-
Notifications
You must be signed in to change notification settings - Fork 5
Param_Comments
This file contains notes on what adjusting certain parameters has done to the HOC code. Brian Griffin; December 17, 2005.
C1 (low skewness): lowering decreases rcm for RF02; raising increases rcm for RF02. As long as I could change the other parameters, the simulation turned out better the more I could raise C1. It provided for a better slope between the peak of rcm and the cloud base, which also seemed to cut the cumulus below the stratocumulus a little bit. If C1 is too low, wp2 at low skewness will be unrealistically high. C1 is also important in determining wp2 below the cloud base in the cumulus cases.
C1b (high skewness): Not much of any effect on RF02. Having it too high produced instability in the Wangara case. As a result having to lower C1b, the stratocumulus clouds in the ATEX case dropped from about 60% cloud fraction (matching COAMPS LES) to about 40% cloud fraction. However, as it turned out, this was actually closer to the LES average for the case (COAMPS was at the high extreme).
C1c (slope): It made the C1 function work out better for all the cases to raise C1c (thereby decreasing the slope of the function) from 0.500 to 1.000.
C2thl: Golaz provided a good value and I stuck with it.
C2rt: I increased it from 1.000 to 1.500 in order to dampen the rtp2 in the RF02 case, which had been helping to produce too much cumulus underneath the stratocumulus.
C2rtthl: Golaz provided a good value and I stuck with it.
C4: Not used in the model anymore.
C5: Golaz provided a good value and I stuck with it.
Note: The constant C6 had been broken into C6rt and C6thl. I found out that having one C6 Skewness function for both cases was better than having two individual constants.
C6 (low skewness): lowering decreases rcm for RF02; raising increases rcm for RF02. We originally made C6 low for the DYCOMSII RF02 case to make the layer more well-mixed, thereby lessening the amount of cumulus below the stratocumulus.
C6b (high skewness): has hardly any effect on RF02. lowering the value from 6.000 to 4.000 did help in terms of decreasing the ARM cloud fraction at cloud base when it was too high while increasing the ATEX cloud fraction in the stratocumulus section when it was too low.
C6c (slope): Setting the slope to 1.000 seemed to work well from the start.
C7 (low skewness): lowering increases rcm for RF02; raising decreases rcm for RF02.
C7b (high skewness): The higher the value of this constant, the better it made BOMEX by increasing the cloud fraction at cloud base. Increasing this value also seemed to decrease the ATEX cloud fraction in the stratocumulus section.
C7c (slope): Changing this parameter from 0.500 to 1.000, thereby lessening the slope of the function, helped a lot in making things work out for all cases.
Note: This function had been originally placed in the code by Chris Golaz and behaves differently than the other skewness functions.
C8: Needed to be lowered significantly in order to prevent ARM from going unstable. Unfortunately, lowering the value of it also significantly decreased the ATEX cloud fraction in the stratocumulus section. This effect on the ATEX case was counteracted by increasing the value of C1b. Read the notes about C1b to read about the conflict between the ATEX and Wangara cases. Raising the value of this parameter also made a very small increase in rcm in RF02, while lowering the value made a very small decrease in rcm in RF02.
C8b: Golaz provided a good value and I stuck with it.
C10: Not used in the model anymore.
C11 (low skewness): raising made a very small increase in rcm in RF02; lowering made a very small decrease in rcm in RF02. A higher value seems to make the stratocumulus cases (FIRE, RF01, RF02) run better.
C11b (high skewness): A key parameter for the BOMEX, ARM, and ATEX cases. BOMEX and ARM seemed to have too sharp of a slope for cloud fraction from the top of the cumulus to near the cloud base, and then a sudden change to a sharp increase in cloud fraction near the cloud base. Increasing the value of this parameter smoothed out the slope somewhat. However, too much increase meant that the ARM case became unstable. It also helped the ATEX case when the value of this parameter was lower, for it increased the cloud fraction in the stratocumulus section.
C11c (slope): I tried to change the value of this parameter from 0.500 to 1.000, as I had with the slopes of all the other skewness functions. However, when I did that, it had the same effect on the ARM case as increasing C11b too much did. That is, it made the case unstable. The value of 0.500 that Chris Golaz provided is the right value for this parameter.
C12: C12 is still in the code, but it is not being changed from it's original value of 1.000.
nu_r: The coefficient for rain water and rain drop diffusion was set by making the value for RF02 DS HOC approximately the same as RF02 DS COAMPS at the ground.
All other constants were set by Chris Golaz and they are good matches. They will remain the way they were set.