Notes: Notes: Notes: Notes:

Transcription

1 NR406 GIS Applications in Fire Ecology & Management Lesson 2 - Overlay Analysis in GIS Gathering Information from Multiple Data Layers One of the many strengths of a GIS is that you can stack several data layers on top of each other for visualization or analysis. For example, if you overlay a map of the habitat for an endangered species and a land stewardship layer you can find out who is managing the land where the species exists. The concept of overlay analysis is easy, however it can get complicated because real world objects in a GIS can be represented by different features such as points, polygons, lines and pixels. Each type of overlay requires a unique analysis procedure in a GIS. For example point over polygon overlay is different from point over pixel overlay etc. Vector Data Coordinate-based data structures commonly used to represent map objects. Each object is represented as a list of X,Y coordinates Remember the three types of vector data: points, polygons and lines? Examples - tree, poles, roads, housing developments, zoning districts Points Polygons Lines Raster Data Cell-based representation of map features. Each cell has a value. A group of cells with the same value represent a feature..and then there is raster data where the real world features are represented by pixels (cells). Examples - satellite imagery, aerial photography and some come from software packages like GRID and ERDAS. Point / Polygon overlay This map illustrates a point over polygon overlay. The polygons here represent different vegetation types created via aerial photo interpretation. The points are GPS coordinates for vegetation plots characterizing the vegetation in the area. The GPS points will be used to estimate how accurate the photo interpreted map is.

2 Point / GRID overlay This is a similar map where the vegetation layer was created via image analysis of a satellite image resulting in a raster dataset (grid) where all green pixels represent vegetation type 1, the brown pixels vegetation type 2 etc. Again, the orange points are GPS locations for ground control points. Piercing Needle Approach Multiple layers of data are in relative position with one another - inserting a "digital pin" through the stack of overlayed data. This allows questions to be answered concerning everything occurring at a particular location. This view of the world limits focus to a particular point and precludes wide-area landscape analysis. Two overlay analysis principles will be described here: 1 the piercing needle approach 2 the cookie cutter approach. Both methods involves extracting data from multiple GIS data layers at locations of interest. In the piercing needle approach information from one or more data layers are extracted at one single point (point polygon overlay). For example, you may want to know the habitat type and canopy cover at bird nest locations. In this case the bird nests are represented by a point layer and the habitat type, elevation and canopy cover is represented by individual polygon data layers. The result of such an analysis will be a table that contains the habitat type and canopy cover for each bird nest location in your dataset. GIS Overlay Analysis - Points Overlay type Point to Polygon Point to Grid ArcView 3x Geoprocessing Wizard Spatial Join Analysis Tabulate Areas or Script: samplegrids.ave ArcGIS 9x Join and Relates: Join Join data based on spatial location Raster Calculator Toolbox Spatial Analyst Tools- Extraction-Extract by Points or Sample ArcInfo Workstation Arc: identity GRID: sample This table gives you information on what tool in ArcGIS (or ArcView 3.x or ArcInfo) you can use to create a point to polygon or point to grid overlay. For example in a point/polygon overlay, to find the polygon characteristics of the locations where the points are located, use the Join and Relates features. In Exercise 2 which is part of this lesson you will perform these different kinds of overlay analyses. What covertypes burned in the Selway- Bitterrot Wilderness in year 2000? Polygon Overlays on polygons or grids Sometimes we want to overlay a polygon layer over another polygon layer or a raster. In the map above the polygons outlined in black represent fire perimeters and the raster map represents different cover types. An overlay analysis can here tell you the cover types within the fire perimeters. Fire perimeters 2000

3 Region Wide: "Cookie Cutter Approach" Vector data Clip Intersect Union Raster data GRID clip (masking) Zonal statistics (GRID) Combine The other overlay approach is the cookie cutter approach. This is an example of a polygon to polygon overlay analysis. Going back to the bird nest example, rather than extracting data from the exact bird nest location it might be interesting to analyze the conditions within a 100 meter buffer around the bird nest. The first step in such an analysis would be to create a 100 m buffer around the points (nests). The buffers will be represented in a new polygon layer in GIS. Next you would clip or intersect the habitat and canopy cover layers with the buffer layer and finally summarize the habitat types and canopy cover classes within the buffers. If your habitat and canopy cover layers are represented by raster data you would use the raster tools masking, zonal statistics or combine rather than the vector tools clip and intersect. Vector Overlay Table Overlay type Poly to Poly Poly to Grid Grid to Grid ArcView 3x Geoprocessing Wizard Intersect or Union Analysis Tabulate Areas Spatial Analyst Zonal Statistics Analysis Tabulate Areas or Extension: Grid Transformation Tools Transform Grids - Combine ArcGIS 9x Toolbox Analysis Tools Overlay Union or Intersect Convert the polygon cover to a grid or the grid to a polygon Spatial Analyst Zonal Statistics Raster Calculator Combine(grid1, grid2) Analysis Zonal- Tabulate Areas ArcInfo Arc: intersect Arc: union Arc: identity GRID: zonalstats GRID: combine Fire Start Locations & Vegetation (GAP) Fires > 640 acres S Miles W N E fire st art lo c a tio n s Vege Urban Agric ulture N a tiv e g ra s se s Exotic grasses Herbaceous clearcut Wetlands/wet meadows Other shrubs Sagebrush/rabbitbrush Broadleaf forests Xeric coniferous Mesic coniferous Water Riparian Barren Land Alpine meadow Sn o w, Ic e, C lo u d s Broadleaf/coniferous mixed Burned, standing timber N o D a ta The table above guides you as to what tool to use when performing a polygon to polygon overlay (vector overlay) or a poly to grid or grid to grid overlay. We will work with these tools in the lab exercise for this lesson. This map shows the vegetation cover types of Idaho overlayed with the fire start locations for fires larger than 640 acres during the time period What cover types do you think has the most fire starts (> 640 acres)? When you estimate the relative abundance of fire in different cover type you must consider how much of each cover type there is in the state. If the entire state was in the sagebrush cover type we would expect all fires to start in sagebrush. If half the state was sagebrush and half was forest you would expect half the fires to start in sagebrush and half in forest if the start of fires was a random event. Is it?

4 Relative abundan Idaho State , Fire and Vegetation Exotic gr Native g Agriculture Fires > 640 acres vegetation Mesic c Xeric co Broadleaf Sagebrush Wetland Other sh Vegetation classes fire frequency Riparian This graph shows data extracted from an overlay analysis of the two layers in the previous map (cover types and fire start locations). First, the black bars here show the distribution of vegetative cover types in Idaho. For example ~14% is in agriculture, 8% in native grasses, 3% in exotic grasses etc. Second, the red bars show the proportion of fires (> 640 acre) that started in these cover types. If the red and the black bar are the same height, the number of fires started are in proportion to the land area covered by that vegetation type, ie. random. If the red bars are taller than the black bars there are more fires than expected started in that cover type (more than random). What cover types have a higher proportion of large fires than you would expect if the distribution was random? Answer: (Native grasses, exotic grasses, other shrub, sagebrush) Large fires (>640aacres) by aspect class In this graph the start locations of large fires in Idaho ( ) are compared to aspect. On what aspects do large fires occur more often than would be expected if the distribution was random? Answer: south, southwest, west Example of Fire-Atlas Data: Selway Bitterroot Wilderness In exercise to you will perform a few different types of overlay analysis for the Selway-Bitterroot Wilderness area in Idaho.

5 Potential Vegetation Types Western redcedar Douglas-fir Grand Fir Lower Subalpine Dry Lower Subalpine Moist Upper Subalpine Dry Upper Subalpine Moist Rock/Alpine/Barren Land This map displays the potential vegetation types in the Selway-Bitterroot Wilderness. What is a potential vegetation type? Answer: A potential vegetation type is a habitat type, i.e. represents the climax vegetation in an area. How is cover type different from potential vegetation type (habitat type)? Answer: The potential vegetation type (PVT) always represents the climax vegetation while the cover type is the vegetative cover currently occurring in an area. The cover type on a Douglas-fir PVT may be grasses or shrubs a few years after a fire, however, given time Douglas-fir will dominate the site. Spatial Analyst in ArcGIS in ArcGIS Zonal statistics is a tool that is used for overlay analysis in GIS. This tool is part of the Spatial Analyst extension. You can here find out what the mean elevation is for forest stands. Statistics are calculated for each zone in a zone dataset based on values from another dataset Example: Calculate mean elevation within forest stands Zone dataset: forest stands Value dataset: elevation Zonal Statistics Explained Zone Layer The Zone layer is in this case the forest stands while the Value layer is the elevation layer. The output table will show the mean, max, min etc elevation within each forest stand. Value Layer The output table can be joined to the zone layer to display a statistic per zone

6 Zonal statistics Zones can be continuous or non-continuous The use of zonal statistics will be illustrated in Exercise 2. The zone layer can be raster or vector The value layer must be a RASTER Many statistics are computed: mean, median, standard deviation, min, max, variety, majority, range Selected statistic can be charted Tabulate Areas in Spatial Analyst Another useful overlay analysis tool is Tabulate Areas in the Spatial Analyst Tools in ArcToolbox. You can here create a cross tabulation based on two data layers. Let s say that you would like to know how much there is of each landcover type in the watersheds within a study area. You must input the two data layers (watersheds and landcover) in this case, but you must also tell the tool what field in the attribute table that you want to tabulate. Output from Tabulate Area Watershed 1 Watershed 2 Watershed 3 Watershed 4 Forest Grassland Agriculture The output table describes how much area there is of each land cover type in each watershed. The results are in square meter (the numbers above are just an example these watersheds would be extremely small!) After completion of Exercise 2 you will master all types of overlay analysis in ArcMap and ArcToolbox!