Simulating convection sensitivity to atmospheric state during AMIE-Gan

Description

The relationship between moist convection and tropospheric humidity is thought to be central to the existence of the Madden-Julian Oscillation. The difficulty that many GCMs have in simulating the MJO may therefore be diagnostic of insufficient coupling between convection and humidity in their cumulus parameterizations, associated with weak entrainment and/or rain evaporation. The AMIE-Gan deployment of the second ARM Mobile Facility (AMF2) in the Maldives over the past fall and winter offers an unprecedented opportunity to observe the onset of the MJO and constrain cumulus parameterizations. Nature has cooperated by producing at least two, and perhaps three, MJO events during the first half of the six-month AMIE-Gan deployment.

Eventually, a forcing data set will be created to drive SCMs and CRMs during the AMIE-Gan period. However, we need not wait for that product to test certain aspects of cumulus parameterizations. One of the highlights of AMIE-Gan is the presence of a long record of three-hourly radiosonde data. Over the first three months the soundings clearly show moistening of the lower troposphere by about 10% relative humidity in advance of MJO onset, followed by 20–30% humidity increases at upper levels after the disturbed phase begins. The temperature signal of the MJO is weaker, but for at least the second observed MJO in November 2011, there is a detectable 1 K warming of the lower troposphere before MJO onset and a 2 K upper troposphere warming thereafter.

Since cumulus parameterizations require only a temperature and humidity profile to diagnose moist convection triggering and depth, we can use the available soundings directly in semi-prognostic SCM simulations that test the response of the parameterizations to the observed thermodynamic structure. We will use this approach to diagnose convection depth using several different versions of the GISS Model E2 SCM: the operational CMIP5 version, experimental versions with stronger entrainment and/or rain evaporation, and a new version that reduces entrainment only after convective downdrafts inject cold air into the boundary layer. ARM radar observations of convective depth during AMIE-Gan will be used to document the actual relationship of convection to thermodynamic structure in advance of MJO onset. We will also attempt to understand whether the previously observed large variance in precipitation and convective depth at intermediate values of column water vapor is due primarily to lapse rate or humidity profile variations.