1. The Hydro-Estimator and Hydro-Nowcaster: Satellite- Based Flash Flood Forecasting Tools Robert J. Kuligowski NOAA/NESDIS Center for SaTellite Applications and Research (STAR) Camp Springs, MD USA with special appreciation to Rod Scofield (1942-2006) who pioneered much of this work International Workshop on Flash Flood Forecasting 14 March 2006

2. Outline The Hydro-Estimator (HE) The Hydro-Estimator (HE) The Hydro-Nowcaster (HN) The Hydro-Nowcaster (HN) Future Work Future Work Where to Get the Data Where to Get the Data International Collaboration International Collaboration

3. The Hydro-Estimator (HE): Introduction The HE is NESDIS’ operational algorithm for estimating rainfall rate from infrared (IR) window (~11 µm) brightness temperatures. The HE is NESDIS’ operational algorithm for estimating rainfall rate from infrared (IR) window (~11 µm) brightness temperatures. Real-time estimates are produced globally (60°S to 60°N) at the spatial and temporal resolution of the available geostationary satellite data Real-time estimates are produced globally (60°S to 60°N) at the spatial and temporal resolution of the available geostationary satellite data –4-5 km resolution generally every 15-30 minutes, depending on location.

4. Hourly rainfall estimates for 0000-0900 UTC 5 January 2005 HE Example—Global Coverage

5. HE Example—Fine-Scale Detail 15-minute rain rate estimates estimates for 1815-2215 UTC 10 July 2005 (Hurricane Dennis landfall)

6. The Hydro-Estimator: Basic Assumptions Assumes that cloud-top brightness temperature (T b )  cloud-top height  strength of convective updraft  rainfall rate Assumes that cloud-top brightness temperature (T b )  cloud-top height  strength of convective updraft  rainfall rate –Colder, higher clouds are associated with stronger updrafts and heavier rain –Warmer, lower clouds are associated with weak updrafts and light or no rain Reasonable assumption for convective (thunderstorm) clouds Reasonable assumption for convective (thunderstorm) clouds

7. 200 250 290 T (K) T b =230 K T b =224 K T b =212 K T b =200 K Illustration of the IR signal from different rainfall intensities

8. 290200250 T (K) Cumulonimbus T b =200 K Nimbostratus T b =240 K Cirrus T b =205 K Exceptions to the Rule...

9. The Hydro-Estimator: Rain/No Rain Differentiation The original NESDIS automated IR algorithm, the Auto-Estimator, routinely assigned rainfall to cirrus clouds and thus greatly exaggerated the spatial extent of heavy rainfall. The original NESDIS automated IR algorithm, the Auto-Estimator, routinely assigned rainfall to cirrus clouds and thus greatly exaggerated the spatial extent of heavy rainfall. The Hydro-Estimator avoids this problem by considering temperature relative to the local average: The Hydro-Estimator avoids this problem by considering temperature relative to the local average: –Colder than average=active rain area –Warmer than average=inactive cold cloud

10. Illustration of the HE Rain-No Rain Differentiation 290200250 T (K) T b < T b Rain T b ≥ T b No Rain T b < T b Rain T b ≥ T b No Rain

11. H-E Rainfall Rates Initial rain rates are a function of both T b and its value relative to surrounding mean Initial rain rates are a function of both T b and its value relative to surrounding mean Adjustments using numerical weather prediction model data, including: Adjustments using numerical weather prediction model data, including: –Precipitable water to enhance rainfall in regions of high moisture availability –Relative humidity to reduce precipitation in arid regions where raindrops evaporate before reaching the ground –Wind fields interfaced with digital topography to determine orographic effects where wind blows: up slope (moistening and enhancement of rain) or up slope (moistening and enhancement of rain) or down slope (drying and reduction of rain) down slope (drying and reduction of rain)

12. HE PW Adjustment Sample Hydro-Estimator Rain Rate Curves as a Function of Precipitable Water PW (mm) Wetter environment

13. The Hydro-Nowcaster: Introduction The Hydro-Nowcaster (HN) is an algorithm for producing 0-3 h nowcasts of precipitation based on extrapolated Hydro-Estimator estimates of rainfall rate. The Hydro-Nowcaster (HN) is an algorithm for producing 0-3 h nowcasts of precipitation based on extrapolated Hydro-Estimator estimates of rainfall rate. Both advection and growth/decay are considered. Both advection and growth/decay are considered.

14. How the Nowcaster Works— Extrapolation Identifies clusters (regions bounded by fixed brightness temperature values) on two consecutive IR images Identifies clusters (regions bounded by fixed brightness temperature values) on two consecutive IR images Determines cluster motions based on the shift of the coldest 25% of pixels within a 100x100- pixel area that produces the best correlation between the two images Determines cluster motions based on the shift of the coldest 25% of pixels within a 100x100- pixel area that produces the best correlation between the two images Cloud motions are extrapolated out to 3 h at 15-min intervals based on the resulting motion vectors Cloud motions are extrapolated out to 3 h at 15-min intervals based on the resulting motion vectors

15. t=t 0 -Δt t=t 0 Vector for maximum correlation between t=t 0 -Δt and t=t 0 Extrapolated motion vector Illustration of the motion vector identification in the Hydro-Nowcaster

16. How the Nowcaster Works— Growth/Decay Each cluster on the current image is matched with one on the previous image according to the computed motion vector Each cluster on the current image is matched with one on the previous image according to the computed motion vector Three factors to determine growth/decay: Three factors to determine growth/decay: –Change in size of the cluster –Change in temperature of the coldest pixel –Change in mean temperature of the cluster The growth/decay factor linearly decays to zero over the 3-h forecast period to avoid unrealistic results The growth/decay factor linearly decays to zero over the 3-h forecast period to avoid unrealistic results

17. Illustration of time change information in the Hydro-Nowcaster t=t 0 -Δt t=t 0 Growing cluster; enhance rain rates during extrapolation Shrinking cluster; reduce rain rates during extrapolation Warming cluster; reduce rain rates during extrapolation

18. Example: Hurricane Katrina on 29 August 2005 1 h nowcast: 1200–1300 UTC 3-h nowcast: 1200–1500 UTC

19. Future Work The Hydro-Estimator is currently being recalibrated to address weaknesses, particularly underestimation of heavy rainfall from warm clouds The Hydro-Estimator is currently being recalibrated to address weaknesses, particularly underestimation of heavy rainfall from warm clouds Hydro-Nowcaster improvements are planned: Hydro-Nowcaster improvements are planned: –Improve the scheme for depicting cloud growth and decay—currently empirical –Develop an advection scheme for circular storms— motion vectors for multiple lags? –Account for the effects of orography on nowcasts of rainfall

20. Additional Plans Beginning collaboration between City College of New York, NOAA/National Weather Service, and NESDIS to evaluate and test multiple nowcasting frameworks (HN, TITAN, RDT) over the New York City metropolitan area. Beginning collaboration between City College of New York, NOAA/National Weather Service, and NESDIS to evaluate and test multiple nowcasting frameworks (HN, TITAN, RDT) over the New York City metropolitan area. Focus is on NYC, but results will be considered in operational satellite-based nowcasting development in the US. Focus is on NYC, but results will be considered in operational satellite-based nowcasting development in the US.

21. Where to Get Hydro- Estimator Data Graphic images of Hydro-Estimator fields worldwide are available in real time at http://www.orbit.nesdis.noaa.gov/smcd/emb/ff/ http://www.orbit.nesdis.noaa.gov/smcd/emb/ff/

22. Where to Get Hydro-Nowcaster Data Nowcasts for 1, 2, and 3 h will be updated every 15 minutes for the entire CONUS on the NESDIS Flash Flood Web page: http://www.orbit.nesdis.noaa.gov/smcd/emb/ff/

23. International Collaboration The Instituto Meteorológical Nacional in Costa Rica is running the Hydro-Estimator in real time as input to the Central American Flash Flood Guidance (CAFFG) system, which supports 7 Central American countries. The Instituto Meteorológical Nacional in Costa Rica is running the Hydro-Estimator in real time as input to the Central American Flash Flood Guidance (CAFFG) system, which supports 7 Central American countries. The Comisión Nacional del Agua in Mexico is also running the Hydro-Estimator in real time for forecaster support. The Comisión Nacional del Agua in Mexico is also running the Hydro-Estimator in real time for forecaster support. The Hydro-Estimator and Hydro-Nowcaster source code can be made available to interested parties—contact me at Bob.Kuligowski@noaa.gov The Hydro-Estimator and Hydro-Nowcaster source code can be made available to interested parties—contact me at Bob.Kuligowski@noaa.govBob.Kuligowski@noaa.gov

24. Questions?