RFAC, SMDFR and the effect of soil moisture on N uptake

[chporter]

This change was made in NUPTAK.for in CROPGRO.
commit #2370bd2bcf9bcff7689eb9859da4f194fadfd7b3

    ! FO/KJB - Change for Cotton
          !RFAC = RLV(L) * SMDFR * SMDFR * DLAYR(L) * 100.0
          RFAC = RLV_2D(L,J) * SQRT(SMDFR) * DLAYR(L) * 100.0

Several other crop models still use the (SFDFR * SMDFR) instead of the SQRT(SFDFR) in the calculation of RFAC. (Taro, CASUPRO, rice, CERES-teff, CSCER models, forage, NWheat, and potato.) Should we make this change there, too?

Ken's explanation:
This was made to reduce the severity of the effect of soil moisture on N uptake. The SMDFR*SMDFR was a strange “peaked” function that made N uptake much too sensitive to soil water in each given layer. The SQRT(SMDFR) makes it softer and more like code for CERES models for maize, wheat, and sorghum (as programmed by Joe Ritchie).

Answer: yes, the other models (at least forage) should be changed like this as well. But not just now. Do it only when we have interested scientists ready to slightly re-calibrate.

By the way, this has a relatively small effect, important only if N is limiting and under water-limited conditions as well. Fabio and I evaluated it for cotton when I was working to improve the cotton model response to N. It turns out that N mineralization (SOC method) is much more important. When you are out of N, you are out of N. No amount of minor coding helps in that situation.