Updates to icloud3

[gthompsnWRF]

Updates to improve icloud3 (cloud fraction scheme) based on 72 all year simulations validated against solar radiation observations.

TYPE: enhancement
KEYWORDS: cloud fraction, radiation

SOURCE: gthompsn

DESCRIPTION OF CHANGES:
Problem:
From numerous real-time and retrospective WRF simulations, the verification of shortwave radiation has consistently shown too much solar radiation reaching the ground due to a lack of cloudiness. The existing icloud=1 option (attributed to Xu and Randall) is ineffective at creating a realistic histogram of values of cloud fraction between zero and one.

Solution:
Compared to the previously-released (initial) version of this cloud fraction scheme, these updates have now been verified (by Mei Xu) by 72 WRF simulations spread every 5th day through the year for 42-h forecasts using almost same grid as HRRR and verified with multiple sources of solar radiation reaching the ground.

ISSUE: None previous

LIST OF MODIFIED FILES: phys/module_radiation_driver.F

TESTS CONDUCTED:

1. Conducted 72 simulations of 42h duration with 3km spacing (approx HRRR-CONUS grid) and validated against USCRN (~100 sites) and SURFRAD (7 sites) solar radiation at the ground.

RELEASE NOTE: Improvements to icloud=3 option based on extensive (year-long) experiments with 3km spacing. Journal manuscript is in preparation.

[dudhia]

Largely parameter changes. Added capability for vapor change when initializing clouds.

[weiwangncar]

@gthompsnWRF The latest change is verified with 3 km simulations, do you know how it would work for ~ 10 km grid sizes? Was the 3 km simulation run using albedo from geogrid?

[gthompsnWRF]

@gthompsnWRF The latest change is verified with 3 km simulations, do you know how it would work for ~ 10 km grid sizes? Was the 3 km simulation run using albedo from geogrid?

Wei: yes, I am pretty certain the 3-km results had the geogrid-albedo as constant (my personal preference). Meanwhile, I did not test at 10km spacings but Pedro Jimenez has been doing his tests at either 9 or 10km in MADWRF project. Of course there could be further tunings in the future for the DX grid spacing dependence on the RH-critical value.

[weiwangncar]

@gthompsnWRF Thanks, Greg. Do you have a figure to show some results? Any comparison from previous code? Is T2 also verifying better with the new code?