Mapping Geomorphic Changes in Black’s Fork from 1996 to 2004
Geomorphology is the study of the processes that creating landforms and the evolution of landscapes over time. This project focuses on short term geomorphic changes in small mountain watersheds.
Background information:
In the mid 1990’s eight study reaches were developed in
stream drainages in the
Figure 1: Black’s Fork is the Red Reach (map from David Galbraith)
Methods:
Georeferencing:
Georeferencing is when data from one map is a lined with another map’s coordinates system. The original maps (1996) had been previously georeferenced, with respect to the ’96 and ’04 survey data. The original maps and survey data exist in an arbitrary coordinate system created when the first surveys was taken. The location of the first benchmark is recorded as (0,0) then the coordinates system is built around this point. With a survey grade GPS the reach could be mapped in a real world coordinates, but that technology was not available at the time. While the maps were being georeferenced personal geodatabase was using the arbitrary coordinates system to store all the feature classes that were to be created.
The ’04 map sections were scanned then georeferenced by choosing a control point that exists on both years’ mapping. The points were used to create links between the map and the (x,y) coordinates created when surveying, by entering the x and y manually. This was done for one point at the top of the section and then repeated for a point at the bottom; GIS then scaled, rotated, and shifted the map according to the two points. Continuing, I matched the plotted mapping control points, to their location in a shapefile, GIS was able to calculate the amount of error associated with the transformation. For the Black’s reach there were five different sections of map that had to be processed this way to create a reach-long georeferenced ’04 map (Figure 2).
Figure 2: Georeferenced ’04 Black’s Forks map
Digitize Maps:
The next step was to digitize the hand drawn maps. Starting with the original hand drawn maps from the 1996, feature classes were created. Line and polygon classes were created to represent all geomorphic features. Line and polygon classes were also created to represent all man made features important in a map, but not necessarily affecting the geomorphic features. For example the road and road cut are man made features but included in the geomorphic features because they affect the geomorphology of the reach. A separate feature class was created to represent large woody debris. In the “geomorph_poly” a large coverage polygon was created to cover the study reach. Using the editor tool bar polygons were cut out of the large polygon to form the individual geomorphic features. This proved to be a difficult task, due to the faintness of the original maps, my lack of knowledge of the mapping, my limited knowledge of geomorphology, and the fact that I am just learning GIS. The create new polygon function in the editor tool bar was used to make the non-geomorphic polygons.
Attributing:
The next step in the process was attribute each new polygon. Each polygon was given a description, and index code. The index codes used are listed in Table 1. The index codes were chosen in respect to the geomorphic features location in the reach. All in subsurface features were give negative numbers. Slopes were assigned a .5 increases from the surface they slope up from, for example the bank is 1.5 because it is sloping up from the bars and silt surfaces (silt bars) index code 1. Independent feature on larger features were give a .5 increases from the surface they were located on, for example the mound of gravel is 3.5 is located on the active flood plain index code 3. Index codes were assigned in order to quantify the geomorphic changes over time.
|
Description |
Index Code |
|
Pool |
-3.00 |
|
Run |
-2.00 |
|
side channel |
-1.50 |
|
riff/run |
-1.50 |
|
Riff |
-1.00 |
|
silt surface |
1.00 |
|
Bar |
1.00 |
|
Bank |
1.50 |
|
low flood plain |
2.00 |
|
soloufuction |
2.50 |
|
Slope |
2.50 |
|
active flood plain |
3.00 |
|
landslide |
3.25 |
|
Weir |
3.50 |
|
Slope |
3.50 |
|
Road cut |
3.50 |
|
mound of gravel |
3.50 |
|
Levee |
3.50 |
|
outwash terraces |
4.00 |
|
Road |
5.00 |
Table 1: Table of index code
Figure 3 represents the geomorphic polygons after they have been attributed and symbolized according to their description.
Figure 3: Map of Black’s Forks Polygons
Creating the ’04 Map:
Next the ’96 feature class was exported into the personal geodatabase and renamed changes_04. Using the editor tool bar changes were mapped by cutting and merging polygons. When merging it was important to merge according to the correct attributes so that the changes could be assigned the new index code representative of the new geomorphic feature. The active flood plain became part of the side channel the new polygon created has to be merged with the side channel, so it would be assigned the index code of -1.5. Figure 4 shows some of the changes that were made in the reach, the yellow line represents the ’04 features and the colored polygons are the ’94 features. Then there exists a mapping of the reach in ’04 with attributes and index codes.
Map of the Geomorphic Changes ‘04
Figure 4: Example of the Changes in the Black’s Fork, the yellow lines are the mapping from ’04.
Mapping the Geomorphic Changes:
Two maps were created using the ’96 and ’04 map to show the geomorphic changes in the Black’s reach. The first map (Figure 5) shows the surfaces that experienced some sort of change. The map is not specific in the change experienced and does not show the area that had undergone changes. This map was created by overlaying the ’96 map and the ’04 then selecting by location the areas that had not changed, invert the selection, export the selected features in the to the personal geodatabase. After symbolizing according to description Figure 5 was created. Figure 5 is not an extremely useful map and was not corrected for mapping errors.
Figure 5: Map of the surfaces that experienced changes from 1996 to 2004
The second map created (Figure 6) illustrates the exact areas that have experienced erosion or deposition in the Black’s Fork reach from ’96 to ‘04. This map is much more useful then Figure 5 because it allows the area of change to be calculated and shows the exact location of deposition and erosion. It was created by using the tool box analysis, overlay, intersect, the polygons created show the where the two layers intersect. Before this process it is important to change the field names in attribute table for ’04 so after intersect was complete the new index codes for ’96 and ’04 could be distinguished. The intersect coverage (change_96_04) attribute table contained all the fields from the two input coverages. The field change was then created in the attribute table of change_96_04. Change was calculated using the field calculator by subtracting the ’96 index code from the ’04 index code (Equation 1). If the result was positive then there was deposition, negative erosion, zero then there was no change.
Equation 1: ’04 index code - ’96 index code =
If
Positive = deposition
Negative= erosion
Zero= no change
Example: Active flood plain erodes to side channel
active flood plain index code = 3
side channel index code = -1.5
so -1.5 - 3 = -4.5
Therefore erosion has taken place in this intersect polygon between ’96 and ‘04
In a new field, deposition_erosion, negative numbers were assigned the string erosion and positive number the string deposition then symbolized to create Figure 6.
Figure 6: Map of the areas that were eroded or experienced deposition between ’96 and ‘04
To ensure the precision of this map each deposition and erosion polygon were checked against the hand drawn maps.
Results:
From ’96 to ’04 the area that eroded was 1,722. m2 and the area was deposited was 836 m2. Using GIS the area of the active channel in ’96 was 13,231 m2. The area of the active channel in ’04 should be the area of the active channel ’96 plus the eroded area minus the deposition (Equation 2).
Equation 2: area of original active channel + eroded – deposited = area of new active channel.
Calculations: 13,231 m2 + 1,722 m2 - 836 m2 = 14,117 m2
According to the the ’04 mapping and GIS the area of the active channel in ’04 was 14,065 m2. There is a 52 m2 difference in the area of the active channel in the two different calculations. This can be attributed to errors in digitize data.
Discussion on Digitize Data:
When considering digitized data two important terms to be aware of are precision, “the level of measurement and exactness in a GIS database relating to the quality of data and the number of errors in a dataset or map,” (Foote and Huebner, 1995) and accuracy, “the degree to which a map matches the real world” relating to the survey and attribute data (Foote and Huebner, 1995). Accuracy and precision are a function of the scale of a map. As the scale of the map decreases the accuracy and precision increases. When looking at a map a point or line is probably with in a certain area depending on the scale of the map and the accuracy and precision used when creating the map.
To consider precision and accuracy in the digitized mapping of Black’s Fork there are three sources of error to discuss survey error, plotting error, and actual mapping, or line error. Survey error is minimal due to the equipment used. Plotting error can be significant because the plotting was done on graph paper using a ruler and compass. Mapping line error is the most significant source of potential error. It is a measurement of the difference between real world lines and the map lines that represent the real world, it is difficult to quantify. The mapping line error depends on the number of survey control points. It is like when a child plays connect the dots the more dots or control points in our case, the more accurate the final product. Mapping line error also depends on the feature being mapped; a hard edge of a rock is easer to map then the changing edge of a sandbar. The last element that mapping line error depends on is the skill of the cartographer.
In this project the survey error is very small due to the type of equipment used. Plotting error could be large but at this point it can’t be quantified because I have only been to the location once. Mapping error in this location should be relative small. Mapping error has a direct relationship to the size of the area being mapped; Black’s Fork is .2km2. Therefore accuracy should be high. When thinking about mapping error it is important to take into account the surface being mapped, the hard line of a rock or bank is easer to map then the sinuous lines that compose cobble bars or vegetation. Water level when the cartographer was at the reach is an important factor in mapping error. I can’t comment on the water level because I have no knowledge of the exact dates of the two mappings. If the date were know it would be possible to retrieve the gage information from the USGS on the Black’s Fork and make hypotheses about water level as related to mapping error. Another factor directly relating to accuracy of the map and mapping error are the number of survey points. The side channel (Figure 7) did not contain any control points in the ’96 survey therefore was not very accurate in the original mapping. In ’04 the side channel was included in the survey. The change in the side channel location can be attributed to mapping error and not geomorphic forces. In the final map of geomorphic changes map the side channel change from ’96 to ’04 noted as mapping error.
Figure 7: Example of Mapping Error relating to Control Points
Processing error in the use of GIS needs to be taken into account for this project. Processing in this case comes from the GIS user having limited knowledge of the reach. Each step builds upon the previous step to create results. Processing error in the first step would be propagated throughout the project. An example of error propagating is in the ’04 map polygons shifted in an edit secession. Then the ’04 map was used to create a deposition/erosion map by overlaying the maps from ’96 and ‘04, on this map there were polygons labeled deposition and erosion that experienced no change according to the hand drawn maps. Using an edit session these errors were removed from the final map, but there still exist a difference between the active channel area as calculated using the deposition and erosion areas and the area found using GIS and ’04 map, the differences was an area of 52 m2.
Acknowledgement:
David Galbraith (Frontier Corporation, UT )
Mark Smelser (Dept of Environmental Quality, CA)
Work Cited:
Foote, Kenneth E., and Donald J. Huebner. "Error, Accuracy, and Precision." 14 Oct. 1995. 26 Oct. 2006 <http://www.colorado.edu/geography/gcraft/notes/error/error_f.html>.