Breakout Summary Report

ARM/ASR User and PI Meeting

19 - 23 March 2018

Session Title: CAUSES: Clouds Above the United States and Errors at the Surface
Session Date: 21 March 2018
Session Time: 10:30 AM - 12:30 PM
Number of Attendees: 28
Summary Authors: Cyril Morcrette

Breakout Description

This session focussed on the CAUSES project, where 11 models from 9 institutes participated in a model inter-comparison project to better understand the reasons for why so many climate and weather forecasting models have a warm screen temperature bias over the American Midwest in the spring and summer. There were presentations covering material from four CAUSES papers recently accepted in JGR-A and about work on the same subject carried out by Lin et al.

Main Discussion

From the presentations given, it was shown that errors in the evaporative fraction explain 0-5 Kelvin of a model's warm bias, while radiation errors explain 0-2 Kelvin of the bias. The evaporative fraction bias is correlated with soil moisture bias. And radiation biases are often related to the representation of convective clouds.

Key Findings

These are summarised in the attached "one-slide summary".

Note that the four CAUSES papers have been highlighted in an article in EOS news: https://eos.org/editors-vox/diagnosing-the-warm-bias-in-the-central-united-states

Issues

Major (M)

M1. (UKMO) Can one improve the radiative properties of deep convective clouds?
M2. (LLNL ongoing) How sensitive is the warm bias to properties of the land surface including stomatal/bare ground resistances and surface albedo?
M3. (LLNL to organize, but would require a collaborative effort particularly from those that have CPMs) Since about ½ of the summer rain comes from mesoscale convective systems, would convective-permitting models (CPMs) have smaller warm biases?

Minor (m)
m1. (LLNL?) What can be learned about the warm bias from analysis of seasonal hindcasts (Experiment 2 in CAUSES)?
m2. (UKMO?) What can be learned from an analysis of simulations stratified by days since last precipitation event?
m3. Why do some models have their largest warm bias during the day while other models have their largest warm bias at night?
m4. What causes the striking East-West gradient in the warm bias in some models?
m5. (LLNL) How sensitive is the theoretical analysis to the assumed value of d(SH)/dT2m?

Decisions

The main part of the CAUSES project is complete. Detailed analyses of many models have been carried out, indicating the areas of model physics that should be focused on to help tackle the warm bias.

Future Plans

UKMO are planning to look at Major1
LLNL are planning to look at Major2 and are considering the feasibility of collaboration using convection-permitting models (Major3)

For Minor 2. (stratifying by days since last precip), UKMO have plans to look at this, staff resources permitting.

It was agreed that LLNL/UKMO will try to analyse any model data provided in a similar format to what was provided before in order to repeat analysis on new models versions, staff resources permitting.