GEOG 400, Advanced GIS, Fall 2020; Week 4 Lecture 1

Transcript

1. GEOG 400: Advanced GIS - Raster Week 4 – Lecture 1 Error and Uncertainty
2. GEOG 400: Advanced GIS - Raster Error and Uncertainty This week, we’ll talk about what we don’t know…and how to deal with it. Today, we’ll talk about UNCERTAINTY and ERRORS in raster data. Next time, we’ll talk about ERROR MODELING and CHANGE DETECTION.
3. A ~15 Minute Exercise A total station is a survey instrument that measures the locations of points. point_1 = (423426.532 N, 42098427.234 E, 294.213 Elevation point_2 = (423743.937 N, 42098937.275 E, 296.581 Elevation point_3 = (423962.472 N, 42098496.234 E, 289.283 Elevation Northing = Latitude = Y Easting = Longitude = X Elevation = Altitude = Z
4. GEOG 400: Advanced GIS - Raster Error and Uncertainty LONGITUDE, EASTING, X LATITUDE, NORTHING, Y
5. GEOG 400: Advanced GIS - Raster Error and Uncertainty ELEVATION, ALTITUDE, Z
6. You (and a partner) are a survey team. You’ve got time to survey 100 total station points. A total station is a survey instrument that measures the locations of points. point_1 = (423426.532 N, 42098427.234 E, 294.213 Elevation point_2 = (423743.937 N, 42098937.275 E, 296.581 Elevation point_3 = (423962.472 N, 42098496.234 E, 289.283 Elevation Northing = Latitude = Y Easting = Longitude = X Elevation = Altitude = Z
7. ANIMAS RIVER SILVERTON
8. You’ve got ~2 hours, or 100 survey points. Your boss wants a 1 m resolution DEM of this area. Where will you put those 100 points? ~800 FEET ~400 FEET
9. You’ve got ~2 hours, or 100 survey points. Your boss wants a 1 m resolution DEM of this area. Where will you put those 100 points? ~800 FEET ~400 FEET tinyurl.com/yyyq5y42 (in class) tinyurl.com/y6t435fa (at home)
10. How do we get from this…to this? discrete points  continuous raster Alex Walker, USU Alex Walker, USU Vector shapefile Raster DEM interpolation
11. How do we get from this…to this? discrete points  continuous raster Vector shapefile Raster DEM interpolation We’ll have a whole lecture on interpolation methods, but for today…one example.
12. How do we get from this…to this? discrete points  continuous raster Vector shapefile Raster DEM “Inverse Distance Weighting” [IDW]
13. How do we get from this…to this? discrete points  continuous raster Vector shapefile Raster DEM “Inverse Distance Weighting [IDW]”
14. How do we get from this…to this? discrete points  continuous raster Vector shapefile Raster DEM “Inverse Distance Weighting [IDW]” (12/350) + (10/750) + (10/850) (1/350) + (1/750) + (1/850) = 11.1
15. But just because we say a point is 10 m high…is it really?
16. But just because we say a point is 10 m high…is it really? Remotely Sensed/Aerial Surveys: Ground-Based/In Situ Surveys: Boat-Based/Bathymetric Surveys: Airborne light detection and ranging (e.g., lidar) Aerial photogrammetry Total Station Real-Time Kinematic Global Positioning System (RTK-GPS) Terrestrial Lidar Multibeam and Singlebeam sonar Acoustic doppler
17. Joe Wheaton, USU
18. Joe Wheaton, USU
19. But just because we say a point is 10 m high…is it really? Joe Wheaton USU
20. Not only do these things vary in cost, they vary in accuracy Remotely Sensed/Aerial Surveys: Ground-Based/In Situ Surveys: Boat-Based/Bathymetric Surveys: Airborne light detection and ranging (e.g., lidar): +/- 12-25 cm Aerial photogrammetry: +/- 10-15 cm Total Station: +/- 2-10 cm Real-Time Kinematic Global Positioning System (RTK-GPS): +/- 3-12 cm Terrestrial Lidar: +/- 0.5-4 cm Multibeam and Singlebeam sonar: on par with terrestrial lidar
21. Accuracy: How close is a value to the “truth”? Precision: How close is a value to the other values? “Our GPS is accurate to within 1 meter” “1451 meters elevation” vs. “1451.584 meters elevation”
22. Not only do these things vary in cost, they vary in accuracy Remotely Sensed/Aerial Surveys: Ground-Based/In Situ Surveys: Boat-Based/Bathymetric Surveys: Airborne light detection and ranging (e.g., lidar): +/- 12-25 cm Aerial photogrammetry: +/- 10-15 cm Total Station: +/- 2-10 cm Real-Time Kinematic Global Positioning System (RTK-GPS): +/- 3-12 cm Terrestrial Lidar: +/- 0.5-4 cm Multibeam and Singlebeam sonar: on par with terrestrial lidar
23. Some really important distinctions in sampling and error: Sampling Pattern: Random sampling randomly pick spots and survey at those locations Feature-based sampling: more effort devoted to particular features (slopes, trees, river banks, cliffs) Uniform grid sampling: cover the area in a regular grid of survey points LEVEL OF THOUGHT REQUIRED
24. Some really important distinctions in sampling and error: X, Y, and Z errors: X,Y Errors [positional error]: errors in your points’ locations on a 2D map Z Errors [elevation error]: errors in the height of your points
25. Some really important distinctions in sampling and error: X, Y, and Z errors: X,Y Errors [positional error]: errors in your points’ locations on a 2D map Z Errors [elevation error]: errors in the height of your points Most of the time, our instruments are better at measuring X & Y than they are at Z And cells encompass much bigger areas than individual points on the ground …so Z errors are the ones that we’re concerned with more than X and Y X Y Z
26. But just because we say a point is 10 m high…is it really? …or is it 10.01 m, 9.99 m, 11 m, 20 m, 2 m, or something else?
27. But just because we say a point is 10 m high…is it really? …or is it 10.01 m, 9.99 m, 11 m, 20 m, 2 m, or something else? If we surveyed the same point three times with GPS we’d probably get three different elevations
28. But just because we say a point is 10 m high…is it really? …or is it 10.01 m, 9.99 m, 11 m, 20 m, 2 m, or something else? Because our rasters are built from point measurements every raster cell inherits those points’ errors! If we surveyed the same point three times with GPS we’d probably get three different elevations
29. We can correct for bias (if we know it) ‘we subtracted 0.25 m from all points’ We can account for random error in our analysis ‘an elevation of 135.5 m +/- 0.1 m’ …or bias …or uncertainty
30. If you get nothing else from today… it’s that you should be healthily skeptical of all raster data because (mostly) all rasters are interpolated products from point data Error in the points Error in the points Error in connecting the points Error in interpolating the surface Error in the points Error in connecting the points ERRORS ERRORS ERRORS
31. If you get nothing else from today… it’s that you should be healthily skeptical of all raster data because (mostly) all rasters are interpolated products from point data Error in the points Error in the points Error in connecting the points Error in interpolating the surface Error in the points Error in connecting the points ERRORS ERRORS ERRORS