|
TEACHING GEOG 385.02/GTECH 785.02 |
GIS SG course homepage |
|
CLASS DEMO 2-2. DATA DISPLAY AND MAP COMPOSITIONS
If you have not done so yet, download Demo2.exe. Downloading class demo data.
|
Viewing palette and symbol files |
Underlying values and color enhancement in autoscaling
|
|
UNDESTANDING PALETTES AND SYMBOL FILES
In the computer, GIS spatial data is organized using vector and raster data models. The attribute data is also stored digitally as text and numbers. When visualizing (displaying) the attribute data, we use colors to represent attribute values.
- Colors represent data values; there is no color without a value.
- Same colors represent the same values, different colors represent different values.
- The choice of colors is often limited by: computer architecture and human perception of color variation.
Palettes are color groups used in conjunction with particular images or themes. Graphic symbols for points, lines, and polygons are recorded in symbol files, where, in addition to colors, the shape, size, style, and fill pattern for each symbol can be specified.
- In Idrisi, each displayed color is defined by a palette color code.
- The color codes are numbers between 0-15 or 0-255 (artifact of a computer architecture).
- When you display data values, you establish a correspondence between these values and color codes.
- If the number of values to be displayed is small, use a 16 color palette, if the number of values is large, use a 256 color palette.
Viewing palette and symbol files
256 color palettes
- Display AFFAOSOL with
qualitative palette. This image shows soil classes in
- Query the image, check out Layer Properties or Metadata and scroll down the legend. Palette file used to display this image is Qual256.
There are 133 soil classes shown on this map. In other words, cell attribute values range from 0 to 133 (total of 134 values) and indicate the location of these soil types. These 134 values are displayed with 134 different colors. The colors are defined by color codes (numbers) stored in a Qual256 palette file. Out of 256 palette colors 134 are used to render these data. Each attribute value from 0 to 133 is put in correspondence to first 134 color codes, also numbered from 0 to 133. We will now open the palette file to see the how the color codes and colors are organized.
- Open Symbol workshop under Display menu or click sixth or seventh icon from the left (Symbol workshop or Palettes). This will open a default symbol or palette file.
- File/Open in Symbol Workshop. Choose to open a palette (for a raster image), and then find \Idrisi\ Symbols\Qual256.
This palette file has 256 (16x16) colors represented by color codes from 0 to 255. This is one of the standard (default) Idrisi palettes.
- Move the cursor over the palette to see the color code behind each color. Check out the color code in the first and the last box (0 and 255).
First 134 colors of this palette are used to display values from 0 to 133 in the AFFAOSOL image.
16 color palettes
16 color palettes are organized in a similar way. Only first 16 colors are defined however, while the rest is not used to display values.
- Open Layer properties for AFFAOSOL and change palette to Qual16.
The display changed dramatically. Scroll the legend down. You can see that now only the first 16 soil classes are displayed with unique colors, while the remaining color codes all produce white background color.
- Open palette file Qual16 and look at how colors are distributed throughout the palette.
- Explain why 256 color palette better displays the attribute data in AFFAOSOL image.
You can modify the each palette color by clicking on it or by using Blend and Copy functions to the left.
- Click on some colors but do not modify them. Then close the palette without saving it and close the image.
User-defined palettes
User-defined (custom) palettes usually have the same name as the image for which they are created. They are stored in the folder with the image itself (as opposed to \Idrisi\ Symbols\ folder, that stores default palettes). When copying an image, you should also copy the palette. To see a custom or user-defined palette:
- Display WORCWEST with Worcwest palette.
- Open Symbol workshop and open palette Worcwest.
Worcwest palette was edited in such a way, that most color codes display white color while first 15 color codes (0-14) correspond to colors that best convey how 13 landuse classes are distributed throughout WORCWEST image. Where possible, colors correspond to natural landcover colors (green for forest, blue for water, etc.).
- What is the range of attribute values in WORCWEST image? (1-14)
- The first color code 0 corresponds to black color. Are there pixels in the image colored black? Would the image display change is we changed black to white?
- Close WORCWEST image and palette.
Symbol files
- Open vector file KVROADS using Qual256 symbol file.
This map shows a road network in
- Open Qual256 line symbol file in Symbol workshop.
Here lines are symbolized to show differences in color. In addition, one can change line size and thickness (click on a symbol to be modified).
- How would you symbolize lines to display with the same symbol disregarding their attribute value?
Point and polygon symbols are stored in corresponding files and can be rendered according to the content of the data.
Quantitative versus qualitative data
Different color schemes should be used for discrete data and continuous data (buildings or elevation), and for qualitative and quantitative data.
- Handout Idrisi Display basics – Palettes. Review the differences between qualitative and quantitative data and corresponding data display conventions.
Display SIERRADEM Idris256
This is quantitative data, legend shows one blending color bar.
- Use the Cursor Inquiry Tool to query data values
- Open Idris256 palette in Symbol workshop
Colors change gradually and blend to emphasize continuous change in data.
DISPLAY KVLANDU with Qual 256
This time the legend shows discrete categories. The legend labels are read from
the documentation file. This is qualitative data.
- Use the Cursor Inquiry tool to query data.
- Open Qual256 palette in Symbol workshop
Colors are contrasting to emphasize the differences.
Using palettes with image data
Drag the two windows so they are both visible.
- Give SIERRADEM focus and Choose Layer Properties on Composer
- Change the palette to GREY256, then QUANT256, then NDVI256, the QUAL256.
- Explain which palettes look bad and why.
Choose CANCEL on the Layer Properties dialog to return to
the original display.
- Now give KVLANDU focus. Go to Layer Properties.
- Change the palette to QUAL16, then IDRISI16, then KVLANDU.
Which palette works best? Which does not work well?
- Then change it back to KVLANDU.
Data display summary
- Choose a QUANTITATIVE palette when you have quantitative data and want you want to get a sense of the pattern of the values.
- Use a QUALITATIVE palette when you have qualitative data and you want to emphasize the differences between adjacent values.
Handout Choosing Palette Workshop
- Figure out which palettes
should be used with which images.
Close everything.
AUTOSCALING DATA FOR DISPLAY
Autoscaling defines the relationship between range of data and color representation. Specifically, it refers to the automatic division of a range of data values into a new range of values that correspond to color codes for display. Autoscaling is used to display any range of values in a raster or vector layer with the number of colors (color codes) in a specified palette or symbol file. Autoscaling performs a linear stretch of the data values, which means that the lowest data value is assigned to the lowest palette or symbol index number, the highest data value is assigned to the highest palette or symbol index number, and all other data values are assigned to an index in direct proportion to their position within the data range. (See HELP)
Autoscaling is invoked when available color codes do not match the data in number or value.
When data is autoscaled
1. Data range is smaller than the range of color codes
- Display raster layer DEC88C with NDVI256 palette.
Note the poor contrast in this image. Why is this happening? Open Layer Properties. Note that the min and max attribute values are 0 and 15. Ndvi256 palette, however, has 256 colors and only first 15 colors are used to display the attribute values. Because the palette is quantitative, colors change gradually. In this end of the palette they all are shades of dark blue. This poor contrast can be remedied with Autoscaling.
- Click
Before values 0-15 were shown with first 16 colors of 256 color palette (all very similar), but now 0 corresponds to 0 color code and 15 corresponds to 256 color code. In other words, data values are stretched over the entire palette which dramatically improves contrast.
If you where to use NDVI16 palette with this image, would you need to autoscale the data for display?
- Check if you were right by changing the palette to NDVI16.
- Close this image.
2. Data range is larger that the range of colors ( beyond 16 or 256 values),
- Display SIERRADEM with default palette.
Autoscaling is invoked by default because there are more attribute values than available color codes. In this case, the attribute values will be distributed between color codes 0 and 255, and each color will display a range of attribute values. Thus, 0 of the palette will correspond to the min attribute value while 255 - to the max attribute value.
- Check the documentation and find out min and max data value.
Therefore, value 410=0 color code; value 1999=255th color code.
- Try to disable Autoscaling.
You get an error message that numeric characteristics of the image require autoscaling. Which characteristics are these? (data range)
3. Color codes do not match the data in values
For example, data may range from –7.2 to +4.6 (temperature data?). Real values can never be matched to byte color codes, because fractions have no clear boundary. Thus, when we display real numbers, autoscaling is always invoked.
- Display DRELIEF with default palette. This image is autoscale by default.
- Try to disable Autoscaling.
You get an error message that numeric characteristics of the image require autoscaling. Which characteristics are these? (data type)
Byte or integer data can be autoscaled upon request (check Autoscale on), if the data range is under 255 levels or 16 levels (depending on a palette).
Underlying values and color enhancement in autoscaling
Note that the underlying data values are not changed by any of these display enhancements.
- When you are doing a lab and you get a result that looks like this, don't immediately assume that you have made an error. The problem could simply lie in the display.
- First, go to Layer Properties and check the min and max values. Are they as you expected?
- If so, check which palette is being used. Is it the appropriate type for the layer you produced (i.e., is it qualitative or quantitative to match the values in the image)?
- If you have quantitative data that has a range much smaller than the number of levels in the palette you are using, click on autoscaling.
- Open Layer Properties and note the min and max values. Then click on autoscaling.
Autoscaling Summary
· Use autoscaling to increase visual contrast when the data range is smaller than the range of colors in the palette.
· Autoscaling is always invoked when the values are real or exceed the 0-255 range.
· Autoscaling does not alter the data values. It only affects the display.
Final question:
- Display SIERRADEM (default palette autoscale) and add a vector layer COUNTOURS (default palette autoscale).
Why we cannot see these contour lines?
MAP COMPOSITIONS AND MAP PROPERTIES
Map components
A good map must include certain components:
- Map itself
- Map title - a brief description of the content
- Scale (verbal, representative fraction, or graphic)
- Legend
- North arrow
- Authorship information
- Commentary and explanatory text, if needed.
Composing a map
Cartography is defined as both art and science of map making. Science implies precision of location and competency in display of data. Art means that as any visual language, map is a powerful way of communicating the information. Maps must be composed in such a way that its language is effective, not deceiving, and esthetically pleasing, which all enhance the power of maps. In particular:
- Most important components should be easily seen (e.g., map itself as opposed to the giant legend).
- Only the absolutely necessary information should be included. Too much data on one map only obscures the content of the map.
- Choice of colors and symbols is absolutely critical. If chosen in a wrong way, not only they obscure the content, they might dramatically misrepresent what the author was trying to show.
Go to Map Compositions. Compare the two maps and discuss what is wrong with the Bad Map.