2nd IPWG

25-28 October 2004

Monterey, CA

  1. General Remarks
  • The WMO Space Programme – Don Hinsman
  • Review of WMO’s Global Observing System; about 10 partners
  • IPWG program co-chairs comments – Gruber and Levizzani
  • IPWG established June 2001
  • First meeting – Sept 2002 – Madrid; established 3 working groups (Operational Applications, Research, Validation); 21 WG recommendations
  • Accomplishments:
  • Web Site:
  • Requirements for MW frequency protection
  • Climate data sets inventory
  • Validation & links
  • Links to training
  • Co-sponsor with GPCP on IPCC Assessment
  • Algorithm inventory and downloadable code
  • IPWG/GPCP validation intercomparison project – CONUS, Australia, Europe
  1. International Projects and Satellite Programs
  • Report of the Rapporteur to the CGMS – Jim Purdom
  • Review of linkages between IPWG and bigger picture CGMS activities. Plans are to move forward with virtual laboratory and “electronic workbook”. IPWG members should revisit material on web site and make sure information is up to date.
  • CGMS high priority items to IPWG – IPCC Assessment, solid precip, precip over complex terrarin, ongoing validation activities
  • Research & Operational Application of TRMM-based, fine time scale precipitation analysis – George Huffman for Bob Adler
  • V6 converging with PR better than V5; now within 5% over ocean
  • TRMM is working well and used in operations; awaiting NAS recommendation on future of TRMM. TRMM has enough fuel for controlled reentry through 2005, without that, it could go into 2011 (and overlap with GPM). TRMM costs $4M/yr to operate (full cost, include FTE’s). NASA safety group does not require controlled reentry.
  • A variety of application examples of TRMM rainfall shown – tropical cyclone rainfall, ENSO onset Index, etc.
  • Microphysical Considerations in Remote Sensing – Vincenzo for Rosenfeld
  • Rosenfeld and Lensky (BAMS, 1998) – classification of convective clouds into microphysical zones according to the shape of the temperature profile (IR) and effective radius (near-IR) – applied to MSG data
  • Differences between maritime and continental cloud systems (and orographic); continental has larger drops, but smaller rain rates while maritime smaller drops, but higher rain rates. Translates into lower reflectivity (Z) in maritime rain, factor of 2-4. Can use this information to develop improved passive MW databases/libraries; something that will be addressed via RAINCLOUDS, if funded by E.C.
  • GPCC Assessment for IPCC – Arnie Gruber
  • Review of outline/proposal; draft to be ready by Jan 2005 with a follow-up workshop in March/April 2005, with a final version due by Sept. 2005.
  1. Operationally-Oriented Precipitation Datasets
  • Integration of the EUMETSAT MPE in an Operational realtime processing environment – T. Heinemann (EUMETSAT)
  • Flash floods are rare, but a real concern in Europe. There is a need for real time processing. Plus, for Africa and Asia, there is no other data but satellites.
  • The MPE – IR and MW combined algorithm (METEOSAT+SSMI); within the Nowcasting SAF; real-time. Same ideas and philosophy as we employ with MSPPS.
  • Will eventually bring in TRMM data
  • The SEVIRI Precipitating clouds product of the Nowcasting SAF – A. Thoss (SMHI) + Bennartz
  • Two products – Precipitating Clouds (PC) and Convective Rain Rate (CRR)
  • PC product – 3 classes – no rain, 1-5 mm/hr, >5 mm/hr; developed from AVHRR and BALTEX radar data + surface temperature data and adopted for MSG (vis, near-IR and IR)
  • Daily mapping of 24-hour rainfall at pixel scale over South Africa – G. Pegram (South Africa)
  • Merging data sets; S. Africa is between GOES and METEOSAT and has sparse gauges and radars as well
  • Multispectral rain technique to identify raining pixels (vis, IR, WV), then use IR for rain rate
  • They generate daily maps and it gets posted on the S.A. Met Service web site
  • EPPrePMEx: A real time rainfall estimation system based on GOES-IR Imagery – J. Sanchez (IMTA/Mexico)
  • Roots from CST originally adapted for hurricane rainfall, later, adapted for general use
  • Working collaboratively with NOAA/OH
  • Real time generation and web posted ( basin generated amounts as well
  • Comparisons with Autoestimator during 1998; appears to do better than AE. Also, comparisons with RFC gauges in Texas/Mexico region
  • Moving towards multispectral, multiplatform operational rainfall at NESDIS – Bob K.
  • Overview of GOES based rainfall work; SAB customer, automated, real-time use, etc.
  • Need to move towards blending measurements and platforms – SCaMPR; transitioning to run in real time
  • SCaMPR appears to outperform AE and GMSRA for rainfall >20 mm in six hours
  • Possible use of lightning data; GOES-R possibility?
  • Status of NOAA/NESDIS AMSU algorithm – Ralph F.
  • TRMM TMI Algorithm – V6 and Beyond – Chris K.
  • V5 to V6; new databases (ASCII, available on web)
  • Validation – AMSR-E and TMI similar, V6 land on Melbourne,FL -10%; regional biases
  • V7. Why? 10-15% difference between PR and TMI. Also, many new sensors like SSMIS and Windsat  parametric algorithms. Also, V7 should eliminate all screening routines and instead use probability of rain and confidence of using the correct database. So, like a 50% probability of rain or greater will be close to what V6 is producing now.
  • Future … more attention to land and extratropics. More coordination needed; how do we transition from empirical to physical methods? Version numbers, standards, QA, etc.
  • CMORPH – Bob J.
  • Uses 8 satellites – POES, DMSP, AMSR-E, TMI for MW; 5 for GOES
  • Cloud motion from IR to move MW rain
  • Validation looks great
  • Comparisons with CMORPH with GMORPH
  • The status of the MPA and insights gained from adding new data sets – George H.
  • TRMM 3B42RT and 3B42
  • Testing AMSU-B and AMSR-E
  • Discussion on validation and error estimates
  • Cloud motion winds diffusion scheme for quantitative rainfall estimation – F. Tapiador (Spain)
  • Motion vector approach – similar in concept to CMORPH, but uses IR derived winds and model winds (like GMORPH) to propagate MW precip.
  • A more physical method than an image processing technique – consider “diffusion” approach where cloud motion is treated as a fluid.
  • A combined TMI-SSMI rainfall estimation technique for W. Africa – J. Schulz (Germany)
  • PR, TMI, SSMI, MSG
  • Uses PC analysis and temporal averages to remove background MW signature (Bauer, Petty); 7 day average used
  • Rain Rate – PR calibrated rain index; Petty HSS method
  • Operational rainfall retrieval based on cloud microphysical properties – T. Nauss (Germany)
  • Active Convective Technique (ACT) – vis, near-IR, IR, WV used for conv./stratiform, advective scheme and cloud optical properties
  • Plans to implement for SEVERI
  • Precipitation observational capabilities at the Brazilian Inst. for Space Research – C. Angelis
  • They run the HE; want to develop new techniques to improve on the HE for different climate regimes
  • They maintain 16 weather radars in Brazil, Argentina and Paraguay
  • Web access to obtain gauge data
  • The global satellite mapping of precipitation project (GSMAP): Integration of MW and IR radiometers for a global precipitation map – T. Ushio (Japan)
  • HOAPS II: Global ocean precipitation database – C. Klepp (Germany)
  • HamburgOcean Atmosphere Parameters and Fluxes from Satellite Data = HOAPS
  • 15 years of data – 1987-2002; primarily from SSMI (precip, OWS) + AVHRR (SST)
  • Use all satellites available, intersatellite calibration
  • HOAPS-G – 0.5 degree, pentad, monthly, climo; HOAPS-S – orbital
  • GPCP is greater than HOAPS
  • No trends on global scale, but regional trends in 15-year period.
  • Klepp et al, 2003
  • Something seems wrong with his implementation of SSMI/Ferraro; contact him and follow up with code check.
  1. Validation and Error Analysis
  • Estimating bias of satellite-based precipitation estimates relative to in-situ meas. – Tom Smith
  • For climate scale, biases need to be assessed and adjusted for. Use gauges as reference.
  • Bias depends on both spatial and time scale; for climate scale, e-folding distance ~750 to 1000 km
  • Uses OI scheme to analyze bias fields; bias for various satellite techniques ~1-2 mm/day, but it has spatial structure
  • Ocean is a real problem since no gauges to adjust to; use TRMM?
  • Rainfall Estimation using satellite data – C. Oria Rojas (CPN/Peru)
  • Peru – very sparse gauge network and no radar; satellites extremely important for rainfall monitoring
  • Use 2 IR techniques – AE and GMSRA; showed validation. GMSRA appears to do better in Peru.
  • Developed some corrections for the algorithms – cirrus filters, TPW, etc.
  • IPWG Validation-Intercomparison Study – Beth Ebert (BMRC/Australia)
  • Stimulated by first IPWG plus precipitation needs in Australia gauge void areas
  • Also model precip. validation, but interestingly, model and satellite are complimentary – NWP better in stratiform, satellite better in convective
  • Focus is only daily rainfall and on data sets that are produced daily
  • CONUS Validation-Intercomparison Study – John Janowiak
  • Modeled after Beth’s web site
  • Showed some results; merged MW/IR seems to do the best, but the NCEP model still is better in winter seasons
  • Case study over Montana showed how the satellite estimates were in error due to low level RH/evap.
  • European Validation-Intercomparison Study – Chris Kidd (Univ. Birmingham, U.K.)
  • European continent – lots of coastlines, diverse topography and precipitation types
  • Also, plenty of light rain intensities
  • Main data to validate so far – ECMWF, CMORPH, CPCMW, 3B40-42 + locally generated algorithms
  • Data remapped to common projection; stats at 20 and 50 km
  • Radar as validation
  • Add AMSU to his validation
  • Joyce/AMSU low bias in Europe?
  • Verifying Precipitation Events using composite statistics – J. Nachamkin (NRL)
  • Looking for rain > 25 km; rather large events; compare PDF’s
  • BMRC gauge analysis vs. Turks blended product. Results show that for winter, satellite grossly low while in summer, satellite did very well, maybe too high in heaviest rainfall
  • Verifying TRaP over Australia for 2003-04 – Beth Ebert
  • 2003-04 season – 6 landfalling storms
  • Appears to be issues with SSM/I algorithm w.r.t. max rain rate; AMSU seems to do fine
  • Those TRaP’s that were vetted by SAB did seem to be worse than those that were operational
  • By sensor, AMSU seems to be best closely followed by TRMM, then SSMI worst
  • CRA/blob verification –pattern error was most important (48%), followed by volume (34%), then track error (18%)
  • Several areas of improvement – include storm rotation, better algorithms, atmospheric moisture/shear corrections, smoother to remove stripes in TRaP, etc.
  • Future – ensemble TRaP, probability forecasts
  • Describing diurnal cycle of precipitation from satellite observations – Phil Arkin
  • Use of CMORPH to look at various diurnal cycle features over CONUS and globe; interesting results. Even showed diurnal cycle trends in Hurricane Isabelle (midnight max)
  • Land gauge analysis – Bruno Rudolph (DWD, Germany)
  • Discussion of GPCC
  • Monitoring product – 7000 stations – available 2 months delay
  • Severe problems due to wind (up to 20%) and solid precipitation (factor of 3)
  • Rainfall estimation using an image classification approach – Y. Hong (UC-Irvine)
  • PERSIANN – for hydrological applications
  • Cloud classification scheme – location, cloud top temp, cloud size, texture and dynamics
  • Develop a cloud type – rainfall data base (ground and MW rain rates)
  • Doing snowfall retrievals – Sfc. Temp + elevation; future to use PR and NN technique
  1. Research Activities
  • The need for experimental data to support IPWG research activities – B. Bizzari (CNR-Italy)
  • An action from IPWG#1 and CGMS – Space agencies need to support the development of experimental data sets (e.g., aircraft) for precipitation/cloud modeling and instrument definition.
  • Geared towards need for GOMAS (Geostationary Observatory for MW Atmospheric Sounding) – scan 1/12 disk every 15 minutes with 10-30 km resolution. It would have bands around 54, 118, 183, 380 and 425 GHz.
  • Working with ETL/Gaisewski – promoting use of PSR; hope for flights in 2005 and 2006
  • NASA GEOS Rainfall Assimilation – Art Hou
  • Focus on TRMM on global model
  • A model needs physical consistency between rainfall and other variables; if the rainfall is incorrect, then something else is incorrect. Satellite retrieval algorithms should be concerned with the same issues. NWP rainfall is thus a forecast, not an analysis of precipitation based on observations and model information.
  • Results pulling in TMI, SSMI and AMSR-E translates into a 10% reduction in OLR errors
  • Mesoscale convective system life cycle over Del Plata Basin – D. Vila (INA, Argentina)
  • The FORTRACC technique – based on the fact that many GOES images are missed, so how do you account for this?
  • Case studies/validation/applications
  • Precipitation characteristics of landfalling typhoons in Taiwan – D. Jou (TaiwanUniv.)
  • 4-doppler radars on small island of Taiwan
  • Tropical storms – devastating floods
  • Focus of this talk – use radar to study precip. structure changes of Typhoons
  • Typhoon Nari – Sept 16, 2001 – affected Taiwan for many days.
  • RR increase 2%/hr during 12 hours before landfall, but 8%/hr 2 hrs before and 3 hr after landfall
  • Also looking at TRaP (tell Sheldon)
  • High Latitude Precip – Ralf Bennartz
  • Snowfall represents a large fraction of precipitation at mid and high latitudes
  • We can observe radiances and reflectivities, but not cloud microphysics, thus, we need models that relate the cloud microphysics to the radiometric measurements
  • Bennartz and Petty microphysics model (2001) – ice density and size of ice relative to liquid particles – can get consistent description of Z/fall speed/particle number concentration
  • Using AMSU and AMSR-E with Baltex radar data for RTE simulations; then used Tripoli CRM
  • Get fairly good agreement at 89 and 150 GHz; can expand to other frequencies to see how they might respond on future sensors (TB Jacobians) – example, 118 GHz
  • These cases had a precip. layer thickness of no more than 2 km.
  • Urges that snowfall algorithm cooperation be guided by IPWG, in particular, cloud microphysical models – Gruber mentions that the recent GRP is also suggesting a workshop on solid precipitation.
  • We need channels that are surface blind!
  • EGPM/Snowfall simulations – E. Smith for A. Mugnai
  • EGPM; ESA has put it on hold and redirected to another funding pot; review of mission
  • CRM simulations/working with Tripoli
  • 3 case studies presented (East coast U.S. –1/25/2000; Gotland; U.K.)
  • Summary – Snowfall over land/snow cover – high frequency and sounding channels needed; over ocean, have to deal with melting near SST. Radar will aid in the retrievals. For light rainfall over land, use of sounding channels is important.
  • Cloudsat snowfall/light rain algorithms – T. L’Ecuyer
  • Review of CloudSat – 94 GHz radar with -20 Dbz sensitivity; flown as part of “A-Train”
  • Cloudsat mainly for clouds, but shows promise for light precipitation; CRYSTAL-FACE, WakasaBay examples
  • TRMM radar (18 DbZ) misses quite a lot of light rainfall than contributes ~2-5% of rainfall in tropics and Colorado (but presumably, much more at higher latitudes)
  • Retrieval algorithm – Uses Rodgers optimal estimation formulation of inverse problem; get vertical profiles of IWC/LWC
  • Synthetic retrievals – seems like you can do well for rain rates <= 5 mm/hr
  • Snowfall retrieval algorithm developed in similar manner; uses 89 and 150 GHz (WakasaBay)
  • Both algorithms require PIA or PWP constraints
  • DSD and Surface Wind using dual-freq and X-band Radiometer – S. Brown (Univ. Michigan)
  • 13.4 and 35.6 GHz radar
  • DSD from pointing rain radars – A. Battaglia (Univ. Bonn/Germany)
  • Microwave Rain Radar (MRR)
  • GPM Era Precip. Algorithms – E. Smith
  • Formulation phase started 10/01 to be completed by 12/05
  • Review of mission status: GMI selection 12/04; DPR expected 4/05; H2A core launch mid 2010-11, still TBD, so that puts hold on EGPM (ESA’s Earth Watch Program, new, but not funded yet).
  • GPM Algorithm structure – include L2 DSD profiles, rainrate profiles and latent heat profiles. GV program for L2 products. L3 products include core satellite prodoucts, blended/intercalibrated passive MW products, Morphed MW/IR products
  • Discussion on GPM core reference algorithm – DPR-GMI Bayesian Algorithm
  • Looking for guidance from IPWG on algorithm requirements, formats, etc.
  • Evaluation of RTTOV8 at AMSU frequencies – Una O’Keeffe (UKMO)
  • RTTOV8 vs. N16 and ARTS (another RTE model)
  • Use mesoscale model as input into RTE
  • RTTOV 8Results – good at 89 GHz, but did not do well at 183 +/-7 GHz – could not handle precip. scattering; ARTS appears to do better. Why? Most likely, treatment of cloud microphysics/ice. Probably not accounting for enough large ice particles
  • Possible new size distributions to use: UK field campaign data (Field, 2004, QJRM, submitted)
  • Model works well for AMSU-A  CLW retrieval
  • Radiometer channel optimization for precipitation remote sensing – Peter Bauer (ECMWF)
  • Window (18, 23, 37, 89, 157) vs. sounding channels (50.3, 118.75 GHz); sounding channels have gotten more recent attention because of their constant weighting functions and being surface blind in clouds/precipitation.
  • 1-D var; advantage over cloud profile database since they are not tuned to CRM’s, however, they still require constraints (and may not converge). They are more computationally expensive.
  • Examination of signal to noise for a variety of precipitation systems; results as you might expect; 89, 118 and 150 GHz best over land, lower over ocean. Lower channels over land suffer from noise/errors in surface emissivity.
  • Channel optimization for rain, cloud water and snow: sounding channels seem to be better over land, window over ocean
  • Rainfall type estimation from the information of life stage of deep convection – T.Inoue (Japan)
  • Recognize that IR is less physical than MW, but we have GOES based temporal sampling. If IR information on life stage of convection is obtained, then maybe better rain rate estimates can be made.
  • Satellite precipitation estimates and the Central America FFG system - Rosario Alfaro (Costa Rica)
  • Running AE/HE in Central America.
  1. Working Groups (Operational, Research, Validation)

Other Actions:

  • Look at Jorge Schulz’s EUMETSAT paper on TPW kriging
  • Tristan L./ESSIC Seminar, CICS, etc.
  • GRP/IPWG/GPM/JCSDA sponsored solid precip workshop w/Bennartz

Operational Working Group

Attendees: BobJoyce (chair), Ralph Ferraro, Joe Turk, Jim Purdom, George Huffman, Arnie Gruber, Vincenzo Leviazzani, Deborah Smith, Jorge Sanchez-Sesma, Yang Hong, Rosario Alfaro, Tomoo Ushio, Thomas Heinemann,

  • Review of last years AI’s:

STANDARD/COMMON PROCEDURES: