Cloud Forcing Studies using CERES and GERBlike data

Cloud Forcing Studies using CERES and GERB(-like) data Joanna Futyan, Jacqui Russell and John Harries GIST 22 RMIB, Brussels, 09/11/2004 Page 1 © Imperial College London

Contents • Background • Results - ERBE-like data from CERES TERRA – Comparison with western tropical Pacific – Limitations of monthly mean gridded data • Results – GERB(like) data – Comparison with CERES – Cloud forcing by cloud type • Summary 2 © Imperial College London

Background • CRF - difference in flux between clear and cloudy sky conditions • Balance between LW and SW depends on macro- and micro-physical properties • Quantify via: • ERBE - R~1 in tropical convective regions – i. e. monthly mean LW and SW CRF ~ equal and opposite – Studies tend to focus on the western tropical Pacific – How does behaviour in African/ Atlantic regions compare? 3 © Imperial College London

Results – ERBE-like data from CERES on TERRA Distribution of 2. 5 o grid-box monthly mean R values for the Pacific, African and Atlantic convective regions for 2001 & 2002 • Pacific & African regions show high degree of cancellation • Lower degree of cancellation over Atlantic Seasonal variations in R for each region • Spread in part due to seasonal variations – least cancellation for Atlantic in summer – SWCRF >20 Wm-2 larger than LWCRF – Similar to eastern Pacific 4 © Imperial College London

Results 2. • Variability over Africa can be explained by combination of cloud and surface albedo effects • SWCRF low over bright surfaces – +ve net. CRF (low R) where high cloud occurs over the desert A • Some regions have a mixture of convective and low non convective clouds during the month – Low cloud increases SWCRF but not LWCRF -ve net. CRF (high R) – Need higher time resolution data to separate convective cloud – GERB & SEVIRI 5 © Imperial College London Futyan, J. M, Russell, J. E. and Harries, J. E. , J. Climate, 17 (16), 2004

Monthly mean CRF for June 2004 • GERB-like SEVIRI (GERB footprint resolution ~0. 5 o) LWCRF SWCRF net CRF • CERES TERRA, ERBE-like (Ed 1, FM 1 only), 2. 5 o resolution LWCRF 6 SWCRF © Imperial College London net CRF

Using SEVIRI to select the convective region SW flux LW flux Cloud type • EUMETSAT MPEF ‘CLA’ cloud classification – 3 x 3 SEVIRI pixel scale, 3 hourly – meteorological cloud types – simplified here to high-, mid-, low- • Gridded to GERB scale and used to select data on an instantaneous basis 7 © Imperial College London

Cloud forcing by cloud type • Calculate instantaneous CRF (clear sky flux – obs flux) – Use time-step mean estimate for clear • High cloud CRF = instantaneous CRF if flag as high cloud = zero otherwise SWCRF LWCRF 8 Standard definition © Imperial College London High clouds Mid/ high clouds Convective? Low clouds

CRF in the ‘convective region’ • Select footprints with LWCRF>30 Wm-2 and find average CRF land CRF 0 -20 ocean 100 CRF 60 net. CRF LWCRF - SWCRF -60 • GERB-like agrees reasonably well with CERES • Esp over ocean have significant contribution to net. CRF from low and mixed level clouds • For ocean net. CRF is always –ve • Over land - high clouds have a +ve net forcing 9 © Imperial College London

40 N Cloud forcing ratio land R 1 – Reduces R esp over ocean 0. 5 R 2 1 0 10 0 40 W 0 40 E 40 S • If low clouds are excluded explicitly R falls further 0 3 • 30 Wm-2 limit removes some low clouds ocean – No longer need LWCRF limit to select region • Over land R~1 in spatial average – Not at higher spatial scales • Over ocean R>1 even when low clouds are excluded © Imperial College London

Summary & future work • Monthly mean CRF averages the effects of all cloud regimes present during a month – Hard to attribute differences between regions • Synergy between GERB and SEVIRI allows calculation of cloud forcing by cloud type – Separate effects of low stratocumulus cloud from active convection • Cloud forcing ratio for ‘convective’ clouds ~1 over African land (on average) • Significant departures over Atlantic (-ve net CRF) even when low clouds are excluded • Future work: – investigate behaviour at daily and higher timescales 11 © Imperial College London