Increasing Sediment Delivery Ratio (SDR) in SWAT

[Bendik Hansen]

Hello,

I am working on simulating sediments on an hourly time-step in a catchment. I am calibrating on observed suspended sediment concentration in the river. 

My problem is that in order to get the right values for my simulated concentration, I have to tweak my parameters so that I get a very high upstream sediment yield. The Sediment Delivery Ratio (upstream yield/sediments going out of reach) is typically in the range 5-10% for my simulations, while my observed data and other studies indicate it should be closer to 20%-25%. 

Does anyone know which parameters I should be adjusting to increase SDR? There is apparently a lot of erosion going on in my simulation, it's just that the sediments never reach my river (I do not have river deposition/erosion enabled).

Best regards,

Bendik

[Jim Almendinger]

Turning off channel degradation (I think) only means that SWAT does not track how channel erosion or deposition affects bed elevation or slope.  Without the channel degradation routine, SWAT normally assumes the channel is an infinite source or sink for sediment.  

I often try to minimize channel processes to start with, so that they don't obscure sediment yields (erosion) from the uplands.  You can (in theory) shut off channel erosion by setting the erosivity and cover factors to zero.  To minimize deposition, you can increase spcon by an order of magnitude (or two, I think), to increase transport capacity in the channel.  If this works, then sediment in your reaches is mostly all from the uplands (hrus and subbasins).  

However the SWAT default parameters (especially spcon=0.0001) often result in some channel deposition.  You should check what your channel deposition is by checking the values of sed_in minus sed_out for each reach (and add them up to get the net watershed-wide channel deposition or erosion).  Sed_in minus sed_out should be close to zero for each reach, if channel erosion and deposition are minimized.  It could be that your reaches are trapping sediment, making your sediment yields appear smaller than they really are from the hrus and subbasins.  

Further (as a first step, really), check if your channel widths are reasonable.  I've generally found that ArcSWAT's estimate of channel widths are about 3 times wider than actual widths.  So I routinely divide my initial channel widths by 3 before calibrating.  When you decrease the channel widths, the cross-sectional area decreases and flow velocities increase.  That will also reduce sediment deposition in the channels, even without increasing spcon.  

Hope that helps,

-- Jim

---

James E. Almendinger, PhD Director, St. Croix Watershed Research Station 16910 152nd St N, Marine on St. Croix, MN  55047 e: jalmen...@smm.org o: (651) 433-5953 x 11

--
You received this message because you are subscribed to the Google Groups "SWAT-user" group.
To unsubscribe from this group and stop receiving emails from it, send an email to swatuser+u...@googlegroups.com.
To post to this group, send email to swat...@googlegroups.com.
Visit this group at https://groups.google.com/group/swatuser.
To view this discussion on the web visit https://groups.google.com/d/msgid/swatuser/95d180d9-28c8-4eb5-9ad9-b736851ad2a7%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Bendik Hansen]

Thanks Jim, these sound like great suggestions. I will check them out now. 

I see you posting a lot of good answers here, so maybe you know the answer to this as well: Do you know how to increase the Potential Evapotranspiration in SWAT? So far I have been using the built in tools for estimating PET, but they underestimate it by approximately half. I would like to tweak some settings so that the calculated PET gets a little (or a lot) higher, if possible. I'm using the weather generator for RH and solar radiation. I know it's possible to use a .pet input file, but it's lumped for the entire basin which I would prefer to avoid.

Best regards,

Bendik

[Jim Almendinger]

I haven't really investigated changing PET.  Like many users, I've chosen different PET methods and used the one that gives me the best overall water balance, without really checking how PET affects AET (actual ET).  There are certainly parameters that affect AET (like soil moisture capacity -- the larger it is, the more ET will occur).  But I'm not sure about PET.  

A colleague has warned me that the ET calculations are sensitive to RH (at least Penman-Monteith method), and that it's much better to give a time series of actual RH values rather than letting the weather generator guess.  I personally have never had that much of a problem, and getting a good RH data record is not nearly as easy as getting good precip and temperature data.  Even here, I'm not sure if the RH values affect the PET, or the AET.  I am thinking that for Penman-Monteith, RH would affect the PET, leaving only the crop stature (relative to the reference crop) to alter PET into AET.  Perhaps other readers can confirm or correct this guess.  

-- Jim

---

James E. Almendinger, PhD Director, St. Croix Watershed Research Station 16910 152nd St N, Marine on St. Croix, MN  55047 e: jalmen...@smm.org o: (651) 433-5953 x 11

--

You received this message because you are subscribed to the Google Groups "SWAT-user" group.
To unsubscribe from this group and stop receiving emails from it, send an email to swatuser+u...@googlegroups.com.
To post to this group, send email to swat...@googlegroups.com.
Visit this group at https://groups.google.com/group/swatuser.

To view this discussion on the web visit https://groups.google.com/d/msgid/swatuser/70bd00dc-c96c-40af-b4c9-58ce95c5e26e%40googlegroups.com.

[Bendik Hansen]

You were correct, sediments were being stored in my reaches. However, tweaking SPCON and SPEXP had a rather small effect on reducing this. I tried decreasing channel widths (which, as you said, were overestimated) but it had very little impact. 

While investigating this I found several odd things: First, all my water velocities in output.vel are 0. My water depths in output.dep are okay, and so are my discharge simulations. Second, there is a discrepancy between the sediment concentrations in my output.rch file and my custom output file for hourly simulations. This is driving me insane, as I can't for the life of me figure out why. The discharge values (FLOW_OUT) are exactly the same in the two files, but SEDCONC is entirely different. SEDCONC in output.rch is based on SED_OUT and FLOW_OUT in the same file. This means that either SED_OUT from output.rch is wrong or SED_CONC from the custom print below is wrong (the HYD_NUM is correct, as indicated by the correct discharge). I've tried for diffrent HYD_NUMs with the same result. I use IPRINT = 3 in file.cio for hourly output in the output.rch file.

saveconc      14    57     2     1

          sub_18_hr.dat

I don't expect you to solve this for me, just wanted to give an update. 

On Tuesday, May 14, 2019 at 8:32:10 PM UTC+2, Jim Almendinger wrote:

Turning off channel degradation (I think) only means that SWAT does not track how channel erosion or deposition affects bed elevation or slope.  Without the channel degradation routine, SWAT normally assumes the channel is an infinite source or sink for sediment.  

I often try to minimize channel processes to start with, so that they don't obscure sediment yields (erosion) from the uplands.  You can (in theory) shut off channel erosion by setting the erosivity and cover factors to zero.  To minimize deposition, you can increase spcon by an order of magnitude (or two, I think), to increase transport capacity in the channel.  If this works, then sediment in your reaches is mostly all from the uplands (hrus and subbasins).  

However the SWAT default parameters (especially spcon=0.0001) often result in some channel deposition.  You should check what your channel deposition is by checking the values of sed_in minus sed_out for each reach (and add them up to get the net watershed-wide channel deposition or erosion).  Sed_in minus sed_out should be close to zero for each reach, if channel erosion and deposition are minimized.  It could be that your reaches are trapping sediment, making your sediment yields appear smaller than they really are from the hrus and subbasins.  

Further (as a first step, really), check if your channel widths are reasonable.  I've generally found that ArcSWAT's estimate of channel widths are about 3 times wider than actual widths.  So I routinely divide my initial channel widths by 3 before calibrating.  When you decrease the channel widths, the cross-sectional area decreases and flow velocities increase.  That will also reduce sediment deposition in the channels, even without increasing spcon.  

Hope that helps,

-- Jim

---

James E. Almendinger, PhD Director, St. Croix Watershed Research Station 16910 152nd St N, Marine on St. Croix, MN  55047 e: jalmen...@smm.org o: (651) 433-5953 x 11

On Tue, May 14, 2019 at 7:07 AM Bendik Hansen <marek...@gmail.com> wrote:

Hello,

I am working on simulating sediments on an hourly time-step in a catchment. I am calibrating on observed suspended sediment concentration in the river. 

My problem is that in order to get the right values for my simulated concentration, I have to tweak my parameters so that I get a very high upstream sediment yield. The Sediment Delivery Ratio (upstream yield/sediments going out of reach) is typically in the range 5-10% for my simulations, while my observed data and other studies indicate it should be closer to 20%-25%. 

Does anyone know which parameters I should be adjusting to increase SDR? There is apparently a lot of erosion going on in my simulation, it's just that the sediments never reach my river (I do not have river deposition/erosion enabled).

Best regards,

Bendik

--
You received this message because you are subscribed to the Google Groups "SWAT-user" group.

To unsubscribe from this group and stop receiving emails from it, send an email to swat...@googlegroups.com.

[Bendik Hansen]

CORRECTION: The problem is when printing daily output to output.rch and comparing it to the hourly output from the custom file (or from output.rch with hourly output, for that matter). The daily discharge in output.rch is identical to the daily values from sub_18_hr.dat when aggregated to daily values. Sediment is not, however. When output.rch is printed on an hourly basis both are identical. I don't know where SWAT is getting the daily values for sediments from when performing hourly simulations and printing daily output to output.rch. They are not based on the average of the hourly values (unlike discharge, which is).