1 Program to Evaluate High Resolution Precipitation Products (PEHRPP): An Update Matt Sapiano P. Arkin, J. Janowiak, D. Vila, Univ. of Maryland/ESSIC, College Park, MD Joe Turk, Naval Research Laboratory, Monterey, CA (Presenter) E. Ebert, Bureau of Meteorology, Melbourne, Australia

2 Outline Brief explanation of PEHRPP PEHRPP activities Workshop highlights Recommendations to IPWG

3 What is PEHRPP? A collaborative effort to understand the capabilities and characteristics of High Resolution Precipitation Products High Resolution = Daily/sub-daily; <1 degree Sponsored by IPWG with broad voluntary participation Capitalizing on existing research and operational activities/datasets Providing a link between the observational and application communities

4 PEHRPP Strategy PEHRPP is designed to exploit four kinds of validation opportunities: 1.Networks based on national or regional operational rain gauges or radar networks 2.High-quality time series from ongoing research programs GEWEX CEOP, TAO/TRITON buoy gauges Ethiopia, Sao Paolo 3.Field program data sets NAME, BALTEX 4.Coherent global scale variability as depicted by the various data sets - the big picture

5 Real-time radar/gauge comparisons (slide courtesy of C. Kidd, with additions) (See Ebert et al., BAMS, 2007)

6 Guangdong Validation Site: Jianyin Liang, CMA with Pingping Xie, NOAA April – June 2005 period of initial data (394 hourly real-time gauges) Guang-Dong Seasonal Mean Bias GaugeCMORPH 3B42RT 3B42MWCOMB

7 Sub-daily, high quality time-series Comparison against sub-daily TAO/TRITON buoys and US SGP sites –Correlations are generally high –Under-estimates over ocean –Over-estimates over US in Summer in absence of gauge correction –Models included and perform OK at daily, less well at sub-daily Sapiano and Arkin, J. Hydrometeorology, 2008 (in press) % Bias

8 North American Monsoon Experiment Precipitation Daily Evolution: NERN vs Satellite over NAME Domain (Nesbitt)

9 First PEHRPP Workshop Hosted by the IPWG 3-5 December 2007, WMO, Geneva 40 attendees from 12 countries Presentations and working group reports on applications, validation and error metrics

10 Presentations and discussions Talks: –Precipitation Products (5) –Regional Validation (10) –Applications (9) –Error Metrics (6) Advanced Blending Methodologies –Use of Kalman filter Updates on 8 Validation Sites –Northern Europe, Southern Europe, Japan, Brazil, Australia, Mozambique, Continental US, Western Africa Use of HRPP’s by Users of Hydrological and Mesoscale Forecast Models Improved and Relevant Error Metrics –Focused on user requirements Summary due to appear in December BAMS (Turk et al.)

11 Key recommendations Multiple recommendations were made, the key ones being: 1.Several high resolution precipitation products exhibit useful skill, but clear superiority for one is not yet evident: continuing activities are useful to this end 2.IPWG should establish a continuing effort to conduct, facilitate and coordinate validation and evaluation of such products 3.A concerted validation/intercomparison campaign, covering multiple climatic regimes and seasons, should be designed and conducted

12 Discussion PEHRPP has become a useful framework for validation activities on high resolution data –Not all elements have been addressed – Synthesis of results is still lacking – hence the need for a concerted campaign New leadership is required for activities to continue –Mandate needs to come from IPWG Currently working on joint proposal to WGNE to collaborate by including more model precip in PEHRPP –Ebert, Huffman, Kidd, Sapiano and others

13 Recommendation 1: Recommend an intercomparison project (similar to PIP,AIP) for the evaluation of HRPP. Products should aim for a standard of three-hourly, 0.25 degree resolution with global coverage, with validation done at the regional scale. Details of the inter-comparison (locations, temporal scale, etc.) will be charged to an intercomparison working group in association with the GPM working group to maximize the impact of such a comparison. The intercomparison should be completed in the next 24 to 36 months. Working Group Key Recommendations: VALIDATION Historical Background Precipitation Intercomparison Program (PIP), sponsored by NASA’s WetNet Project PIP-1: First assessment of SSMI precipitation algorithms on a global scale, Aug-Nov PIP-2: Examined SSMI precipitation algorithms on a case basis for multiple years, seasons, and meteorological events (Jul 1987-Feb 1993). PIP-3: Examined global scale precipitation algorithms over an entire year (1992). Algorithm Intercomparison Program (AIP), sponsored by the Global Precipitation Climatology Program AIP-1: Japan and surrounding region during Jun–Aug 1989, covering frontal and tropical convective rainfall. AIP-2: Western Europe during Feb–Apr 1991 with rainfall and snowfall over both land and sea regions. AIP-3: Tropical Pacific Ocean region (1°N–4°S, 153–158°E) during Nov 1992–Feb Kidd, 2001: Satellite Rainfall Climatology: A Review, Int. J. Climatol.,

14 Recommendation 2: Recommend that the outputs of the current and future validation efforts are better utilized: a working group should be formed under IPWG as a PEHRPP activity, and should report by the next IPWG meeting (October 2008). The co- chairs should be a product developer and validation site developer. Recommendation 3: Recommend the use of existing HRPP in hydrological impact studies, such as the EUMETSAT H-SAF and HydroMet testbeds in the US, to assess the usefulness of the HRPP products in hydrological models. Recommendation 4: Recommend that we include and/or encourage the development of high- latitude sites such as the BALTEX, LOFZY, high latitude maritime radar sites, and/or the Canadian sites. Recommendation 5: Recommend that countries or weather institutions with high quality ground validation dataset actively participate in IPWG sponsored validation activities. Working Group Key Recommendations: VALIDATION

15 Recommendation 1: Product developers should be encouraged to formulate and produce error estimates for the products, by: Engaging end users IPWG should investigate the forms of error required for applications Engaging other product developers Since full error estimates will take time to obtain, developers should be encouraged to make other information available such as the main source of data (i.e. SSM/I F-13 GPROF V6) and the latency of PMW data (time since last MW overpass) Working Group Key Recommendations: APPLICATIONS

16 Recommendation 2: PEHRPP/IPWG should make satellite organizations aware of the fact that PMW data are useful for a broad range of applications and that these applications would benefit from more data, faster data delivery and the maintenance of all existing data streams. Recommendation 3: Product developers should be encouraged to pursue other assimilation and/or downscaling methodologies which exploit all available information (satellites, NWP, gauges, lightning estimates), particularly those which are optimized for specific applications. Working Group Key Recommendations: APPLICATIONS

17 There is a general feeling that the current understanding of HRPP quality/certainty/errors suffers from a lack of adequate error metrics that are pertinent to users and well-understood Long-term Recommendations: Physically based error characterization of retrievals (key element of GPM) Consistent set of “basic” metrics Comprehensive quantitative error model that allows users to specify time and space scales, give the space-time… coefficients associated with a precip data set, and obtain estimated RMS error (diagnostic) or create synthetic precip fields (prognostic) Work towards an assimilation-like method for combinations Working Group Key Recommendations: ERROR METRICS

18 Working Group Key Recommendations: ERROR METRICS Short-term Recommendations: Develop a standing working group on error metrics Agree on a short list of error metrics – each needs confidence intervals -“traditional” metrics that give insight at the scales of interest -other metrics suggested by the long-term vision -fuzzy validation framework -WWRP/WGNE Joint Working Group on Verification list of metrics -diagnostics (PDFs, conditional statistics, …) -examine using transformed data in metrics Test practicality of these metrics for producers and utility for users -Inter-satellite errors (Joyce/NOAA subsetted gridded (30-min, 0.25°) precipitation data sets from ~15 satellites/sensors) -Characterizing errors by regime -Establishing some minimum set of space/time correlations that are needed