GEOG 591Adam Migliore MeyerFebruary 24, 2012
Exercise 3: Flow path mapping with Arc Hydro
Q1. How many subcatchments were formed in your watershed partitioning? What was the drainage area in hectares that you used to define the beginning of each channel?
There were 305subcatchments formed in the watershed partitioning using 3500 pixels as the threshold.
This is a 30m resolution DEM. So the drainage area is (30m * 30m) *(2000~3500 pixels) *0.000 1 m2/ha = 180 ~ 315 ha (-1.0)
Q2. Remember this is all a model. What key assumptions were made, and how would you want to improve the watershed delineation to better match field conditions?
The most glaring assumption was the threshold approximation used in delineating the streams. The pixel threshold was determined by how well the stream delineated network matched with the NHD flowline. As Map 1 illustrates, the threshold used for determining the beginnings of channels (3500 pixels) did not map exactly onto the NHD flowline.Likely a closer threshold approximation could have been used. This method also introduces another assumption by using the NHD flowline as the standard for comparison. By considering the NHD flowline as the standard, this assumes that either the NHD flowline is 1) without error or 2) has less error than the stream delineation from pixel threshold. As noted in class discussion, the NHD is quite prone to error itself especially from digitizing error.
A third assumption is D8 connectivity for determining flow direction. D8 connectivity assumes that streams will only flow in one of eight directions. Therefore the model does not consider that streams may flow in any direction or that streams may diverge.
By filling pits, the capacity of them to store water is neglected. Lakes, ponds, and stormwater infrastructure may store water or redirect it, which is not considered in this model. The pits themselves are also prone to error as the DEM is also a model. While not perfect either, LIDAR would reduce error.
To improve the model, more stream delineations could have been done using more thresholds to find a better match with the NHD flowline. Furthermore D-infinity connectivity could have been used. Instead of filling all pits, lakes, ponds, and stormwater devices could have been excluded by adding additional layers containing these features.However, these improvements must be balanced with the marginal benefit that they would provide in relation to the time and computational costs needed to execute them.
Also, burning process by vector stream data will help to improve the result (-1.0).
Q3. Make impervious cover and land cover zonal statistics maps with your catchment boundary shape file. According to the zonal maps and tables, what is the range of % impervious area for the catchments?
The New Hope Creek catchments range froma mean of 0 to 60.72 percent for impervious cover. This is not the range from the lowest pixel to the highest pixel, but rather the range at the catchment level.
(23/25)
1