TOPMODEL - m parameter

[Raphaël]

Hi everybody !

I really need help regarding the hydrological model of CAESAR-Lisflood. 

I just noticed that there is very little difference between simulationsin terms of water flows whatever the m variable (topmodel) value is. 

I have attached to this mail the discharges obtained with spatially and temporally constant values of 0.02 (forested basin)  and 0.005 (deforested basin).  I was expecting the difference to be much larger in terms of behaviour of the modelled flood hydrographs with much lower flood peaks and slower rate of recession of the hydrograph for the 0.02 value (figure 1).

I wonder if I have set up the hydrological model correctly (figure 2). 

I have tried to use a time varying M file with a constant value by thinking that the value in the box was not enough, but it does not change the results. 

What I have done wrong ?Note that the discharges collect at the outlet are daily. Does the daily nature of the water flows would risk to hide the influence to the m parameter ? 

On the other hand, I am a bit confused by the option : 'if checked, discharge is read direct from rainfall file'. Does the activation of this option lead to the desactivation of TOPMODEL and the m parameter ? I seems that the discharges are the same whatever the value of the m parameter. On the other hand, water flows seem to be approximately equal to the value of rainfall (figure 3)

What is the interest of this option ?

Thank you for your help.

Raphaël.

[Tom Coulthard]

Hi Raphael - that might be correct. You're using daily rainfall values - so low intensities spread out over 24 hours means there may be little difference in the output...

Do you have anything of smaller resolution to drive the model?

Tom

--
You received this message because you are subscribed to the Google Groups "caesar-lisflood" group.
To unsubscribe from this group and stop receiving emails from it, send an email to caesar-lisflo...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

<Fig1.PNG>

<Fig2.PNG>

<Fig3.PNG>

[Raphaël]

Hi Tom,

I noted that for the simulation results posted in the previous message, the water balance between the rainfall input and the flow output are not respected with far more water that leaving the catchment over a year (input (900mm) << ouput (15000mm) - which might be physically impossible). 

So, it seems that the model is not managing the daily data properly.

By contrast, when I break the daily data in hourly data, i.e. divide by 24 the daily rainfall amounts and allocate this value over 24 hourly time steps for conserving the water amounts (wich leads as a result to 24 times more lines in the rainfall file), the water balance is coherent with output lower than input (~900 mm input against ~800 mm output) (the output are lower than the input because the subsurface flows produced by TOPMODEL are lost, isn't it ?)

Though, in absolute terms, it should not produce any change because the daily amount of rain is always the same.

In addition, with this configuration, the m parameter influence is clearly visible with much lower flood peaks and slower rate of recession of the hydrograph for the 0.02 value (figure 1).

Sounds inexplicable, doesn't it ?

I netherless expect that the annual runoff volume is much lower for the forested scenario (m=0.02) than for the desforested scenario (0.005) with thus higher infiltration. The annual runoff coefficient remains very high for the forested scenario (~90%). It is equal to ~95% for the deforested scenario.

In contrast the smallest flowrates for the forested scenarios are not systematically reflected in the sediments yields with higher sediment yields for the forested scenario for some period of times (figure2).

This would not make sense.

What it is due to ? Did I do something wrong ?

Thank you very much Tom for your patience and advices and please excuse the length of this message.

Kind regards,

Raphaël

[Tom Coulthard]

Are you accounting for the basin area in your calcs comparing Qw to precipitation inputs?

[Raphaël]

Yes, I take into consideration this information.

But, the water balance appears correct when I divide the initial rainfall amounts in mm/j by 24 (rainfall rate) to obtain amounts in mm/h and mataining the time step to 1440minutes (herein, the number of lines in the rainfall file is not multiplied by 24). Thus, the model appears to realise the operation that I have did manually, i.e apply 24 times the rainfall rate over one day. I get very close results (almost the same) with this two methods.

This aspect seems to be resolve.

Thus,  I get flows consistent with m paramter (vegetation percentage). But what bothered is that the sediments yiels are greater for the forested scenario than for the deforested scenario during recession limb, which offsets the lower sediments yields during floods and results finally in total sediment yields very similar for the forested and the deforested scenario (figure 1 in my previous message)

I have seen in your studies "A cellular model of holocene upland river basin and alluvial fan evolution" (2002) and "Modelling long term basin scale sediment connectivity, driven by spatial land use changes" (2017) that you have used the m parameter to test the influence of vegetation, as I try to do.

But I see that you have sediment yields very different for forested and deforested scenario which is logical.

I don't understand why my results are different from your own.

Have you used the vegetation growth model while it was most suitable for the reach model as you have said to me in a previous message ?

On the other hand, I am not sure about the role of the 'max velocity used to calc Tau from' option ? What purpose does it serve ? Have you used this option ?

Thank you very much for your help,

Raphaël

[Tom Coulthard]

[Tom Coulthard]

OK – text to go with the image in the earlier reply

 

The reason there is a difference between the volume of water output and input is due to the in-out difference. In the image – the blue line is the actual discharge and the pencil line is modelled. It gets the peaks right, but the low flows are influenced by the in-out difference. In other words the model speeds up when the difference is below the in-out difference, meaning that the hydrograph does not drop as fast. This leads to an over prediction of low flows (the shaded area) which leads to there being more water in volume output than input.

[Raphaël]

However, I have set up this parameter to 0m3/s.

I have attached the xml config file, the rainfall record and the dem refer to.

That would be great if you having a look at that and give your feedback about my configuration.

[Raphaël]

However, I have set up this parameter to 0m3/s.

I have attached the xml config file, the rainfall record and the dem refer to.

That would be great if you having a look at that and give your feedback about my configuration.

Le mardi 16 mai 2017 12:59:00 UTC+2, Tom a écrit :

[Raphaël]

On the other hand, the erosion deposition differences between the intial dem and final dem (5370 days ~14.7 years) seem to be disproportionately high with erosion and deposition up to 27.19m and 9.42m respectively (figure).

[Tom Coulthard]

OK – so I just ran your sim (for a short period – 373 days just to test the hydrology)

 

My figures are:

Rainfall. 0.038m total * 24 (as its mm/hr) * 30 * 30 (the grid size) * 60501 (number of cells in DEM) = 18 522 889 358

 

Discharge 350 total * 60 * 60 * 24 (to give a total per day) * 373 (days) = 11 279 520 000

 

So quite a bit less…. Which is what I would expect…

 

Now erosion.

 

20m of incision might sound like a lot – but its in only a few places – and your catchment is 9km long, with a 0.5m drop – gradient of 0.0555’

So that’s pretty steep (gradient = energy of course). Additionally, your coarsest grainsize is 0.002m.. coarse sand. So, if you put even a moderate amount of water down these slopes – you’ll get lots of erosion!!!

[Raphaël]

Thank you very much Tom. 

Yes, it clear that with such slopes, there is many erosion.

Finally, a last interrogation. Have you used - in addition to the m parameter - the vegetation growth model in Coulthard & Van De Wiel (2017) and Coulthard et al. (2002) to reflect the vegetation influence ?

Thank you again for your time and effort.

Raphaël