click Processing Extent, and navigate the Extent to the Soil_Variable layer created earlier. Click OK twice.
Fuel Cover 16. Open the Reclassify tool, and set the input raster to Cntys_Rast. • The old values are the previous FIPS numbers. The New Values should all be set to 1. Leave NoData as NoData. • Name the Output Raster as Cnty_Bool. 17. Open the Raster Calculator tool, and input the following expression: Float(“Cntys_Bool”*” FBFM40_10N”). • This step isolates the fuel types within our two counties and eliminates surrounding counties. • Name the Output Raster as Fuel_StyArea. • Open the attribute table of FBFM40_10Nand take note of the FBFM40 column which gives a USGS fuel code for various land cover types. Use the help material on the LANDFIRE website to give a detailed description of each of these categories to be able to assign appropriate weights. Open the Properties of Fuel_StyArealayer, and click Unique Values tocompute the values. • Open the Reclassify tool, set the Input Raster to Fuel_StyArea. Enter the new weight values in the New Values column. See the table below for our chosen weights. Set the Output Raster to your working folder and name the layer Fuel_weight. • Open Raster Calculator, then enter the following expression: Float(“Fuel_Weight”*0.01) • Navigate the Output Raster to your working folder. • Name the final image Fuel_Variable.
Proximity to Roads • To access roads data for desired state, navigate to: http://www.fhwa.dot.gov/planning/processes/tools/nhpn/ • Click to download the ISRN Zip File, which is the Integrated Statewide Road Network. • Open this layer in ArcGIS 10, then reproject the layer into WGS 84 UTM Zone 10N using the Project tool. Call the output Roads_10N. • Open the Dissolve tool from the Geoprocessingmenu bar. • Set the Input Features to Roads_10N. • Navigate the Output Feature Class to the working folder and label the new layer as All_Roads. • Open the Clip tool from the Geoprocessingmenu bar. • Set the Input Features as All_Roads. • Set the Clip Features to Cntys_10N. • Set the Output Features to your working folder, and label the layer as Roads_StyArea. • Open ArcToolbox, navigate to Conversion Tools→ToRaster→Feature to Raster. • Set the Input Feature to Roads_StyArea. • Set Field to FID. • Set the Output Raster to your working folder and label the layer as Roads_Rast. • Set Cellsize to 30. • Click Environments. Click Processing Extent, and navigate the Extent to the Soil_Variable layer created earlier. Click OK twice. • Open the Reclassify tool, and reclass all old values to the new value of 1. Leave NoData as NoData. Label the Output Feature as Road_Master. • Open the Reclassify tool again, and reclass all old values of 1 to NoData and the old value of NoData to the new value of one1. Label the Output Feature as Road_Bool. (This layer will be used in a step to follow.)
Proximity to Roads • To calculate distance from roads, navigate to the following tool: Spatial Analyst Tools→Distance→Euclidean Distance. • Set Input raster of feature source data to Road_Master. • Set Output distance raster to your working folder, and label the layer as Roads_Dist. • Leave all other defaults. Make sure the Output Cell Size is set at 30. Click OK. • Open Raster Calculator, then input the following expression: Float(“Roads_Dist”*”Roads_Bool”*” Cntys_Bool”). • Set Output raster to your working folder, and label the layer as Dist2Roads.tif. • To apply membership, open ArcToolbox and navigate to the following tool: Spatial Analyst Tools→Overlay→Fuzzy Membership. • Set Input Raster to Dist2Roads.tif. • Set the Output Raster to your working folder, and label this layer as Roads_Variable. • Set the Membership Type to Linear. • Set the Minimum to 200 and the Maximum to 30. Click OK. The roads layer now has membership and will be used in the Fuzzy Overlay after the other layers have been completed.
Slope • Open the ASTER Global DEM products that were downloaded at the beginning of the tutorial into ArcGIS 10. There are three tiles for this study. • Navigate to the following tool in ArcToolbox: Data Management Tools→Raster→Raster Dataset →Mosaic To New Raster. • Input all DEM layers. Set Output Location to your working folder. • Set Raster Dataset Name with Extension to DEM_all.tif. • Set Number of Bands to 1. Leave all other defaults. • Open Raster Calculator and input the following expression: Float("DEM_all" *”Cntys_Bool”). • Set Output raster to your working folder, and label the layer as DEM_Sty.tif. • ReprojectDEM_Sty.tif into WGS 84 UTM Zone 10N, and label this layer as DEM_Sty10N.tifusing the Project Raster tool. • Navigate to the following tool in ArcToolbox: Spatial Analyst Tools→Surface→Slope. • Set the Input Raster as DEM_Sty10N. Set the Output Raster to your working folder and label this layer as Slope_Perc. • Set the Output Measurement as Percent. Set the Z factor as 0.5. Click OK.
Slope • To apply membership, navigate to the Fuzzy Membership tool. • Set Input Raster to Slope_Deg. Set the Output Raster to your working folder, and label this layer as Slope_Vari1. • Set the Membership Type to Large. Leave the default midpoint value. (Alter this value in order to get various results for membership.) • Open Raster Calculator and input the following expression: Float(“Slope_Vari1" *”Roads_Bool”). • Set Output raster to your working folder, and label the layer as Slope_Variable. The slopes layer now has membership and will be used in the Fuzzy Overlay after the other layers have been completed.
Wild-Urban Interface • Load the previously created Landsat 7 supervised classification (ensuring you have urban as a class) • If you do not have urban as a class (because there is little to no urban in your study area), this parameter for the Fire Risk Map is not worth using because fires will have little to no interface with the local human population. • Right click on the classification to open its attribute table • Click on “Table Options” and then “Add Field” • Name field “Urban”, type “Float” • Select the urban pixels in the attribute table and click “Show Selected Records” • In the “Editor” toolbar select “Start Editing” • Assign a value of “1” to the urban column then save and stop editing
Wild-Urban Interface • Search for the “Raster to Polygon (Conversion)” tool • Use your classification as the input raster • Use “Value” field option • Then save your new Urban polygon to a good place • You should have a new layer which has several new “urban” polygons • Use the “Editor” toolbar to delete any polygons you know are not properly classified as urban • Once all of your polygons are urban, use the “Polygon to Raster” tool to put the layer back into raster form • Use “GRIDCODE” as value field, “CELL_CENTER” as Cell Assignment Type, and “30” as Cellsize because the cell size needs to match the pixel size of the imagery you are using (in this example, Landsat has 30x30 m spatial resolution. If this doesn’t work try other options as value field and cell assignment type • Search for “Euclidean Distance (Spatial Analysis)” tool • Use “200” as maximum distance and “30” again as Output Cell size. Save to a good place • Your Wild-Urban Interface parameter is complete!
Fire Risk Map • Add all parameters (soils, slope, roads, fuels, wild-urban interface) to ArcMap • Open “Fuzzy Overlay (Spatial Analyst)” tool and input all layers using the “Input Rasters” field • Save to a good place • The Overlay Type field is hit and miss; use the process of elimination to get the best result • Your Fire Risk Map is complete!
Cartography • At the top select “View” Layout View • Ensure your classifications, NDVI’s, Fire Risk Map, etc. is centered in the layout • To insert a nice background, click “Insert” “Data Frame” • Extend the new data frame to the entire background. • Right click and do “Order” “Send to Back” • You can right click on the layer to change the color and border • Feel free to insert a legend, north arrow, scale bar etc. to your map to be more informative of the area • The cartographic elements are up to you and the design you want to present. The next slide shows the final Fire Risk Map we created for our project
Cartography • This map has a title, legend, compass, and scale bar to give descriptions of the area • We used the “National Geographic” basemap and colored our second data frame green to give a more natural element to the map • We intentionally left Bend, Oregon in the bottom right portion of the map to give a frame of reference • Also be sure to list the projection, organization and date the map was created at the very bottom or wherever you see fit • Fire Risk Map is for Descheutescounty for where Pole Creek Fire took place
Fire Susceptibility Map • Ensure your previously created NDVI, NDMI, and NBR are loaded into ArcMap • Use the “Image Analysis” and “Composite Image” button from slide 13 to create a composite • Right click on the layer properties Symbology • Ensure your NDVI is green, NDWI is blue, and NBR is red • Use the steps from slide 14 to create new classifications of your areas • This makes the end user (you) to think hard about what the different colors on the image mean. For example snow should be bright purple, which means there is high amounts of water (NDMI) and low potential for burn (NBR), with no vegetation • Use the color wheel to decide what different colors on your image represent. Forests in March should appear bright aqua, because there is high amounts of vegetation (NDVI)and moisture (NDMI) with low potential for burn (NBR) • Once you have decided on correlated areas, decide on the training samples you feel best represent the specified areas of land cover
Fire Susceptibility Map • Use the Supervised Classification from slide 14, and you should have a somewhat accurate Fire Susceptibility Map • Once again, classifications require many attempts to be accurate, and remember that no classification is perfect. Images are from Pole Creek Fire in Oregon. • NDVI/NDMI/NBR composite image - Classified image/Susceptibility Map