GEOG 400, Advanced GIS, Fall 2020; Week 3 Lecture 2

Transcript

1. GEOG 400: Advanced GIS - Raster Week 3 – Lecture 2 Raster Resolution
2. GEOG 400: Advanced GIS - Raster Raster Resolution This is the second of two closely-related lectures Last time, we talked about BIT DEPTH This time, we’re talking about CELL RESOLUTION
3. BIT DEPTH CELL RESOLUTION Number of memory bits used to store color data for each pixel in a raster image All pixels require the same number of bits The number of bits determines the range of colors the image may have The dimension of the cell size representing the area covered on the ground Resolution is constant across a raster (with very few exceptions) Resolution balances the size of features that can be represented with file size, processing time “A 32-bit integer raster” “A 5-meter resolution raster”
4. BIT DEPTH CELL RESOLUTION Number of memory bits used to store color data for each pixel in a raster image All pixels require the same number of bits The number of bits determines the range of colors the image may have The dimension of the cell size representing the area covered on the ground Resolution is constant across a raster (with very few exceptions) Resolution balances the size of features that can be represented with file size, processing time “A 32-bit integer raster” “A 5-meter resolution raster”
5. 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”
6. 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”
7. The dimension of the cell size representing the area covered on the ground NOT The actual size of the pixel on your screen
8. There are many types of resolution; just so you’re familiar with the terms… Spatial Resolution: the area on the ground covered by each cell Radiometric Resolution: synonymous with bit depth; number of possible values in each cell Spectral Resolution: you’ll see this in remote sensing; the ability to distinguish between parts of the electromagnetic spectrum [a grayscale photo versus a color and near-infrared photo] Temporal Resolution: how frequently measurements/images are collected through time [daily data is a higher temporal resolution than annual data]
9. There are many types of resolution; just so you’re familiar with the terms… Spatial Resolution: the area on the ground covered by each cell Radiometric Resolution: synonymous with bit depth; number of possible values in each cell Spectral Resolution: you’ll see this in remote sensing; the ability to distinguish between parts of the electromagnetic spectrum [a grayscale photo versus a color and near-infrared photo] Temporal Resolution: how frequently measurements/images are collected through time [daily data is a higher temporal resolution than annual data]
10. Spatial Resolution: the area on the ground covered by each cell
11. But resolution is not the same as scale! smaller scale  more detail shown higher resolution  more detail shown Spatial Resolution: the area on the ground covered by each cell
12. But resolution is not the same as scale! smaller scale  more detail shown higher resolution  more detail shown Spatial Resolution: the area on the ground covered by each cell
13. Spatial Resolution: the area on the ground covered by each cell For the rasters we’ll deal with, resolution is constant across the dataset.
14. Spatial Resolution: the area on the ground covered by each cell For the rasters we’ll deal with, resolution is constant across the dataset. 10 m 10 m 10 m 10 m 10 m 10 m Cells are the same size everywhere
15. 30 m 30 m 10 m 10 m Radial Distortion: cells vary in size
16. Radial Distortion: cells vary in size
17. Knowing the resolution can help you answer a few (important!) questions 1. Where am I?
18. Terminal 4, Phoenix Sky Harbor International Airport
19. Terminal 4, Phoenix Sky Harbor International Airport
20. 1 m Resolution: the south end of Concourse B, near gate B17
21. 5 m Resolution: the south end of Concourse B
22. 10 m Resolution: probably in a concourse, maybe on a plane
23. 30 m Resolution: I can maybe tell it’s an airport
24. Knowing the resolution can help you answer a few (important!) questions 2. What’s the value where I am?
25. 10 m Resolution: I’m at the top of a mountain
26. 10 m Resolution: I’m at the top of a mountain
27. 30 m Resolution: I’m at the top of a mountain
28. 200 m Resolution: I’m at the top of a mountain
29. 200 m Resolution: I’m at the top of a mountain
30. Knowing the resolution can help you answer a few (important!) questions 3. How long is this gonna take? 4. Am I gonna need a bigger hard drive?
31. 10 m Raster 5 m Raster 1m: almost 1 GB! For each doubling in resolution, we increase file size by a factor of 4
32. Raster Considerations 2: Resolution Raster resolution is a delicate balancing act between the information you want to convey and the size (and processing time) of the file you’re dealing with. “1 m resolution” “0.5 m resolution” 1 m 1 m 1 m 1 m To DOUBLE the resolution, we have to QUADRUPLE the cells So be very careful, because files can get really big, really fast.
33. For a 1,000 x 1,000 cell raster: 1 bit: 0.125 MB 8 bit: 1 MB 16 bit: 2 MB 24 bit: 3 MB 32 bit: 4 MB Doubling bit depth doubles our file sizes (recall that it’s 2n) Bits are relatively cheap resolution is expensive Bit depth: factor of 2 Resolution: factor of 4
34. How we go from this: …to this: Isn’t a trivial matter! 10 m resolution 200 m resolution Arc will give you four choices for resampling: 1. Nearest Neighbor: whatever value was in the new cell’s center is assigned to the new cell This really quick, and never introduces new cell values But it’s best for discrete data
35. How we go from this: …to this: Isn’t a trivial matter! 10 m resolution 200 m resolution Arc will give you four choices for resampling: 2. Bilinear: values of the four nearest cells in the original raster are weight-averaged and the value is assigned to the new raster cell This is a little slower, but preserves continuous data (elevation, slope, temperature, precip, noise data)
36. How we go from this: …to this: Isn’t a trivial matter! 10 m resolution 200 m resolution Arc will give you four choices for resampling: 3. Cubic Convolution: like bilinear, except with 16 nearest cells I can’t recall an instance where I’ve had to use these. 4. Majority like nearest neighbor, except with 16 nearest cells
37. Two Quick Ways to Change Resolution 1. Resample Tool The raster you want to change cell sizes of The new cell size (in the original raster’s coordinate system units!!!!) The technique to use for computing the cell values of the new raster There’s nothing stopping you from making a raster a finer resolution… Except (a) file size and (b) whether it’s actually meaningful
38. Two Quick Ways to Change Resolution 2. Copy the raster to a new file New raster filename/type The new cell size (in the original raster’s coordinate system units!!!!) You need to click ‘Settings’ to define the resample method (bilinear, nearest neighbor, etc.) that you want to use.
39. Two Quick Ways to Change Resolution 2. Copy the raster to a new file New raster filename/type The new cell size (in the original raster’s coordinate system units!!!!) You need to click ‘Settings’ to define the resample method (bilinear, nearest neighbor, etc.) that you want to use.
40. A Quick Exercise to Review…
41. For Monday’s Class… Read pages 66 – 86 on data models!