Comments on The Use of GPM/IMERG at The Weather Company Todd Hutchinson Director of Numerical Weather Prediction Global Forecasting Services The Weather Company

The Weather Company The Weather Company is one of the world’s largest private weather companies. We provide services to a large number of consumers and broad spectrum businesses Direct Consumer Portals The Weather Channel Weather.com Weather Underground Indirect Consumer Portals Apple Yahoo! Google Majority of US local TV broadcasters B2B services Aviation: 55,000 flights a day, 85% in US. Energy: Forecasts and analytics to Energy Traders Insurance and Reinsurance: Broad suite of weather svcs Services to Governments and Agencies

Forecasting at The Weather Company We create our own global forecasts. We collect data from governments and partners to fuel our forecast generation Standard WMO data NWP data from global centers Remotely sensed data (Radar, Satellite, profilers, etc) Short-fuse government warnings We augment the acquired data with our own proprietary content (global lightning, mesonets, A/C data). We run our own NWP models (WRF-ARW, with 3D Var/Warm Cloud initialization) and climate models (CSM). Humans can influence or augment the forecasts through an efficient editing environment. We package the end result in useful and compelling ways. Broad Input Data We acquire weather and forecast data from governments and partners around the globe Broad Input Data We acquire weather and forecast data from governments and partners around the globe Proprietary Data We have our own sources for additional weather data, and we run our own NWP models Proprietary Data We have our own sources for additional weather data, and we run our own NWP models Forecast Engines We have invested heavily in market specific analytical and statistical forecasting processes Forecast Engines We have invested heavily in market specific analytical and statistical forecasting processes Human Touch Humans can improve, augment and enrich the forecasts Human Touch Humans can improve, augment and enrich the forecasts Useful Products We contain our forecasts in compelling ways in market specific products Useful Products We contain our forecasts in compelling ways in market specific products

General Use of Satellite Data at TWC In general, satellite data drives TWC content in the following expected ways: Forecasting Situational Awareness: Location and evolution of important weather features, primarily using geostationary data. Through Indirect Forecasting Applications: Assimilation of satellite imagery and derived products into NWP in our own models; Use of satellite-initialized models from government centers; Use in automated nowcasting systems. “Story Telling”: Satellite imagery and derived products as a direct method to convey weather and forecasting stories and their impacts.

But, Use of POES data is increasing Mapping the Global state of precipitation o We want to tell users what the weather is “now”: o “Outside, the temperature is currently 75 degrees with light rain falling” o And, if precipitation is imminent (nowcasting) o Requires melding of POES, GOES, NWP (and other) data sources Graphical animations of global precipitation o Provides unique content (oceans, international) o Compelling global precipitation graphics Numerical weather prediction initialization o Currently, TWC incorporates POES through analyses from National NWP centers o By assimilating POES data directly, we can o Reduce forecast latency o Incorporate the latest science from state-of-the-art data assimilation systems

Recent Analysis of GPM/IMERG: Stats How well does distribution of GPM/IMERG precipitation match radar (“truth”)? o Analyzed GPM/IMERG data valid at 06 and 18 UTC between 12 March and 30 June 2014 o Converted GPM precipitation rate to reflectivity (dBz) using Marshall-Palmer (Z = 200 * R ^ 1.6) o Binned number of GPM “reflectivity” values depicted at each observing site (~1000 CONUS sites) o Similarly, analyzed WSI/TWC NOWRad Mosaic: o Binned reflectivity obs depicted at same observing sites X 100 TWC NOWRadIMERG Same counts of “no precip” = same count of all precip IMERG too much heavy precip – Z-R might explain a few dBz IMERG too little light precip

GPM/IMERG Analysis 1.Nice representation of overall precipitation field 2.Correspondence of Kalman Weights to snow/cold areas 3.Some areas of lighter precipitation well depicted 4.Some areas of lighter precip not well depicted 5.Some discontinuities in GPM 100% GEO 100% GPM 1” 0 GPM/IMERG Final – 27 Nov Kalman Weights Precip Accum.

NOWrad (“truth”) 12 UTC June

Compare “Early” to NOWrad 12 UTC 5 June 2015

Compare “Late” to NOWrad 12 UTC 5 June 2015

Current usage at TWC and Wish List for future GPM will offer many new products that enhance the usability of the products for TWC use: – Better global near real-time precipitation analysis – Better data assimilated within models – Potential particle discrimination / volcanic ash (?) Broader use at TWC is occurring with these recent enhancements: – Regular Real-time internet data streams of GPM/IMERG – Self-service portal to “research quality” data analysis and product/graphical generation. – Relationships with GPM core science organizations for better “use of opportunity”. – Availability of “Early” run Even Broader use at TWC would occur with: – High-availability data streams – Reduced latency for “Early” run to < 4 hrs – More MW coverage and smoother transition to IR coverage – Smoother transition from MW to IR – Improved depiction of light precip in GPM/IMERG