Alpha and n for specific soil textures
Alpha and n for specific soil textures
>>> Spencer Anthony Uminski <sau@et.byu.edu> 02/12/03 01:04PM >>>
Rien:
I will be pursuing HYDRUS1D and/or 2D. I have a really quick question that I have been trying to find an answer to for a few days. I know that you will be able to give quick and knowledgeable answer.
I have found many tables throughout various reports with alpha and n values for specific soil textures. Some vary greatly, but I keep finding reference to a table produced by Carsel and Parrish (1988). My question is:
Are VG parameters alpha and n specific to a certain soil type (texture), is there an acceptable range of values for each soil texture, or do they need to be calculated for each site with hydraulic variables?
Regards,
Spencer
Rien:
I will be pursuing HYDRUS1D and/or 2D. I have a really quick question that I have been trying to find an answer to for a few days. I know that you will be able to give quick and knowledgeable answer.
I have found many tables throughout various reports with alpha and n values for specific soil textures. Some vary greatly, but I keep finding reference to a table produced by Carsel and Parrish (1988). My question is:
Are VG parameters alpha and n specific to a certain soil type (texture), is there an acceptable range of values for each soil texture, or do they need to be calculated for each site with hydraulic variables?
Regards,
Spencer
Spencer:
Within the HYDRUS codes, and also separately, we provide Neural Network predictions of the VG parameters. They are based on the work by Schaap and colleagues, leading to the separate Rosetta software (http://www.ussl.ars.usda.gov/models/rosetta/rosetta.HTM) and RosettaLite within HYDRUS. The software will let you predict the VG parameters from soil textural class, or from more detailed information (percent sand, silt, clay, bulk density, wilting point,...). We also include the Carsel table in HYDRUS to allow people to compare the Rosetta predictions with the earlier Carsel estimates. Hope this helps.
Rien van Genuchten
Within the HYDRUS codes, and also separately, we provide Neural Network predictions of the VG parameters. They are based on the work by Schaap and colleagues, leading to the separate Rosetta software (http://www.ussl.ars.usda.gov/models/rosetta/rosetta.HTM) and RosettaLite within HYDRUS. The software will let you predict the VG parameters from soil textural class, or from more detailed information (percent sand, silt, clay, bulk density, wilting point,...). We also include the Carsel table in HYDRUS to allow people to compare the Rosetta predictions with the earlier Carsel estimates. Hope this helps.
Rien van Genuchten

 Posts: 4
 Joined: Thu Apr 25, 2002 12:50 am
 Location: USA
I am afraid that I have a bit of confusion regarding the alpha parameter in Rosetta. I am trying to perform a sensitivity to perturbing alpha paramters in my model. I have found a table that list the average value and one standard deviation (reported in parantheses) in Rosetta. The values are reported as log values; for example log 1.29 (0.65). If you were to take the inverse of those values, the result would be 0.051 and (4.46). Am to understand that the standard deviation reports 4.46 above and below 0.051?
I imagine there is probably an embarassingly simple answer to this; nonetheless, I am compelled to inquire. Thank you for your assistance. I have found both Hydrus and Rosetta to be outstanding programs.
Kindest Regards,
Joseph McCarthy
Joseph C. McCarthy
I imagine there is probably an embarassingly simple answer to this; nonetheless, I am compelled to inquire. Thank you for your assistance. I have found both Hydrus and Rosetta to be outstanding programs.
Kindest Regards,
Joseph McCarthy
Joseph C. McCarthy
dear van Genuchten and all,
I have found some extra information in literature, but the question is, is this still valid or are the values different?
In the MvG equations the use of the n, m, l and the alpha were stated to be dimensionless parameters. They are used as fitting parameters in the closed form equations when m= 11/n (n>1). However I found in some literature the have the following physical meaning;
n = a pore size distribution index
alpha = the inverse of the air entry value (or bubbling pressure)
l = pore connectivity parameter
In the study of Brooks and Corey [lit 5.1] the value of l is assumed to be 2.0. In the study of Mualem {lit 5.2] the n is based on a statistical pore size distribution and the value of l is estimated as being 0.5 for many soils.
The equations are based on statistical data of soils and average values of soils. It has proven to be valid for many soils and accurate calculations can be made.
But my question is are the fitting parameters still connected to the physical properties as mentioned before or is this relation in nonexcisting anymore ?
And if they still have a physical meaning can they be used to calculate pF curves (of nonsoils) more accurate?
I know this questions brings us back to the basics but I am still interested in knowing
I have found some extra information in literature, but the question is, is this still valid or are the values different?
In the MvG equations the use of the n, m, l and the alpha were stated to be dimensionless parameters. They are used as fitting parameters in the closed form equations when m= 11/n (n>1). However I found in some literature the have the following physical meaning;
n = a pore size distribution index
alpha = the inverse of the air entry value (or bubbling pressure)
l = pore connectivity parameter
In the study of Brooks and Corey [lit 5.1] the value of l is assumed to be 2.0. In the study of Mualem {lit 5.2] the n is based on a statistical pore size distribution and the value of l is estimated as being 0.5 for many soils.
The equations are based on statistical data of soils and average values of soils. It has proven to be valid for many soils and accurate calculations can be made.
But my question is are the fitting parameters still connected to the physical properties as mentioned before or is this relation in nonexcisting anymore ?
And if they still have a physical meaning can they be used to calculate pF curves (of nonsoils) more accurate?
I know this questions brings us back to the basics but I am still interested in knowing
IJsbrand, others:
Yes, we tend to think that Rosetta is the best, but it sometimes enlightening to compare things with the Carsel and Parrish data to see the level of (dis)agreement.
The equations Brooks and Corey proposed, and our equations, are purely empirical functions in attempts to describe the water retention properties of soils (the same for most or all other equations; one may argue even for Kosugi's lognormal model). They are empirical, even though people have gone out of their way to assign physical significance to the parameters (n, alpha, hb, l, whatever). Defining alpha as 1/hb is such an attempt, which is formally incorrect: (1) many or most soils do not have a welldefined air entry value (that's why we went to that smooth function), and (2) alpha=1/hb is only an approximation (and a very poor for low n values). The same reasoning applied to l, which Mualem estimated in his equation to be 0.5 based on an analysis of some 40 soils (mostly coarse, repacked media). Marcel Schaap found that l on average is closer to 1.0, which does not make sense if l is interpreted as a poreconnectivity (or tortuosity) factor (physically, Se^l probably should decrease, rather than increase, when the water content decrease). Also, the B&C equations (for K) are based on a different poresize distribution theory (Burdine), and the l=2 value should have no relationship with Mualem's l=0.5. Actually, our equations can be combined also with Burdine's theory (giving the approach Haverkamp and a few others prefer), while the B&C retention function can be combined with Mualem's equation. All those things are worked out in detail in the RETC manual.
One can take this point of empiricism even further with how WCR and WCS are viewed (residual and saturated water contents). For most dynamic soil water flow studies (infiltration especially), WCS is not porosity but some "saturated water content" (after Hillel) that is 10% to 30% less than porosity because of entrapped and dissolved air. Only for longterm saturated conditions (e.g., groundwater) will the effective WCS value approach porosity. All this is reason why I prefer to denote alpha, n and l (and WCR and WCS to the extrem) as empirical parameters taht define the shape of the hydraulic functions (empirical shape factors). Some purists among us may not like this.
Rien van G.
Yes, we tend to think that Rosetta is the best, but it sometimes enlightening to compare things with the Carsel and Parrish data to see the level of (dis)agreement.
The equations Brooks and Corey proposed, and our equations, are purely empirical functions in attempts to describe the water retention properties of soils (the same for most or all other equations; one may argue even for Kosugi's lognormal model). They are empirical, even though people have gone out of their way to assign physical significance to the parameters (n, alpha, hb, l, whatever). Defining alpha as 1/hb is such an attempt, which is formally incorrect: (1) many or most soils do not have a welldefined air entry value (that's why we went to that smooth function), and (2) alpha=1/hb is only an approximation (and a very poor for low n values). The same reasoning applied to l, which Mualem estimated in his equation to be 0.5 based on an analysis of some 40 soils (mostly coarse, repacked media). Marcel Schaap found that l on average is closer to 1.0, which does not make sense if l is interpreted as a poreconnectivity (or tortuosity) factor (physically, Se^l probably should decrease, rather than increase, when the water content decrease). Also, the B&C equations (for K) are based on a different poresize distribution theory (Burdine), and the l=2 value should have no relationship with Mualem's l=0.5. Actually, our equations can be combined also with Burdine's theory (giving the approach Haverkamp and a few others prefer), while the B&C retention function can be combined with Mualem's equation. All those things are worked out in detail in the RETC manual.
One can take this point of empiricism even further with how WCR and WCS are viewed (residual and saturated water contents). For most dynamic soil water flow studies (infiltration especially), WCS is not porosity but some "saturated water content" (after Hillel) that is 10% to 30% less than porosity because of entrapped and dissolved air. Only for longterm saturated conditions (e.g., groundwater) will the effective WCS value approach porosity. All this is reason why I prefer to denote alpha, n and l (and WCR and WCS to the extrem) as empirical parameters taht define the shape of the hydraulic functions (empirical shape factors). Some purists among us may not like this.
Rien van G.
Re: Alpha and n for specific soil textures
"WCS is not porosity but some "saturated water content" (after Hillel) that is 10% to 30% less than porosity because of entrapped and dissolved air."
does anyone know where Hillel said this?
tom
does anyone know where Hillel said this?
tom

 Posts: 4
 Joined: Thu Feb 04, 2021 6:46 am
 Location: Turkey
material dustribution
Hello I am using Hydrus 2D program. I want to make the soil profile 3 layers. I select Domain properties,after,material distribution,table of material and update materials 3 and ok and I select materials in domain properties with mouse after this, I select calculate current project but calculation did not occur, while calculating, it was write: 'Numerical solution has been stopped since it has not converged in 10 consecutiv a time steps'. calculating was stopped.I read all hydrus books but couldn't find the reason for this. After this, I tried another way. I selected ınsert,Domain geometry, surfaces,planar,graphically, I tried to apply the materials here but boundary curves part is empty and I didnt select, When I press the apply button it says error in input field emty entry.Why can't I create a 3layer (eg 1st layer of clay, 2nd sand, 3rd layer) soil profile. can you help me? very important to me ... thank you in advance
Re: Alpha and n for specific soil textures
We do not have any tutorials with multiple material layers since we thought that assigning materials to different part of the transport domain is rather trivial. Only one of our examples, Test2, involves a twolayered system. You can perhaps look at that. You can assign various variables, such as materials either on geometrical objects or directly on FE Mesh. When you want to do it on Geometrical Objects, you need to already define different soil layers as part of the Geometry, and thus it is a little bit more complicated. Assigning materials on the FE Mesh is likely easier, since you can do it anytime once you have the FE Mesh. You can, for example, as a practice, open the Furrow project, increase the number of materials, and assign them to different parts of the domain (e.g., around the drain, different depths, etc). J.
Re: Alpha and n for specific soil textures
I have realized that we have a series of video tutorials (Tutorial 5.06) on the topic “Introduction to Defining Properties on Geometric Objects”:
https://www.pcprogress.com/en/Default. ... d2tut506
You can also look at tutorial 5.17 (2D Domain with multiple layers) at
https://www.pcprogress.com/en/Default. ... tutorials
https://www.pcprogress.com/en/Default. ... d2tut506
You can also look at tutorial 5.17 (2D Domain with multiple layers) at
https://www.pcprogress.com/en/Default. ... tutorials

 Posts: 4
 Joined: Thu Feb 04, 2021 6:46 am
 Location: Turkey
Re: Alpha and n for specific soil textures
Thank you very much for your answer. I wish you good work.