Understanding monsoon precipitation representation in climate models: process-based diagnostics H. Annamalai IPRC, University of Hawaii hanna@hawaii.edu
CMIP3 MMM – GPCP JJAS - Precipitation (mm/day) CMIP5 MMM – GPCP (Sperber, Annamalai et al. 2012) 1. +ve rainfall errors along the path of cross-equatorial flow 2. Errors persist thru’ AC “multiple sources”
Δτ CMIP5 MMM minus ERA_INT (3 o S-3 o N; 40 o -100 o E) Jan Mar May Jul Sep Nov Compared to climatology: In May 40-45% weaker and in November 70% weaker Δτ a measure of Bjerkens’ feedback in the Equatorial Indian Ocean
3 o S-3 o N – surface currents Observations (cm/s) 3 o S-3 o N – surface currents CMIP5 MMM 1. Weak eastward WJs 2. Unrealistic westward currents 3. Pile-up of warm waters WEIO
EXP2 minus CTL Precip / wind 850hPa CMIP5 bias – Precip / wind stress mm/day mm/day Misrepresentation of EIO coupled processes lead to systematic errors in monsoon precipitation Annamalai et al. (2017, J. Climate – revised)
Processes involved in organizing convection over the Bay of Bengal? During the annual cycle: When does spatially organized convection develop and how? and/or What are the processes that anchor onset of self-aggregation of convection
Precipitation averaged over Bay of Bengal (80 o -100 o E; 10 o -20 o N) May - range April “Onset” of convective organization occurs during the month of May over Bay of Bengal Quasi-steady state during June - August
Overview of key physics from cloud-resolving models 1. Held et al. (1993); Tompkins and Craig (1998) 2. Bretherton et al. (2005) J. Atmos. Sci., * 3. Muller and Held (2012) J. Atmos. Sci., * 4. Emanuel and Khairoutdinov (2010). AMS Conf.,* *System for atmospheric modeling developed by Khairoutdinov and Randall (2003)*
CRM simulations for a homogenized SST forcing – Day 20 initiate convection – perturbations in the low-level MSE Bretherton et al (2005) Instability of convection-water vapor – radiation feedbacks: systematically dry and moist
CRM simulations for a homogenized SST forcing – Day 50 “decrease in radiative cooling due to anvil cirrus clouds” Bretherton et al (2005) “increase in THF due gustiness” “mesoscale circulations flux MSE out of dry column”
(near-surface Temperature) Convective self-aggregation : sensitivity to domain size Muller and Held (2012) Low-level clouds –LW radiative cooling
Raymond (2000) and Stephens et al. (2008)…. Emphasized the role of convection – radiation feedbacks in self-aggregation i.e., differential radiative heating and cooling drives mesoscale circulation that fluxes MSE out of dry column “ This feedback is a consequence of convection itself “
Processes that could anchor onset of self-aggregation…. 1. Moisture-convection feedbacks (dry regions become more drier and moist regions become more moist) Meso-scale circulations flux MSE out of the dry column and feed the moist column 2. Radiation-convection feedbacks (differential radiative heating and cooling drives the mesoscale circualtions) This feedback is a consequence of convection itself 3. Surface fluxes (Emanuel 1986; Raymond et al. 2009) 4. Sea surface temperature (Emanuel and Khairoutdinov) 5. Stratiform convective instability (Mapes 2000)
Representation of interaction between cumulus convection and large-scale circulation [Quasi-equilibrium concept of Arakawa and Shubert (1979) ] requires consideration of moisture and temperature, represented by MSE ( m ) m = C p T + gz + Lq Vertically integrated MSE tendency – conserved in convective processes “storage” “adiabatic terms” “diabatic terms” 1. Deep tropics – above PBL – no horizontal T variations Neelin and Held 1987 2. Entropy forcing: LH, SH, LW, SW, moisture variations Raymond et al. 2009 Bretherton et a. 2005 Su and Neelin 2005
Precipitation averaged over Bay of Bengal (80 o -100 o E; 10 o -20 o N) May - range April “Onset” of convective organization occurs during the month of May over Bay of Bengal Quasi-steady state during June - August
1. April end – mid/high 1. April – descent – entire troposphere level ascent 2. May early – ascent – entire troposphere 2. May early – Upwards of 3. Low-level clouds 650 hPa – ascent 4. Mid-higher level clouds “stratiform induced top-heavy profile – exports more MSE per unit mass flux
Vertical profile of vertical velocity…………….. • “vertical velocity is not observed but is assimilated : depends on model physics” • In data void regions (such as the monsoons), model prejudices inject errors • In moist convective regions, vertical profile of vertical velocity is determined by latent heating and radiative cooling • In dry regions, vertical profile of vertical velocity is determined by radiative cooling • Vertical velocity represents large-scale circulation – mass convergence • Top versus bottom heavy – exports or imports moist static energy
A Conceptual model based on vertically integrated MSE budget dry southern Indian Ocean and moist Bay of Bengal moisture-convection and cloud-radiative feedbacks Charging of MSE during late spring and early summer - Surface fluxes and radiative warming Diagnostics performed on……….. Four reanalysis products (MERRA, JRA-25, ERA-I and NCEP) 33+ CMIP5 models
Storage term (dh/dt) over the Bay of Bengal (80 o -100 o E; 10 o -20 o N) Jan Mar May Jul Sep Nov Charging of MSE during March-April and discharging starts during May – indicate convective onset
ERA-Interim MSE budget – Southern Indian Ocean (50-95 o E; 30 o -20 o S) Horizontal and vertical advection of MSE - sharp increase in May
LHF – Bay of Bengal 1. Increase begins in April (SST) 2. Peaks in June 3. Semi-annual cycle 4. MERRA LHF is weak 5. JRA-25 LHF is high “despite similar SST in the reanalyses” Bay of Bengal Jan Mar May Jul Sep Nov LHF – Southern IO 1. Increase begins in April 2. Maximum June-August 3. MERRA LHF is weak Southern IO 4. JRA-25 LHF is high Jan Mar May Jul Sep Nov
F rad - - Bay of Bengal 1. Sharp rise in May 2. Quasi-steady state in JJA 3. MERRA – Increase in April itself 4. JRA-25 weak (despite higher rainfall) Bay of Bengal Jan Mar May Jul Sep Nov Southern IO F rad --- Southern IO 1. Increased radiative cooling – March onwards 2. Quasi-steady state JJA 3. JRA-25 weakest (despite higher rainfall over Bay of Bengal) Jan Mar May Jul Sep Nov
LW component (cloud-radiative feedbacks) – Bay of Bengal JRA – weakest OLR ~ 240 W/m 2 (JJAS)
(GISS) “all-models have cold SST bias in the northern Indian Ocean” Surface fluxes instability for convective aggregation in JRA-25
“cloud-radiative feedback instability in MERRA”
CESM1 CCSM4 GFDL_CM3
Multi – Reanalysis – Mean (precipitation) minus TRMM “no errors in SST”
Concluding remarks Monsoon convective organization over the Bay of Bengal begins in May – (a) Surface fluxes and cloud-radiative feedbacks (diabatic MSE sources) Consistent with CRM results – onset of convective aggregation (b) JRA-25 – surface fluxes instability dominate (c) MERRA – cloud-radiative instability dominates (d) CMIP5 models – cold SST bias over the northern Indian Ocean under estimate surface fluxes Monsoon convection during JJA (quasi-steady state): dibatic terms balance adiabatic terms Need in-situ observations to constrain reanalysis and climate models
Extra slides
MERRA - SIO MSE terms for BoB / SIO (one reanalysis)
CRH versus precipitation Bay of Bengal ERA-Interim JRA-25 MERRA
Precipitation and SST 1. Regional rainfall zones I 2. High-SST/Orography II 3. Different SST threshold (tropospheric T/CRH) III E-W asymmetry NIO QuickSCAT SST – upwelling RW dynamics Jet axis Shear vorticity A/C circulation (dynamic not thermodynamic Control) EEIO – Annual cycle SST-precipitation relationship?
TRMM-based precipitation climatology April May June (mm/day) Convective organization over Bay of Bengal Transition from April to May
(i) Slowly, due to surface heating and attendant sensible heat flux, shallow low-pressure forms over land (ii) This low, gradually generates low-level wind inflow that draws moisture from the ocean to the south (iii) The moisture is carried upward by the convection, condenses, and warms the upper troposphere, where an anticyclone forms
Near-surface wind climatology April May
Muller and Held (2012)
Stephens et al. (2008)
CRM simulations for a homogenized SST forcing – Day 10 Bretherton et al (2005)
TRMM-based precipitation climatology April May June (mm/day) Convective organization over the Bay of Bengal Transition from April to May
L = 198 km PW L = 510 km PW L = 198 km OLR L = 510 km OLR Muller and Held (2012)
Recommend
More recommend
