Spiky Hydrograph with low Baseflow

[Max Brunke]

Dear all, 

I have read about this problem multiple times so i was wondering about your solutions. 

I am recently trying to get a less spiky hydrograph with higher low flows.

But I am kinda lost in the calibration parameters...

I already tried cn2 and alpha_bf. 

How did you solve this problem?

Best, Max B

[Andy Turner]

Hello

I am modelling a mountainous region where the flow in the river according to observational data has been far exceeding that which trickles down from the model which I think is purely based on the precipitation and then water balance within the catchment. I think this is because the ice towers are melting in the region I am modelling and that somehow I need to add data on how much ice there is and what the rate of melt is.

Does the total predicted outflow over the period you are modelling have a similar magnitude to that of your calibration/validation data?

What type of region are you modelling?

Best wishes,

Andy

  

--
You received this message because you are subscribed to the Google Groups "SWAT+" group.
To unsubscribe from this group and stop receiving emails from it, send an email to swatplus+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/swatplus/28bfa148-976c-4b83-ade4-20cfe0554ac5n%40googlegroups.com.

[Andy Turner]

For the mountainous region I am modelling, the SWAT model hydrographs were much more spikey than the calibration/validation data. I did some sensitivity analysis and made an effort to calibrate, but the model remained a poor fit. As well as adding in meltwater flow, I also considered disaggregating the precipitation data I have. Currently I am using modelled climate data that is a raster at ~ 60km x 75km resolution. There are around 30 such cells that cover the study area. Within each cell the elevation can vary significantly. One thing I thought to do was to disaggregate the precipitation data further and have more precipitation fall in higher areas. At least, I thought to try it to see if it made a difference to the general shape of the hydrograph. It makes sense that there will be more precipitation at higher elevations which is typically also further from a channel.

I've found that I can get a much closer fitting shape of hydrograph using Neural Networks.

Anyway, if you are in high mountains where there is a lot of ice, then you might like to look at some of the ice melt parameters too, but I don't think these are as easy to adjust as some of the others.

I suspect that SWAT works much better in less mountainous - more lowland areas, but I'm no expert.

HTH

Andy  

[Max Brunke]

Hi Andy, 

Thanks for the great input!
My overall waterbalance and SurQ look alright. 

I am modelling a low mountain to lowland catchment in germany where ice/snow melt shouldn't effect my model that much. 

But there are 10 big reservoirs in my catchment and I am currently working on a better parametrisation!

Best, Max

[Andy Turner]

Thanks Max,

Reservoirs add further complication. Ideally you want data on reservoir levels, extraction and policy for each reservoir. There may be a policy of minimum flow from a reservoir - a flow guaranteed until the reservoir runs below a certain level. It may be that as a reservoir begins to get full, that more water is allowed to leave. At some level the reservoir is full and all water coming in is matched by the extraction and water coming out.

I also tried varying ESCO, OVN and PERCO.

It might be worth doing some sensitivity analysis to try to find out which parameters your catchment is most sensitive to. Each of the parameters can be moved one way or the other to produce flatter or more spikey hydrographs (it should be clear from the description what the effect is likely to be - basically the more you slow the flow, the less spikey the hydrograph should become - at least, that is the logic I tried to apply).

If it gets colder than freezing then I suspect the model will have more precipitation falling as snow. To my mind, this should slow things and stop them being more spikey. If you think your catchment has a lot of snow at times, then you might try also adjusting the temperature data to make it seem colder. It is only worth trying this if there is a lot of snow at times though...

If you are using weather station data based on observations, then I don't think you should adjust the climate variables, but if you are using modelled climate data, then this might be worth considering.

As well as ice melt another thing that could change things a lot is water springing out from aquifer that is charged from other catchments. This groundwater flow comes from a broader catchment. The catchments that are HRUs are defined based on surface topography and surface drainage.

Cheers,

Andy  

To view this discussion on the web visit https://groups.google.com/d/msgid/swatplus/811500d6-b9e5-4733-98c3-96ff42c43079n%40googlegroups.com.

[Andy Turner]

The nature of material that surrounds a reservoir/lake will also influence how much water seeps out from the water body underground. This might be what results in more steady river discharge more than anything else in your catchment. I am not sure if modifying PERCO increases varies this or not...