comparing ISCCP and GEWEX products Stefan Kinne Max

comparing ISCCP and GEWEX products Stefan Kinne Max Planck Institute for Meteorology Hamburg, Germany Ehrhard Raschke University of Hamburg, Germany MPI Metorolo gy Madison, July 2006

overview MPI-M ¡ available long-term global data-sets for radiative fluxes l l at the Top of Atmosphere (To. A) at the surface (sur) ¡ concept on investigating consistency ¡ assessments of solar flux comparisons ¡ assessments of infrared flux comparisons ¡ recommendations

MPI-M l ¡ Earth’s radiation budget how accurate defined is the radiation budget of our climate system? know your clouds … ¡ size-distribution ¡ cover ¡ (z) know ancillary data … ¡ surface + s-processes ¡ anthop. influences ¡ …on regional and seasonal scales

2 long-term data-sets MPI-M describe radiation budgets at To. A and surface ¡ ISCCP l l l ¡ GOAL: extract data on cloud field characteristics from operational meteorological satellite sensors years: 1983 -2004, res: 250 km (spatial) , 3 hr (temp) processed. C at NASA-GISS (Rossow, Zhang) GEWEX-SRB l GOAL: determine radiation budgets at the surface years: 1983 -2004, res: 100 km (spatial) , daily (temp) processed at NASA-Langley (Stackhouse) l clouds properties are ‘based‘ on the ISCCP climatology ! l l

task at hand MPI-M ¡ two bb-flux data sets l l for same time-period based on the same cloud data ¡ we should expect similar (if not the same) data l ¡ let’s test that l l l stratify data into zonal bands of monthly means display differences (always ISCCP minus GEWEX) interpret differences and highlight issues

MPI-M 75 -90 N ¡ 60 -75 N ¡ 30 -60 N ¡ 0 -30 S ¡ 30 -60 N ¡ 60 -75 N ¡ 75 -90 N ¡ regional temporal choices (1. 7%) (5. 0%) (18. 3%) (25. 0%) (18. 3%) (5. 0%) (1. 7%) use monthly averages

solar fluxes MPI-M solar i To. A ¡ solar i surface ¡ solar h/i To. A ¡ h the ‘solar’ driver solar atm. transmittance surface albedo planetary albedo typical plot: timeseries of monthly averages diff. colors for diff. latitude zones ISCCP GEWEX deviation Time (starting in 1983) g

ISCCP – GEWEX MPI-M WHY DEVIATIONS ? simplified treatment of GEWEX solar insolation at low sun-elevations toa D sol i DECEMBER 2005 for the record: larger deviations are gone In new GEWEX data

conclusion # 1 MPI-M ¡ un-necessary deviation for ‘solar driver’ ¡ low Dt) sun, avg (Dlat, ¡ also an issue in global modeling IPCC-4 AR a ¡ use consistent routines for To. A insolation ! ¡ agree ¡ on orbit and So implement properly! solar insolation of IPCC models

ISCCP – GEWEX MPI-M ¡ text sol isur

ISCCP – GEWEX MPI-M sol isur TOA ¡ text at surface: differences among data-sets are larger ! high lat. peaks are out phase to To. A peaks a cloud issue !

conclusion # 2 MPI-M ¡ ‘sol i To. A’ differences are lost at ‘sol i surface’ and ‘sol i surface’ differences are larger (!) ¡ differences in atmospheric properties dominate ¡ larger differences (season dep. ) at higher latitudes l ¡ most probable explanation ¡ diff. ¡ in cloud-cover / cloud opt. depth (for data-sets) assessment: cloud cover / optical depth differ ! ¡ ‘cloud’ differences have a seasonal dependence ¡ GEWEX cloud (opt. depth/cover) impact is stronger especially during polar summers (particularly in SH) (… yet weaker during mid-latitude summer in SH)

D sol h/i sur ISCCP – GEWEX MPI-M largest differences during NH mid-lat winters - at high latitudes (not shown) even worse ! a snow issue !

conclusion # 3 MPI-M ¡ solar surface albedo in models differs l l l differences have a seasonal dependence sign of diff. varies between high and low latitudes largest differences are linked to snow (alb. / cover) ¡ GEWEX has smaller solar surface albedos at higher latitudes especially in seasons, when snow can be expected … yet larger solar surface albedos in the tropics ¡ assessment on solar surface albedo: accuracy and consistency of ancillary (non-cloud data) data matters !

sol h/i ISCCP - GEWEX MPI-M ¡ toa text a combination of all previous biases

conclusion # 4 MPI-M ¡ diff. in plantetary albedo display combined effect l l l solar insolation biases solar surface albedo atmospheric properties (especially those of clouds) ¡ potential ¡ planetary albedo at To. A differences l ¡ for offsetting errors surface albedo diff. at mid/ high lat. are modulated as expected by cloud impact based on solar transm. - except for tropics: GEWEX clouds less reflective! assessment: cloud microphysics differ

infrared MPI-M IR h surface [emission] ¡ IR i surface ¡ IR h at To. A [OLR] ¡ h surf. temp effect (low) cloud effect (high) cloud effect typical plot: timeseries of monthly averages diff. colors for diff. latitude zones ISCCP GEWEX deviation Time (starting in 1983) g

ISCCP – GEWEX MPI-M ¡ text sur D ir h

ISCCP – GEWEX MPI-M ¡ D ir h sur can this trend text be detected at - ir i sur ? - ir h toa ? ‘false’ trend due to the use of incorrect surface temperature data for ISCCP in the tropics

ISCCP – GEWEX MPI-M ¡ there D ir i sur NO: atm. effects (clouds) dominate

ISCCP-GEWEX MPI-M toa NO: atm. effects (clouds) dominate ¡ text D ir h

ISCCP-GEWEX MPI-M toa/sur toa ¡ text lower GEWEX opt. depth/cover higher GEWEX opt. depth/cover sur D ir h

MPI-M conclusion # 5 ¡ atmospheric properties are main IR modulators ¡ surface emission differences vs OLR differences l l ¡ usually consistent with cloud (opt. depth/cover) bias … though not always ! cloud boundary temperatures matter l atm. temp. profile or altitude placement of cloud? ¡ assessment: cloud altitude placement differs ¡ other important ancillary data: l surface temperature / atm. temperature profile

conclusions MPI-M ¡ ISCCP and GEWEX radiations products often disagree on cloud ancillary data ¡ significant difference for cloud properties surprise, given the same cloud data-source ¡ larger disagreements at high-latitudes ¡ potential offsets can dilute severity of problem ¡ ¡ careful validation to quality data are needed ¡ ground-based network (BSRN) ? ¡ use synergy of advanced space sensors (A-train) ¡ collaboration of data/analyzing groups needed

recommendations MPI-M ¡ develop a reference algorithm for To. A solar insolation ¡ ¡ re-evaluate cloud properties and ancillary data (T, snow) ¡ ¡ ¡ compare to in-situ and ground-based quality data identify systematic diff. on regional / seasonal scales treat cloud ancillary data in a consistent manner ¡ ¡ Earth’s orbital data, solar constant, low sun elevation issue implementation ( … to suit model / data-set resolution) document your steps ! ¡ supply complete and detailed explanations on assumptions and methods – including a brief summary to allow a hasty user to understand major characteristics and error sources.

MPI-M

MPI-M extras solar downward surface flux ‘trend’ ¡ solar transmission ratio and ‘trend’ ¡ solar planetary ‘trend’ / ‘trend’ differences ¡ infrared surf emission ‘trend’ / ‘trend’ differences ¡ infrared outgoing ir flux ‘trend’ differences ¡ all-sky vs. clear-sky: the cloud effect ¡

MPI-M

ISCCP – GEWEX MPI-M ¡ sur D sol i MAY 2006 text high latitudes only

sol i ISCCP MPI-M MAY 2006 ¡ text lower latitudes sur

GEWEX MPI-M MAY 2006 ¡ text lower latitudes sol i sur

MPI-M

sol (isur/itoa) ISCCP/GEWEX MPI-M ¡ text

sol (isur/itoa) ISCCP/GEWEX MPI-M ¡ text

GEWEX MPI-M ¡ text sol (isur /itoa)

MPI-M

D sol h/i toa ISCCP – GEWEX MPI-M ¡ text lower latitudes

D sol h/i toa ISCCP – GEWEX MPI-M ¡ text high latitudes

sol h/i ISCCP MPI-M ¡ text toa

sol h/i GEWEX MPI-M ¡ text toa

MPI-M

ISCCP – GEWEX MPI-M ¡ text sur D ir h

ir h ISCCP MPI-M ¡ text sur

MPI-M

ISCCP-GEWEX MPI-M ¡ toa text lower latitudes D ir h

ISCCP – GEWEX MPI-M ¡ D ir h toa text high latitudes

MPI-M

ISCCP-GEWEX cld effect MPI-M ¡ text solisur

ISCCP cloud effect MPI-M ¡ text toa sol h/i

ISCCP cloud effect MPI-M ¡ text toa ir h

ISCCP 91 -95 MPI-M ¡ text Raschke et al. , Int. J. Clim. 2005 sol+ir h toa

ISCCP 91 -95 MPI-M ¡ text Raschke et al. , Int. J. Clim. 2005 sol+ir atm