GEOG 400, Advanced GIS, Fall 2020; Week 10 Lecture 1

Transcript

1. GEOG 400: Advanced GIS - Raster Week 10 – Lecture 1 Image Analysis and Image Classification
2. GEOG 400: Advanced GIS - Raster Week 10 Are we ignoring the syllabus? Yeah, kinda… October 26; Image Analysis and Image Classification Week 14 Week 11 Week 12 Week 13 November 23; Big Raster Data, Google Earth Engine November 2; Georeferencing Raster Data November 9; Habitat Modeling and Vector Extraction November 16; Spatial Statistics Final Project Prompt: posted by 8AM on November 30th – due by 5PM on December 5th
3. GEOG 400: Advanced GIS - Raster Week 14 November 23; Big Raster Data, Google Earth Engine [no lab this week] In order to use Google Earth Engine, you’ll need to sign up with them. Please do this by the end of this week because if there are problems, we need to fix them! earthengine.google.com/signup You can (and should!) use whatever email address you used to sign up for your Google Site …which is most likely your @fortlewis.edu address They’ll send you an email that you need to confirm, so please don’t forget to do this.
4. GEOG 400: Advanced GIS - Raster Week 14 November 23; Big Raster Data, Google Earth Engine [no lab this week] In order to use Google Earth Engine, you’ll need to sign up with them. Please do this by the end of this week because if there are problems, we need to fix them! earthengine.google.com/signup You can (and should!) use whatever email address you used to sign up for your Google Site …which is most likely your @fortlewis.edu address They’ll send you an email that you need to confirm, so please don’t forget to do this.
5. Week 10: Image Analysis and Image Classification A few weeks back, we talked about remote sensing. Let’s chat some more about this, because it’s vital to this week’s concepts Remote sensing is the process of detecting and monitoring the physical characteristics of an area by measuring its reflected and emitted radiation at a distance (typically from satellite or aircraft).
6. Week 10: Image Analysis and Image Classification A few weeks back, we talked about remote sensing. Let’s chat some more about this, because it’s vital to this week’s concepts Remote sensing is the process of detecting and monitoring the physical characteristics of an area by measuring its reflected and emitted radiation at a distance (typically from satellite or aircraft). Active Remote Sensing: energy for data collection is provided by the sensor. The sensor emits radiation which is directed toward the target to be investigated. Passive Remote Sensing: energy for data collection is naturally available. Examples: synthetic aperture radar, sonar, lidar scanners (ground-based and airborne), total stations Examples: radiometers, gravimeters, satellite/ground-based photography (uses reflected sunlight)
7. Week 10: Image Analysis and Image Classification Passive Remote Sensing: energy for data collection is naturally available. Examples: radiometers, gravimeters, satellite/ground-based photography (uses reflected sunlight) LIGHT LIGHT Camera Lens Charge-Coupled Device Digital Image of Pixels CONVERSION Light (a form of electromagnetic radiation) is passed through a camera lens and hits a charge-coupled device (CCD), composed of many individual sensors that store the amount of light hitting each sensor as elements, or pixels, in a digital photograph.
8. Week 10: Image Analysis and Image Classification Some commonly-used imagery sources Satellites: Commercial… WorldView: 1 foot resolution, 8 bands QuickBird: 2 meter resolution, 4 bands IKONOS: 1 and 4 meter resolution, 4 bands plus black/white photos Government… LANDSAT: since 1972, currently on LANDSAT 8; 8 bands @ 30 m resolution MODIS: moderate resolution imaging spectroradiometer; 30 bands @ ~ 250 m resolution AVHRR: advanced very high resolution radiometer; 5 bands @ 1 km resolution
9. Worldview: 30 cm
10. LANDSAT: 30 m
11. MODIS: 250 m
12. Week 10: Image Analysis and Image Classification Some commonly-used imagery sources Satellites: Commercial… WorldView: 1 foot resolution, 8 bands QuickBird: 2 meter resolution, 4 bands IKONOS: 1 and 4 meter resolution, 4 bands plus black/white photos Government… LANDSAT: since 1972, currently on LANDSAT 8; 8 bands @ 30 m resolution MODIS: moderate resolution imaging spectroradiometer; 30 bands @ ~ 250 m resolution AVHRR: advanced very high resolution radiometer; 5 bands @ 1 km resolution Aerial Imagery: - National Agricultural Imagery Program (1 m imagery at 3 year repeat intervals for United States) - Legacy/historic aerial images (often available back to 1950s)
13. Analog aerial photograph of Durango, 6/15/1956
14. NAIP Image of Durango, 9/9/2019
16. You’ve downloaded an image from EarthExplorer (or purchased one from a commercial company) …now what can we do with it? We’ll use LANDSAT as an example
17. Multi-Band Raster (e.g., Satellite Image) Single Band Raster (e.g., Digital Elevation Model) Bands: individual layers in a raster dataset with data from different parts of the electromagnetic spectrum
18. The most common multi-band raster we deal with is color imagery, which has three bands (red, green, blue) Bands: individual layers in a raster dataset with data from different parts of the electromagnetic spectrum
19. You’ve downloaded an image from EarthExplorer (or purchased one from a commercial company) …now what can we do with it? We’ll use LANDSAT as an example
20. You’ve downloaded an image from EarthExplorer (or purchased one from a commercial company) …now what can we do with it? We’ll use LANDSAT as an example
21. Multiband rasters might look a bit different from other rasters when you bring them into ArcGIS!
22. Hmm, this doesn’t look quite right! To generate a true color image, we want red, green, and blue bands displayed in that order Multiband rasters might look a bit different from other rasters when you bring them into ArcGIS!
23. Multiband rasters might look a bit different when you bring them into ArcGIS! Hmm, this doesn’t look quite right! To generate a true color image, we want red, green, and blue bands displayed in that order
24. Multiband rasters might look a bit different from other rasters when you bring them into ArcGIS! Hmm, this doesn’t look quite right! To generate a true color image, we want red, green, and blue bands displayed in that order
25. Multiband rasters might look a bit different from other rasters when you bring them into ArcGIS! Sometimes we intentionally display bands out of order to bring out features Image captured southwest of Durango in March
26. The IMAGE ANALYSIS TOOLBAR in ArcGIS is where the uh, image analysis happens. Access this from Windows  Image Analysis If you’ve got a raster in your map, you can select it to activate all these fun tools If you’ve got more than one raster in your map, make sure you’re working on the one you want to work on! Let’s go through these sets of tools, starting with the display options.
27. Contrast @ - 80 Contrast @ + 80 Contrast refers to the difference between light and dark colors in an image
28. Contrast @ - 80 Contrast @ + 80 Brightness @ - 30 Brightness @ + 30 Brightness refers to the overall lightness or darkness of an image
29. Gamma affects the brightness of middle pixel values without affecting the extreme end-member pixels. Normal gamma is 1; here, gamma is 0.5
30. Gamma affects the brightness of middle pixel values without affecting the extreme end-member pixels. Normal gamma is 1; here, gamma is 9.5
31. Dynamic Range Adjustment (DRA) shows the full range of colors in the image, no matter what extent you’re zoomed in/out to. We’ve used this with DEMs before. DRA off DRA on
32. Histogram Stretch Method: controls how the range of colors are distributed across the range of values in your image Max/Min whole range displayed Standard Deviation middle values emphasized Histogram Equalize emphasize contras
33. Histogram Stretch Method: controls how the range of colors are distributed across the range of values in your image Histogram Equalize emphasize contras Or choose your own adventure!
34. Won’t go through the interpolation type too much, but: - Nearest neighbor and majority are good for discrete data - Bilinear and cubic convolution are good for continuous data
35. Processing Tools Clip does exactly what you’d think it would do [clips out an area of interest for further analysis]
36. Original Image with Polygon Overlay Clip Result
37. Processing Tools Mask does exactly what you’d think it would do [masks out an area we don’t want to include]
39. One thing that’s super important to remember is that just because you ran a tool, the outputs aren’t automatically saved! So you need to save each of these to a new raster.
40. The normalized difference vegetation index, or NDVI = 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 − 𝑅𝑅𝑅𝑅 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 + 𝑅𝑅𝑅𝑅 8 = 𝐵𝐵𝐵𝐵 5 − 𝐵𝐵𝐵𝐵 4 𝐵𝐵𝐵𝐵 5 + 𝐵𝐵𝐵𝐵 4
41. What’s the greenest place in Durango in early March?
42. What’s the greenest place in Durango in early March? Dalton Ranch Hillcrest
43. Finally, there are a number of filtering options we can use Each of these are an example of a FOCAL RASTER OPERATION!
44. Finally, there are a number of filtering options we can use Each of these are an example of a FOCAL RASTER OPERATION!
45. Original Image “Blur” “Sharpen” And again, any edits you make to an image must be saved/exported to a new raster!
46. Original Image “Blur” “Sharpen” And again, any edits you make to an image must be saved/exported to a new raster!
47. You’ve already used the “Composite Bands” tool in the 416 fire lab But here’s a quick review, because it’s important for image analysis as well! Instead of manually selecting the bands to display (and potentially introducing errors if we’re wrong), we can use “Composite Bands” to make a new single-band image out of these three bands.
48. You’ve already used the “Composite Bands” tool in the 416 fire lab But here’s a quick review, because it’s important for image analysis as well! Instead of manually selecting the bands to display (and potentially introducing errors if we’re wrong), we can use “Composite Bands” to make a new single-band image out of these three bands.
49. You’ve already used the “Composite Bands” tool in the 416 fire lab But here’s a quick review, because it’s important for image analysis as well! Instead of manually selecting the bands to display (and potentially introducing errors if we’re wrong), we can use “Composite Bands” to make a new single-band image out of these three bands.