In theory clasees, we all have learnt about drainage density (DD), which is the ratio of 'total length of stream network in a basin (L)' to the 'basin area(A)'. However, when, we are trying to determine DD of some Indian river basins using Arc-GIS, we are facing some confusions. For instance,
1. We have to define some particular break value in Arc-GIS to get streams of differnt orders. Lesser the break value, higher is the stream order, therefore resulting in a higher value of L, and vice versa. For example, when we set the breakvalue as 500 for a particular basin, we are obatining a stream order upto 7, but if we increase the break value to 1500, the max. stream order reduces to 5 - automatically the L value also reduces. Thus, the same baisn may yeild two differnt DD values, under two aforesaid considertions of break values.
2. We also fixed the theoretically least possible break value , i.e., >0 and obtained a extremely dense stream network with high DD value.
So, my question is, what should be the threshold break value for a particular basin in order to get the DD?
3. From literature, we found that, there are five classes of DD with the following value ranges (km/km2), i.e., very coarse (8). However, for 20 river basins across differnt parts of India, we obtained DD values ranging between 1.03 to 1.29, which makes all those basins fall under very coarse category. But, from our visual inspection (one sample bain attached below), it seems to be very less to us.
We want some justification/ clarification/ comment on it.
You may want to look at some of the papers I uploaded, starting with Stream Types and Management Implications. In the early days, I used the 1:24000 topographic map crenulations or indentations aligning to identify channel network. I have since found that LiDAR can add about another two orders due to its added detail. In converting contour maps to DEMs and applying in GIS, you will find there is a lot of editing and hopefully some ground truthing. One of the last papers in 2015, the use of LiDAR on the coastal plain drainages within the National Forest allowed substantial improvement in both hydrologic boundaries and stream networks. If you want a high quality product, editing of the network requires removing road or other blockages, and picking in some instances the main channel in braided networks, or manual digitizing if all the braided channels are desired. The early days (1970s forward to recent times) required a lot of manual digitizing and at times, aerial photo pair interpretation, using linear wheel device for distances, planimeter to even dot grids for drainage size after hand tracing the boundaries. ArcMap has some hydro tools that can expedite the process, but careful checking and editing at times still needed.
I did look up 1.6 miles/sq mile = 1 km/sq km. I remember in the mountains and piedmont 10 miles per square mile was about average using 1:24000 topo DEM. Added info probably in my various papers. The coastal plain had lower DD, and the paper from 2015 shows the drainage Boundaries, mileage and DD improvement when using LiDAR. If you have to choose how many pixels to start a drainage network, at one time I think I used 5 acres or roughly 2 hectares to start a channel network. When in the coastal plain, probably 20 acres would have been a good start. Things like erodibility, gullying, rainfall amounts and intensity, land cover and land use can affect the channel network. Sediment from a Small Ephemeral Gully paper can give an example of how small an active drainage might be in fairly extreme conditions.
William F. Hansen Thank you very much sir for your answer. I will definitely go through the papers to understand it better.
Interesting question!
Not only you will find differents results in DD (drainage density), also in any Horton analysis due to are linked to the ammount of first order channels.
So, I think, the question is ¿how to determine from when to consider that it is a channel?. It's related to the amount of flowaccumulation area (cells in the digital elevation model that drain to the cell of analysis). You called breakvalue. This number represents the upslope area (expressed in surface units or number of cells). If you recognize a channel on the ground from a certain point, you should obtain the contribution surface it would correspond to and then use that value as the threshold.
If an aerial photogrammetric restitution is carried out, the recognition of channels or rivers also responds to a threshold to define when the river "begins".
It is worth mentioning that it is also related to the scale of work.
I hope it has been useful.
Cheers
Dear Victor Hugo Burgos, I completely agree with you. I guess overlapping the developed stream network with some base map (e.g. google terrain map) might serve the purpose to identify where the river begins. However, as we are dealing with rain-fed river basins, where most of the times, the lower order streams (1st or 2nd) are dry, it becomes bit difficult to identify them as stream from the base map. Nonetheless, its a very good suggestion, and thanks a lot for that.
- Badges
- Science topic
- Similar topics
- Geoscience
- Remote Sensing