See here for a discussion about soil pedotransfer functions.
Anne Verhoef at Univ. Reading is currently heading a review of the characterisation of water and heat transfer plus pedotransfer models (PTFs) in land surface models as part of CMIP6 and the GEWEX-SoilWat activity "Evaluation of pedotransfer functions in climate and hydrological models" launched Oct 2016 (see also here).
Some related information about my soil hydraulics paper:
Marthews TR, Quesada CA, Galbraith DR, Malhi Y, Mullins CE, Hodnett MG & Dharssi I (2014). High-resolution hydraulic parameter maps for surface soils in tropical South America. Geoscientific Model Development 7:711-723.
Link to paper
Link to data (EIDC)
A question I've been asked quite a lot about the pedotransfer functions in this paper is how do the quantities I talk about in this paper relate to the soil variables required by the model JULES, i.e. the parameters listed here? Well, here's how:
- What JULES calls b or soil_b is related to the Brooks & Corey parameter lambda by b=1/lambda and to the van Genuchten-Mualem parameter n by b=1/(n-1)
- What JULES calls SATHH (in m) is related to the Brooks & Corey parameter psi_e (in Pa) by SATHH=-psi_e/(rhow*9.81) (where rhow=1000 kg/m3 is the density of water) and to the van Genuchten-Mualem parameter alpha (in m-1) by SATHH=1/alpha.
- What JULES calls SM_SAT is exactly theta_sat in cm3/cm3
- What JULES calls SM_WILT is exactly theta_PWP in cm3/cm3: I put an example calculation in my Table 2 for how to calculate this.
*** SM_WILT is not theta_res, by the way: currently JULES assumes theta_res is zero in all simulations (see footnote below) ***
- What JULES calls SAT_CON is exactly k_sat in my paper: the only one of these quantities Hodnett & Tomasella (2002) didn't provide a pedotransfer function for (my Table 2), so by default most people use the Cosby et al. (1984) pedotransfer function despite the very high uncertainty in that PTF.
- What JULES calls SM_CRIT causes a lot of confusion. It's actually defined as the value of theta below which plants begin to feel water stress (as indicated by photosynthetic rate falling below 100% of optimal). This is not the same as SM_FIELD in general, which is the theta equivalent of field capacity. The assumption in the JULES model is that photosynthesis begins to drop off at suctions below 33 kPa, so in JULES this is effectively equal to theta_33kPa (much confusion comes from the fact that older textbooks used to define field capacity at a level of 33kPa instead of 10kPa).
- I didn't provide pedotransfer functions for HCAP or HCON: such functions do exist but they just weren't the topic of my paper.
This work followed on from my earlier work with the SWEAT model in Aberdeen and my paper Marthews et al. (2008).
Footnote about theta_res: If using the van Genuchten soil hydraulic option, the smc values used in the JULES code are supposed to represent not soil moisture theta, but actually (theta-theta_res) where theta_res is the residual soil moisture in van Genuchten's equations (and you are therefore supposed to 'add on' an estimate of theta_res to any soil moisture outputs at the end of your runs). Unfortunately, the documentation is not explicit about this (no mention here), but it is mentioned clearly enough in Best et al. (2011:eqn60) "In JULES, ... the soil moisture variable is implicitly defined as (theta-theta_res), leaving three parameters" (and in any case it's clear from the absence of theta_res in the van Genuchten equations used in the code).
For me, it would have avoided confusion to put theta_res into the JULES code (and perhaps require it to be set =0 by default if it's a problem at many locations), but this is not the way van G was implemented in JULES and we have to live with it / work around it. This issue is partly why it is frequently stated that people should not use the same ancillaries when using the Brooks & Corey vs. van G soil hydraulic options.