Sediment calibration / TSS

[Zeynep]

Hi everyone!

I am trying to calibrate sediment and I only have the observed TSS (mg/L) values (monthly). I run the model and I compare the observed TSS values with the values that I read from the TablesOut/sed/TSSmg/L. Then, I change a parameter related to sediment like SPCON, SPEXP.. but simulated TSSmg/L values do not change ( I always rewrite Swat input tables after making changes). What should I do to calibrate sediment? There is also SED_OUT value given by the model. Can I compare my TSS values with SED_OUT (tons) after converting the observed TSS values to tons (TSS (mg/L) * flowrate (m^3/s) * 30 days * 24*3600seconds /10^6)?

Thanks is advance,

Zeynep

[Jim Almendinger]

Zeynep --

The sp_con and sp_exp determine how much sediment can be suspended given the stream velocity.  If the suspended load is below that value, the excess is deposited in the channel.  So if you don't change anything else, sp_con and sp_exp determine the deposition of sediment. 

If the suspended load is below what the stream could carry, at a given velocity, then the stream channel will erode and contribute sediment -- BUT ONLY if you've changed the erodibility (ch_cov1) and cover (ch_cov2) factors to BOTH be something greater than zero.  Both start out as zero by default, and if either factor is zero, nothing is eroded (their effects are multiplicative). 

Then there are a host of landscape parameters to adjust to change sediment loads to reaches -- soil slopes, slope lengths, soil erodibility, peak flow factors, and USLE_P factors (among other MUSLE factors). 

In general, I turn off channel deposition (by setting sp_con and sp_exp as large as possible) and erosion (by leaving ch_cov1 and 2 = 0), and adjust the landscape parameters to get my landscape erosion to what I think are reasonable values.  Then I add the channel processes back in to fine-tune things (and hopefully not mess things up).  And if you know the split between field and channel erosion from a geochemical fingerprint, you can try to explicitly match those components. 

In theory, you should be able to match up calculated loads from the reach output table with loads that you calculate using concentration X flow rate X time.  I haven't checked this for the sediment loads and concentrations, but it ought to work, or at least give you similar results. 

Good luck,

-- Jim

---

From: "Zeynep" <ozcanz...@gmail.com>
To: arc...@googlegroups.com
Sent: Monday, August 18, 2014 6:37:03 AM
Subject: [ArcSWAT:6507] Sediment calibration / TSS

--
You received this message because you are subscribed to the Google Groups "ArcSWAT" group.
To unsubscribe from this group and stop receiving emails from it, send an email to arcswat+u...@googlegroups.com.
To post to this group, send email to arc...@googlegroups.com.
Visit this group at http://groups.google.com/group/arcswat.
For more options, visit https://groups.google.com/d/optout.

---

Spam
Not spam
Forget previous vote

--

Dr. James E. Almendinger
St. Croix Watershed Research Station
Science Museum of Minnesota
16910 152nd St N
Marine on St. Croix, MN  55047
tel: 651-433-5953 ext 19

[Zeynep]

Dear James Almendiger,

Thank you very much for the detailed explanation. Your explanations will be a guide for me.

Best regards,

Zeynep Özcan

[Jim Almendinger]

Zeynep and others --

Before I forget -- Check your channel widths and depths! 

Channel dimensions (width W and depth D) and Manning's roughness N are rather important in determining stream velocity and hence channel erosion and deposition.  In my mid-sized models (watershed area ~1000 km2), the default W values (as generated by ArcSWAT automatically) were about 3 times the actual W (measured at about a dozen cross-sections).  Actual D was closer, about 0.9*D generated by ArcSWAT.  So, in my model I divided all default W values by 3, and multiplied all default D values by 0.9. 

Before these changes, the flow was too wide, shallow, and slow.  Afterwards the flow depth and velocity were much closer to measured values, especially after I changed N from the default of 0.014 to about 0.025-0.035, closer to natural sandy and gravelly stream bottoms, according to my engineering friends. 

Reducing W causes a substantial increase in flow velocity, since now the flow must pass through a smaller cross section.  This in turn reduces channel deposition and enhances channel erosion (assuming both ch_cov1 and ch_cov2 are >0).  These changes can greatly affect your sediment calibration. 

So -- get your channel dimensions about right before going through the process of sediment calibration, if you believe channel processes are important in your watershed. 

Cheers,

-- Jim

---

From: "Zeynep" <ozcanz...@gmail.com>
To: arc...@googlegroups.com

Cc: ozcanz...@gmail.com
Sent: Tuesday, August 19, 2014 2:41:21 AM
Subject: Re: [ArcSWAT:6511] Sediment calibration / TSS

[Juan Garcia]

Dear Sir/Madam

 

I have a problem with the snow melt. I am modelling a catchment in central Chile where the upper part of the basin, around 4200 masl, keeps the snow for around three or four months and then at the beginning of the spring it begins to melt down. However, the model is not able to capture that behavior and release the water immediately.

 

So, although the mass balance in an annual basis is correct, monthly speaking the release of the water in form of snow melt, is 3 or 4 months early than it should be.

 

Would you be so gentle to guide me or give me some advice about how can improve the timing of the snow melt?

 

Thanks a lot in advance

 

Juan

--

[David Salas]

Did you try putting in snow bands?  

[kostas...@gmail.com]

Dear Jim Almendinger and friends,

I read your comments with interest. You seem to have a lot of experience with the sediment part of SWAT.

I made a test myself setting CH_COV1 and CH_COV2 = 0, as well as K_USLE=0. By doing this, one should expect zero soil erosion and zero strembed erosion, and hence zero sed_out in the output.rch file. However, I still get a considerable amount of sed_out. What other sediment sources are there in SWAT and how can they be controlled?  

Any feedback on this?

Thanks a lot,

Kostas

[Natalja C.]

Hello,

Well, the statement " expect zero soil erosion and zero strembed erosion, and hence zero sed_out" is not entirely correct. There are multiple sources of sediment in swat (land erosion, bank erosion, management operation, point sources, reservoirs, etc.). Refer to the SWAT I/O manual. Just search for "Sediment" and read every encounter. 

First thing to find is: where do your sediments come from? By reading the I/O you can get the idea what parameters to change to see the effects of different sed sources. Try them ONE BY ONE and examine the difference. 

When you identify your source, then just adjust the related parameters to reduce the erosion/load.

Best of luck,

Natalja

[Jim Almendinger]

Kostas --

I do see your point.  To check, in one of my test models, I chose a headwater subbasin to avoid possible complications from sediment being delivered from upstream reaches.  In that subbasin, I set USLE_K1=0 for all HRUs, and that did indeed stop all erosion from the HRUs, and by extension, from the entire subbasin. 

For some reason, there was still a very small amount of sediment coming out of the reach for that subbasin, which must be from some sort of channel or bank erosion.  I had selected erosion equation 1 (the simple standard option), and had both CH_COV1 and 2 set to zero (I would think either set to zero should stop erosion).  But for some reason there is still a little sediment being generated in the channel, and I don't know why.  It is not much and (hopefully) not really a problem, but it means I'm missing something in some of SWAT's details. 

There are other possible sources of sediment -- do you have ponds or reservoirs?  If you set your NSED too large, you can "create" sediment that is discharged from the impoundment.  The D50 parameter also plays a role; I believe too small can allow too much sediment released (rather than trapped) by an impoundment.  Right now that's the only other obvious source of extraneous sediment I can see. 

Good luck -- let us know what you find.

-- Jim

To view this discussion on the web visit https://groups.google.com/d/msgid/arcswat/a50bdb7f-08db-4c2a-af53-ba26504f107en%40googlegroups.com.

[kostas...@gmail.com]

Dear Natalja and Jim, thank you very much for your replies.

Jim, I have put an inlet to my basin, but I have set sediment equal to zero. Only water comes in.

I have come to realize that ADJ_PRK is a quite influential parameter.

Apart from CH_COV1, which is the channel erosdibility there is also CH_BED_KD (bed erodibility) CH_BNK_KD (bank erodibility). The CH_COV2 is the channel cover factor, and when set equal to zero the channel is supposed to be fully protected against erosion. In any case, all these parameters are 0 by default. LAT_SED (sediment concentration in lateral and ground flow) is also zero by default. Still though, sed_out is high.

I will search more on this and let you know if I find something interesting.

Please also share if you find something new.

Best,

Kostas 

[Jim Almendinger]

Kostas --

-- What CH_EQ are you using?  If it is = 0 (Simplified Bagnold default), that should simplify things.  It is not clear to me if the other parameters beyond CH_COV1 & 2 are used, if CH_EQ=0.  I am unsure how SWAT is defining channel vs bed & bank.  I presume the theory manual would clear this up. 

-- The input / output manual gives the equations for default values of the bed and bank parameters -- zero is generally not the default, even if that is what the tables and files show.  This is confusing, but oftentimes when SWAT sees a zero, it interprets it as "the user did not enter anything, so I will use the default."  And, you need to check the in/out manual to know what that default will be.  To get (almost) around this, try entering a very small, non-zero number. 

-- I believe you are correct that ADJ_PKR can be a sensitive parameter in changing erosion in the basin (from HRUs).  I had forgotten.  It is used in the MUSLE equation.  I presume for testing purposes, you can still eliminate all HRU (& thus SUB) erosion by setting USLE_K1=0, if you want to isolate your investigation to the channel / bed / bank alone. 

So I still don't have a great answer for you, and perhaps the bed & bank are the sources of sediment. 

Thanks for investigating!

-- Jim

To view this discussion on the web visit https://groups.google.com/d/msgid/arcswat/f209f714-c4e1-48ec-87e0-53b0d1efe71an%40googlegroups.com.

[kostas...@gmail.com]

Jim, this is another matter. The different stream sediment transport formulas of SWAT give very different results. In a nutshell, the results of CH_EQN 2,3, and 4 are of the same order of magnitude and approximately 10 times higher than CH_EQN 1. I have not tried  CH_EQN 0 recently, but I will.

I will get back to you if I have any interesting findings.

Thanks a lot for your inputs.

Best,

Kostas

[Jim Almendinger]

Kostas --

Yes, I remember the other equations can give quite different results, at least without tuning the parameters.  CH_COV1 & 2 take on different meanings, and I presume transport is sensitive to grain-size settings.  I tried a few of them but backed away when I really had no data to constrain them. 

-- Jim

To view this discussion on the web visit https://groups.google.com/d/msgid/arcswat/d2e513fc-7c1f-4aa0-9fde-1f88b3cd4a4dn%40googlegroups.com.