GEOL 452/552 - GIS for Geoscientists I

Lecture 22 - Chapter 8 (Raster Analysis, part 3)

• Today:

• Zonal Analysis (statistics) for polygons, lines, points, interpolation (IDW), Effects Toolbar, analysis masks

• USGS Seamless raster data server

• Ch 8 Tut 36 - end

• HW 13: Ch 8 challenge questions (+ extra, see WebCT)

• Class exercise data in follow_along/Ch8C_class_ex folder, copy and start ArcMap with Ch8C_class_ex.mxd

Zonal functions (statistics)

• Needs 2 inputs:

• a zone layer: polygons, lines or points; or integer raster

• a value raster

• Zones in a “zone raster” don’t need to be continuous! (see: regions)

zones (raster, polygons)value raster

• for each zone do:

• get stat for value raster cells that are inside the zone

• Creates many types of stats per zone “id”

• creates a standalone table and a chart

• Table can be joined to zone layer (via OID as key)

• think: Summary but with spatial grouping instead of group

• example:

• for each watershed (“groups)

• get statistics (mean,min, max, etc.)

• of the elevation inside (covered)

value raster = elevation raster zone layer = watersheds

avg. elevation of blue watershed = 674.5 m

average elevation of yellow watershed = 234.6 m

• Which watershed (or type of outcrop) has the lowest average (mean) elevation of any ?

• Spatial Analyst - Zonal Statistics

• zone layer: watershed_zones (or geol_zones.img)

• value raster: dem2 (elevation)

• Join output to zone: Yes, Save as elev_per_watershed.shp

• All Stats are about elevation

• Sort by MEAN

• COUNT: numberof cells with eachwatershed

• Points and linecan also be usedas zones

Spatial Interpolation• point samples (x,y, “value”): here 7 samples, (A) irregular distribution• for a raster (B) fill each cell with a value• Principle: a cell value should be similar to the cell’s closest samples • different spatial interpolation algorithms (IDE, spline, kriging)• distance to sample and sample value matter

A) B)

Interpolation• Need: point samples

(here: elevation, think GPS points)

• Go over all cells of an empty raster, for each cell:

• Grab all the closest point samples (here: within a radius around the cell)

• Do some spatial math with these point samples

• IDW: faster, Spline: smoother

• result: cell’s elevation

• True vs interpolated elevation:

• Usually: loss of detail, smoother results

• Limit: number (density) of input point samples

• Spatial Analyst - Interpolate to Raster - Inverse Distance Weighting (IWD)

• Calculate a 50 m raster (IDE_elev.img) from elevation_point_samples

• Us default parameters for IDE algorithm (ELEV, power 2, 12 closest points)

• Symbolize as stretched, import colors from dem2

True elevation Interpolated

• “blurry” results from IDE interpolation(compared to dem2)

• same raster resolution (50 m) but far less samples were used than when making the real DEM (probably used points from contour lines)

• (re-visit next lecture: DEM from Lidar)

The Effects toolbar• How can we visually compare the two elevation rasters?

• make sure both rasters have the same high/low color values (1965 and 872, same color ramp, stretched)

• Move IDE elevation on top of dem2

• Activate the Effects Toolbar

• Set Layer to IDE elevation layer

• press swipe button and drag left mouse

• or: press flicker (500 ms) button

IDW: search radius

• variable search radius: grab n (12) closest points

• (... up to a max. distance, 0=no distance limit)

• fixed search radius:grab ALL points within a distance of X units

• (... but use at least n points, 0=use all)

Setting an Analysis maskMask grid or polygon

NoData

Data

Applies a “clip with mask” to ALL output grids

Masks may be grids, polygon shapefiles or feature classes.But: rasters are still rectangles, outside is simply filled with NoData (usually transparent) Example: Analysis always within a county but your input rasters are largerSee also: Extract by Mask tool (raster clipping) or DataFrame - Clip to Shape (non-destructive, temporary visual clip!)

mask.img or mask.shp

2.

Aspect of DEM

ortho.gis.iastate.edu raster data

• Home > Map type (here: Lidar Hillshade) > Status map (shown here)

• Recenter vs. Zoom in, may need to press “Refresh Map”

• Change width/height (e.g. to 2000 x 2000 pixels), will keep center but may be too big to view)

• When viewing the right area, press “Download map”

• 10 Mb (10000K) download limit

• Get Geotiff (interanally georeferenced to UTM 15 NAD83)

• Or: get JPEG version AND ArcView Header (.jgw worldfile, coordinates of corners)

• put 1234.jpg and 1234.jpg in same folder, if rename: keep same base name !

Seamless.usgs.gov• Web based raster download (interface can be very clunky!)

• Display = preview in Browser (not all types of data are available everywhere!)

• Request download area: interactive rectangle or extent (lat/long)

• Download data or modify Data request (wait for refresh)

• If possible get geotifs, warning: data can be large, don’t download multiple files!

Wrap up• HW 12, ex 5 gotcha: use zonal analysis (zones = points),

new due data Nov. 17 (check WebCT)

• HW 13, challenge problem +

• in addition to AND'ing (or multiplying) the 6 binary maps (for "where are ALL criteria true?"), also ADD (+) the 6 binary maps together (

• Visualize the resulting 0-6 values in a separate map

• Explain the meaning of the added binary maps

• Next lecture: Lidar Data, ArcScene

•