Hi Priyamitra,

1st- To choose your correct set of parameters you can see for instance papers from literature database (https://www.card.iastate.edu/swat_articles/ ). You can see papers with similar purpose (discharge and sediment).

2nd- If you have chosen your set of parameters, now you have to select their variation ranges (min and max). Normally people don´t know real values to their sets of parameters and fixed the ranges (min/max) suggested by swat-cup (I add image 1).

Don´t worry because these initial ranges are not accurate to your study area. You will obtain the important ranges at the end of the calibration process (after all your simulations and iterations).

Remind you have to choose in your parameters if you want a relative variation or no. for example in this case:

a__GWQMN.gw

v__GWQMN.gw

you see the same parameter but we have two changes methods. You can see these concepts in the official manual of SWAT-CUP: http://swat.tamu.edu/media/114860/usermanual_swatcup.pdf

on the other hand, in the middle of the calibration process, you have to select an objective function (usually for discharge NSE is accurate).

Best,

antonio   

Hello Priyamitra,  

To do the calibration and sensitive analysis:  Are you going to use SWAT-CUP with SUFI-2???

With regards,

antonio

Hi Antonio, 

Thanks for replying 

Yes i will use SWAT-CUP with SUFI-2

with regards,

PM

Hi Priyamitra,

1st- To choose your correct set of parameters you can see for instance papers from literature database (https://www.card.iastate.edu/swat_articles/ ). You can see papers with similar purpose (discharge and sediment).

2nd- If you have chosen your set of parameters, now you have to select their variation ranges (min and max). Normally people don´t know real values to their sets of parameters and fixed the ranges (min/max) suggested by swat-cup (I add image 1).

Don´t worry because these initial ranges are not accurate to your study area. You will obtain the important ranges at the end of the calibration process (after all your simulations and iterations).

Remind you have to choose in your parameters if you want a relative variation or no. for example in this case:

a__GWQMN.gw

v__GWQMN.gw

you see the same parameter but we have two changes methods. You can see these concepts in the official manual of SWAT-CUP: http://swat.tamu.edu/media/114860/usermanual_swatcup.pdf

on the other hand, in the middle of the calibration process, you have to select an objective function (usually for discharge NSE is accurate).

Best,

antonio   

Hi Antonio, 

Thank you for shearing this information. I will go through as you said and let you know the results.

Thanks once again 

with regards 

PM

Dear Priyamitra,

Antonio gave you already some helpful advice to start your calibration. I thought I might add a few things that I experienced over the years with calibrating SWAT models.

Selection of parameters:

With the selection of the parameters and their boundaries I think you can be more generous in a first step. See the first few hundred (maybe thousand) simulations as a first sensitivity analysis. If you use SUFI-2 the results provide a parameter ranking and the sensitivities for each parameter. You can select now the parameters, that are sensitive in your model. This is however very subjective to set a threshold for sensitive and non sensitive parameters.

Parameter boundaries:

As Antonio suggested, if you have no good starting point for the boundaries, you can start with the absolute possible parameter range given in the Absolute_SWAT_Values file. Again if you use SUFI-2 these ranges converge and become narrower after each iteration step (the more sensitive a parameter is, the quicker this happens).

Type of parameter variation:

as already mentioned you can use either v__, a__, and r__ for parameter variation. A rule of thumb here is (or it is basically what I do) to use a relative change or the addition of an absulute increment to parameters that you distributed "spatially" in your model. Typical parameters here are CN2, CNOP, K, AWC, etc... (as you can also see in Antonios screenshot). Other parameters that are either basin parameters or have no spatial distribution (e.g. all have the same starting value (default value) for each subbasin or HRU) you can modify by replacement (v__) as well.

The choice of the objective criterion, simultaneous or sequential calibration etc. are further issues where you have to make decision, that we maybe could discuss for ages :)

I hope my explainations are a helpful contribution here

Cheers Christoph

Hello Christoph, 

Thank you for your valuable suggestions. Now i have identified all the parameters related to discharge and sediment and also their boundaries from the SWAT Input-Output document and from  Absolute_SWAT_Values file as you said.

Now i have to start the calibration using SUFI-2. Here i have one doubt can i select all the parameters for both the variables (discharge and sediment) or should i do calibration separately for flow and sediment??

With regards

PM   

Yes that is indeed a very essential question. I myself prefer a simultaneous calibration of different variables as the sediment transport might be strongly influenced by the quality of your discharge calibration on one hand and can contradict the calibration of some parameters for discharge on the other hand. Therfore in general I prefer to calibrate both variables at the same time and accept the tradeoff you might have for the modelling results of the individual variables.

This means:

1. Either you calibrate first discharge and in a further step sediments (and then P and N if these are your target variables as well) and you only use parameters that affect discharge in your first step and only parameters that affect sediment in your second calibration step or

2. you calibrate both variables at the same time with a weighted objective criterion e.g. NSE for discharge and sediment and each has a weight of 0.5. In this case you use the full parameter set you selected for the simultaneous calibration of discharge and sediements

So, i can do calibration simultaneously  both for discharge and sediment but the template given in SWAT CUP to enter the observed data is only for discharge i.e. (Flow_Out_1) in Observed_rch.text. I am not getting the Sediment template. Where should i give the observed sediment data??

Hello Friends,            

 I used Global weather data (1979-2014) for my watershed. i ran the model using these data and got the output precipitation = 41000 mm and the runoff = 38000 mm. which is too much high values. I did not understand why SWAT  giving these high values.

 please suggests me how to fix this problem??

Hello,

where did you get these values from? Are these long term annual values shown in the SWAT checker, or did you maybe sum up the values from the output.rch file. Please provide some more information so that we know where to start. Despite the magnitude of the variables, a runoff coefficient of over 90% is too large for most situations.

I got these values from SWAT checker. Please find below the image 

OK these are most likely nonsense values. The first thing that comes to my mind is to check the regional settings of your computer and the comma sign in your input data (e.g. "," instead of "."). This could maybe explain why your precipitation is too large by an order of magnitude

OK i will check my computer settings and the input data formats.

one more thing which i was thinking is it because of WGEN file?? i used WGEN_user option in weather generator data which was showing station count 4    

Hello Priyamitra,

I have seen your swat-check values. I got similar errors many months ago, because I didn´t introduce anything in weather dada definition/weather generator data/WGEN_user.

If you select the option “WGEN_user”, you see “station count: 4”. (image1).

These four stations are from USA, if you are in other country you cannot use these stations. These stations are in your SWAT2012.mdb (image2).

You have to introduce other file (file 5) with these climate parameters but for a station of your country. You don´t need to remove the previous files from USA. Your WGEN_user only will use values from the station closest to your watershed.  

To calculate these climate parameters (needed to use your WGEN_user), you can see:

-http://swat.tamu.edu/software/links/ (image3).

-https://www.youtube.com/watch?v=d-2CljcpzEU&index=6&list=PLVU_9Q25QQmix9MTet9sGtNAgw0VnSVpT (Calculate the statistical parameters of weather data for SWAT). This video helped me a lot.

Best, Antonio.

Hi Antonio,

Thanks. I was also thinking that it is related to WGEN_user file. I have one WGEN_user file for my watershed but that is made by IMD_data (1979-2005). Now i am using the CFSR Global weather data so can i use that IMD WGEN_user file with this Global weather data??  Or should i prepare new WGEN_user file using the Global weather data??

Please suggest me

Regards  

I think I don´t know. We calculated the WGEN_user file using climate stations with real data (in Spain).

Sorry :((

OK No problem, Thanks for replying 

As i said i am using the CFSR grid data. In that data i have 35 points (including outside the watershed) so i want to know that how many points should i use to generate the WGEN_user file??

Please tell me how to calculate the Skew coefficient, Probability of a wet day and Maximum 0.5 hour rainfall??

Sorry Priyamitra,

I don´t Know these steps. But did you search it in these links ??:

-http://swat.tamu.edu/software/links/ 

-https://www.youtube.com/watch?v=d-2CljcpzEU&index=6&list=PLVU_9Q25QQmix9MTet9sGtNAgw0VnSVpT

Best, Antonio.

Thank you Antonio, 

i searched the above links and understood the process but the CFSR data which i am using don't have dew_point and Maximum 0.5 hour rainfall data. 

If u know any link about these data please shear it

Dear all

i have run SWAT model and got the simulated and observed discharge relation for two outlets as shown in figure. R2 value for outlet 1 and outlet 2 are 0.3947 and 0.2587 respectively. please suggest me what should i do afterwards for calibration and validation in SWATCUP Sufi2. Iwent through SWAT CUP manual but could not get the idea on how to select parameters, start with what parameters and when to add other parameter?

i have measured sediment data of 6 month for outlet 1 also.

Dear Shekhar,

Which climate data are you using?? You have done daily simulation. Do monthly simulation for discharge you will get better R2 value.

I used station data of rainfall and swat database data for others.

There is no good method to estimate without some hourly local data. Else one could assume taking 1/3 of max daily rainfall of a month. (R. Srinivasan)

https://groups.google.com/forum/#!topic/swatuser/kpZMCke3BvU

good question,thank you very much, and thank you for all this answer too

1. For manual calibration you can use manual calibration helper in SWAT, Edit input and run swat window.

2. You can choose the outlet point location for your specific watershed.

3. WGEN file is available for CFSR DATA also you can use that directly.

4. It depends on for how long is your simulation period. Two months data may not be enough for calibration but you can try and compute the statistics.

Hope this will help you

Best

Hi,

My R2 value is 0.86 but the simulated runoff is lower than the observed runoff. I run only for 5 number of simulation SWAT-CUP SUFI2.