Conceptual model, 3D discretization

MOHID Land forum. Questions and discussion related to MOHID Land
epelde.ane@gmail.com
MOHID Advanced user
Posts: 172
Joined: Thu Jan 01, 1970 00:00

Conceptual model, 3D discretization

Postby epelde.ane@gmail.com » Tue Aug 23, 2011 08:13

Hi again all,
Once run the model with the simplest options, I´d like to follow defining the horizons and layers in the grid. I understood that I´d need a horizontal grid for each horizon, but I´m a little bit lost doing it, I can´t see the way of simplifying the geology. I ´ve seen different ways in modeling books but I´m not sure how do I have to do it in Mohid. Is there any information or guide for this?
Thank you very much,
ane

davidbrito
MOHID Advanced user
Posts: 190
Joined: Thu Jan 01, 1970 00:00

Re: Conceptual model, 3D discretization

Postby davidbrito » Fri Aug 26, 2011 19:18

Hello Anne,

This a long post but
I am publishing some text that i have written in e-mail when the forum was down (that can be useful for other users), and then I will try to answer more specifically


e-mail about layers and horizons in Mohid land:
<<

Code: Select all

There is a difference between i) soil layers and ii) soil horizons.

 

The second are the real soil differences in vertical that can have different textures (sand, clay, …). Usually we do not have vertical profiles on soil so that we can differentiate horizons and we use soil homogeneous in vertical. But remind that it can be helpful to differentiate vertical horizons if your horizons are different in soil hydraulic parameters. If they are very similar, creating horizons or leaving it homogeneous in vertical will not have a great impact.

 
 
The first is the model vertical distribution. As you know MOHID works with cells. With DTM you define the horizontal grid, in geometry you define the vertical grid defining the several cell thickness.

 

Another important factor is soil depth. In Mohid Studio it is computed from slope but you should inspect if soil depths (depth of the porous media) are consistent with what you know about your watershed. If they are not you can change soil bottom grid (soil bottom altitude) or create a new grid from your knowledge and give that to the model.

 


So now we can talk about defining layers in MOHID. As you know layers at surface should be smaller and gently increase up to the bottom (something like from 5cm thickness in surface up to 1m thickness and after arriving 1m maintain 1m up to the bottom).

The model starts distributing vertical layers from soil surface until it reaches the soil bottom (defined by the soil bottom grid) and it there is no problem if in any part of the watershed you have less depth than all the layers that you define. the model will ignore layers that are below soil bottom. But there is a problem if in any part of your watershed there is places where soil depth is higher than the maximum depth of all layers that you define in geometry because the model will not know what thichness should use below. So, the sum of all layer thickness (defined in geometry) has to be higher than the maximum soil depth that you have.

 
 
So now we can talk about defining soil horizons in MOHID. Soil horizons we already discusses this in previous MOHID Forum posts.

You have to define for each layer range what will be the horizontal grids distributions of types of soil.

 

Now you have to see what kind of layer thickness distribution is more suited for the real horizon distribution (reminding that should be small at surface and not go higher than 1m) and create the different horizontal grids with soil distribution (e.g. a grid with surface soil (from surface up to 1m) an a grid with lower soil (from 1 to 2m) and a grid with even lower soil (from 2m and on), I do not now it is just an example I do not know if it suits.


>>


Assuming that you have the layer thickness defined in geometry.dat,

to define soil horizons (in PorousMedia.dat) I link back to the post called "Data needed in the Porous Media module":

<<

Code: Select all

Let's assume that the top horizon for the first soil goes from layer 7 to 10 and the bottom horizon goes from 1 to 6, and that this distribution is the same for the entire area

!Horizon 1 - Top
<beginhorizon>
  KLB : 7
  KUB : 10

  <beginproperty>
    NAME                  : SoilID
    INITIALIZATION_METHOD : ASCII_FILE
    FILENAME              : GridDataSoil_1.dat
    DEFAULTVALUE          : 1
  <endproperty>

  <beginproperty>
    NAME                  : Theta
    INITIALIZATION_METHOD : CONSTANT
    DEFAULTVALUE          : 0.38
  <endproperty>
<endhorizon>

!Horizon 2 - Bottom
<beginhorizon>
  KLB : 1
  KUB : 6

  <beginproperty>
    NAME                  : SoilID
    INITIALIZATION_METHOD : ASCII_FILE
    FILENAME              : GridDataSoil_2.dat
    DEFAULTVALUE          : 1
  <endproperty>

  <beginproperty>
    NAME                  : Theta
    INITIALIZATION_METHOD : CONSTANT
    DEFAULTVALUE          : 0.38
  <endproperty>
<endhorizon>

>>

Layers 1 to 6 and 7 to 10 represent the layers defined in file Geometry.dat (the vertical distribution) and of course they will have a direct link to depths. For instance layer 7 to 10 could represent soil depths from surface to 1m (if the sum of this 3 layers thickhness is 1m) and layer 1 to 6 soil depths from 1m up to the maximum depth. Remind that z axe in Mohid starts in bottom so layer 1 is the bottm most and 10 is soil surface.

Meaning that in the example you have to create two grid data files, one called GridDataSoil_1.dat and GridDataSoil_2.dat. This grid data files (2D) have as center values for the cells the soil ID. This means that they define the 2D spatial variation of soil type one for layer 1 to 6 and the other for layer 7 to 10 (here it is the vertical). This example uses 10 as maximum layers and that has to be the same number of layers as defined in Geometry.dat
The grids can be created using the model horizontal grid and i) points, or ii) shape files. This is just like any grid data creation similar for DTM or for land use grid in vegetation.

Hope that it helped. Maybe too much text and many topics involved but now we can move in the good direction with your replies.

David

epelde.ane@gmail.com
MOHID Advanced user
Posts: 172
Joined: Thu Jan 01, 1970 00:00

Re: Conceptual model, 3D discretization

Postby epelde.ane@gmail.com » Mon Aug 29, 2011 10:33

Thanks for the reply david,
Sorry,I remember these explanations from previous posts, but I think I didnt explain myself clearly,
I can see the way you discretize in vertical in your example, but what happens when soils have different depth in horizontal? Do I have to make more horizontal grids or is it better if I simplify and give a mean depth for all the area?

And, looking to the sigma, cartesian..geometries I see that the upper layer is flat. Is it the water table? so, how does it take into account the layers above? I mean, that is not the same if above it we have 1m clay layer or 4m layer.. Another thing I dont understand is, that if the topography is not flat, how does mohid fit it with the layers?

I had defined a cartesiantop geometry because I had read that it was the one used for mohid land. is it better if I use cartesian geometry rather than cartesiantop?

Best regards,
ane

davidbrito
MOHID Advanced user
Posts: 190
Joined: Thu Jan 01, 1970 00:00

Re: Conceptual model, 3D discretization

Postby davidbrito » Tue Aug 30, 2011 11:22

Hello Anne,

there are some things that are being misunderstood so nothing better than an image to show how mohid handles vertical distribution (layers and horizons) - see the picture below.
Ando also how it should look like the Geometry_x.dat and PorousMedia_x.dat for this example (behind the hood)



MohidLandSoilProfile1.png
MohidLandSoilProfile1.png (60.63 KiB) Viewed 6375 times



The Geometry_X.dat should look like something like this for this example (8 layers example, in the figure only appears areas of the watershed where maximum 4 layers exist, but in the whole watershed may exist a area with 8 layers and 3m depth - sum the layer thickhness and you get the maximum depth allowed in layer definition)

Code: Select all


MIN_LAYER_THICKNESS       : 0.2
<begindomain>
ID                                      : 1
TYPE                                 : CARTESIANTOP
LAYERS                             : 8
LAYERTHICKNESS               : 1.1 0.5 0.5 0.3 0.3 0.2 0.1 0.1
<enddomain>

! remind that Mohid Land vertical axe is from bottom to surface so first layer is in bottom and last layer is in surface.




The PorousMedia_X.dat should look like something like this if the soil horizons were similar in space (horizontal variation). But of course soil distribution may vary in space and so you can give an ascii file instead (2D grid for each horizon).
The initial (start of simulation) aquifer may also be obtained, as the impermeability (the must exist options that appear in PorousMedia_X.dat that are not discussed here.

Code: Select all

BOTTOM_FILE               : ..\General Data\Other\PorousMedia\SoilBottom\SoilBottom_Example.dat

OUTPUT_TIME                 : 0 864000.
THETA_HYDRO_COEF       : 0.98
CUT_OFF_THETA_HIGH    : 1e-15
CONTINUOUS                    : 0
START_WITH_FIELD         : 1

<beginsoiltype>
ID                        : 1
THETA_S                   : 0.3859
THETA_R                   : 0.0476
N_FIT                     : 1.39
SAT_K                     : 3.5556e-6
ALPHA                     : 2.75
L_FIT                     : 0.50
THETA_CV_MIN              : 0.2844
THETA_CV_MAX              : 0.3791
<endsoiltype>


<beginsoiltype>
ID                        : 2
THETA_S                   : 0.3859
THETA_R                   : 0.0476
N_FIT                     : 1.39
SAT_K                     : 3.5556e-6
ALPHA                     : 2.75
L_FIT                     : 0.50
THETA_CV_MIN              : 0.2844
THETA_CV_MAX              : 0.3791
<endsoiltype>


!----- Hydraulic Soil Properties

!Bottom Horizon (remind that Mohid Land vertical axe is from bottom to surface so first layer is in bottom and last layer is in surface)
<beginhorizon>
KLB                       : 1
KUB                       : 6
<beginproperty>
NAME                      : SoilID
DEFAULTVALUE              : 1                                    !here the bottom horizon will have the soil type number 1 (remind constant in space)
INITIALIZATION_METHOD     : CONSTANT
REMAIN_CONSTANT           : 1
<endproperty>

<beginproperty>
NAME                      : Theta
INITIALIZATION_METHOD     : CONSTANT
DEFAULTVALUE              : 0.25
<endproperty>
<endhorizon>

!Surface horizon (layers 7 and 8) remind that this has to be consistent with the maximum layers defined in Geometry_x.dat
<beginhorizon>
KLB                       : 7
KUB                       : 8
<beginproperty>
NAME                      : SoilID
DEFAULTVALUE              : 2                                     !here the bottom horizon will have the soil type number 2 (remind constant in space)
INITIALIZATION_METHOD     : CONSTANT
REMAIN_CONSTANT           : 1
<endproperty>

<beginproperty>
NAME                      : Theta
INITIALIZATION_METHOD     : CONSTANT
DEFAULTVALUE              : 0.25
<endproperty>
<endhorizon>


<beginwaterlevel>
NAME                      : waterlevel
INITIALIZATION_METHOD     : ASCII_FILE
DEFAULTVALUE              : 0
REMAIN_CONSTANT           : 0
FILENAME                  : ..\General Data\Initial Conditions\PorousMedia\WaterLevel\InitialWaterLevel_Example.dat
<endwaterlevel>

<beginimpermeablefraction>
NAME                      : impermeablefraction
INITIALIZATION_METHOD     : ASCII_FILE
DEFAULTVALUE              : 0
REMAIN_CONSTANT           : 1
FILENAME                  : ..\General Data\Other\PorousMedia\Impermeability\Impermeability_Example.dat
<endimpermeablefraction>



Now you have to know that cartesiantop is called like that because the model goes to every horizontal cell and tries to put the user desired vertical geometry (Geometry_x.dat) but starting in TOP. when it reaches bottom soil it stops and it does not mind that it were layers defined that were not distributed. after finding bottom the model knows that does not need to distribute layers even if they were defined in the Geometry_x.dat


Now I STRONGLY RECOMMEND that you start with simple horizon distributions and soil values in that horizons (constant or grid) and progressively rise the complexity. Because you learn and get to know how this stuff works and probably you are aiming to something hard for you to implement (a lot of horizons, a lot of grids to build) that will have very limited impact in your results.

You can use a constant depth soil but that is not needed (if you want to test the difference, be my guest) but can use a variable depth soil computed from slope (as Mohid Studio computes).


I think I have answered it all. Maybe not. Let me know.
Best regards,
David


Return to “MOHID Land”

Who is online

Users browsing this forum: No registered users and 2 guests