The composite Lagrangian cases: LES intercomparison Irina Sandu, Andy Ackerman, Peter Blossey, Chris Bretherton, Johan van der Dussen, Adrian Lock, Stephan de Roode, Bjorn Stevens

Lagrangian analysis of the air mass flow How? Trajectories + Re-analysis + Satellite data HYSPLIT (ERA-INTERIM) ERA-INTERIM MODIS (Terra, Aqua) AMSR-E 2002-2007 (May to October in NE, July to December SE) Starting time: 11 LT, Duration: 6 days, Height: 200m When? Where? Klein&Hartmann (1993) zones : NE/SE Atlantic, NE/SE Pacific NEP NEA SEP SEA Sandu, Stevens and Pincus, ACP, 2010

Composites NEP JJA 2006-2007 An ensemble of composite cases: slow, intermediate and fast transitions CF MODIS ref slow fast Composites NEP JJA 2006-2007 3 days SST LTS D Is the LES able to reproduce, not only the transition, but also differences in the transition?...we divided the transitions in 3 categories: slow, fast, intermediate…they are separated by LTS

Our questions Are the LES able to reproduce: the observed changes in cloudiness induced by changes in the SST/LTS? the transition’s pace and its dependence on the inversion strength? Do they agree in term of : The decrease in cloud albedo and cloud cover during the 3 days The time evolution of the cloud fraction The growth rate of the boundary layer

Outline The fast/slow cases Conclusions & Next steps Simulations : initial conditions, requirements, models First results for the reference case The fast/slow cases Conclusions & Next steps

Composite REF case : NEP - JJA 2006-2007 Initial profiles (10 LT) Forcing l (K) qt (g/kg) SST (K) Calipso Time (days) u (m/s) v (m/s) D (x106 s-1) Time (days)

Initial conditions ql l ref slow fast SST Cts divergence (the same) No advective tendency

Simulations initial time : 10 LT, duration: 72 hours initial date: 15 July (but 15 June for UCLA ) diurnal cycle of solar radiative forcing taken into account cloud droplet number concentration: 100 cm-3 resolution : x = 35m, z = 5m (at cloud top) domain size : 4.48 X 4.48 X 3.2 km (128 x 128 X 428 points)

            Models & participants UCLA-LES DALES UKMO SAM (Irina Sandu) DALES (Johan van der Dussen, Stephan de Roode) UKMO (Adrian Lock) SAM (Peter Blossey, Chris Bretherton) DHARMA (Andy Ackerman) REF FAST SLOW               

Outline The fast/slow cases Conclusions & Next steps Simulations : initial conditions, requirements, models First results for the reference case The fast/slow cases Conclusions & Next steps

Difficult to compare to the observed cloud cover UCLA ( ! Qualitative comparison only)

The simulated SCT (UCLA – big domain) Albedo decreases by 41 %

The simulations capture the major observed features of the SCT, and corroborate the conceptual model proposed by Bretherton (1992) to explain it w’v’ UCLA CF

Do the models agree? (I –time series)

Do the models agree? (I – time series)

Hopefully, it does not matter a lot…

Do the models agree ? (II – decoupling) w’v’ (10-4 m2/s3) UCLA SAM DALES DHARMA

Do the models agree ? (III – cloud fraction) UCLA SAM DALES DHARMA

Do the models agree ? (IV – entrainment rate)

Do the models agree ? (V – FT state) qt ql qr CF w’’v LW w’2 SW

Do the models agree ? (V – FT state) qt ql qr CF w’’v LW w’2 SW

Do the models agree ? (V – FT state) qt ql qr CF w’’v LW w’2 SW

Is there a drift in time ? (UCLA)

Is there a drift in time ? (SAM)

Is there a drift in time ? (DALES)

Is there a drift in time ? (DHARMA)

Outline The fast/slow cases Conclusions & Next steps Simulations : initial conditions, requirements, models First results for the reference case The fast/slow cases Conclusions & Next steps

Slow against fast SCT (UCLA – big domain)

Slow against fast SCT

Role of the inversion strength

Role of the inversion strength Boundary layer growth rate during the first 24 hours

Conclusions LES reproduce well not only the main features of the SCT, but also subtle details like differences between slow and fast transitions (UCLA) The SCT timescale is mostly related to the strength of the temperature inversion capping the Sc topped boundary layer (UCLA) striking resemblance of the 4 simulations of the reference case (differences well rather understood)

Next steps fix l,qt at 3km check why LWD is different in DHARMA (fix LWD) correct surface fluxes in UCLA-LES re-run the 3 cases (same domain) - perhaps just the reference case in the beginning