Meteorologische Zeitschrift, Fast Track DOI 10.1127/0941-2948/2011/0265 Fast Track Article c ⃝ by Gebr ¨ uder Borntraeger 2011 The RMS TC-Rain Model J ¨ URGEN GRIESER ∗ and S TEVE J EWSON Risk Management Solution Ltd., Peninsular House, 30 Monument Street, London EC3R 8NB, United Kingdom (Manuscript received November 11, 2010; in revised form March 3, 2011; accepted March 7, 2011) Abstract Tropical-cyclone rain causes inland flooding, land-slides and mud-slides, and these perils are a risk for human life and property in areas prone to tropical cyclones. We describe a parameterized tropical-cyclone rain model which is simple enough to be run several hundred thousand times and hence can be used to estimate these risks. The model is based on climatological off-shore rain rates parameterized by R-CLIPER, but with a number of important improvements: the model takes into account rain enhancement due to orography and landfall; rain rates are spatially distributed to match the asymmetry of observed tropical cyclone rain, including rain bands and the effects of extratropical transition, and rain drift is parameterized as a function of wind speed. The model is calibrated with respect to observations of 358 US-landfalling tropical cyclones during the period 1948 to 2007. Although not designed for forecasting the model is then used to forecast the rainfall for six US-landfalling tropical cyclones in 2008 (the most recent active season) as an independent test. A statistical track set consisting of tracks of North-Atlantic tropical cyclones covering 100,000 years is used in order to estimate maps of return levels of tropical cyclone rain for the US mainland. Zusammenfassung Die Niederschl¨ age tropischer Wirbelst¨ urme verursachen ¨ Uberschwemmungen, Sturzfluten und Erdrutsche. Sie sind somit eine Gefahr f¨ ur menschliches Leben und Eigentum. Wir stellen ein parameterisiertes Mo- dell f¨ ur den Niederschlag aus tropischen Wirbelst¨ urmen vor. Es ist einfach genug, so daß es mehrere hun- derttausend mal angewendet werden kann und somit f¨ ur Risikoabsch¨ atzungen verwendbar ist. Das Model basiert auf den klimatologischen Regenraten des Models R-CLIPER, geht aber durch die Parametrisierung zahreicher in R-CLIPER vernachl¨ assigter Prozesse weit dar¨ uber hinaus. So wird die lokale Verst¨ arkung des Niederschlags durch Orographie sowie die ¨ Anderung der Bodenrauhigkeit an der K¨ uste parametrisiert. Ty- pische r¨ aumliche Muster des Niederschlags tropischer Wirbelst¨ urme, wie ¨ aussere Regenb¨ ander, das Auge, Abweichungen von der Rotationssymmetrie und Extratropische Transition werden ber¨ ucksichtigt. Gegeben die hohen Windgeschwindigkeiten in tropischen Zyklonen wird auch die Niederschlagsdrift parameterisiert. Das Model ist anhand von 358 Wirbelst¨ urmen kalibriert, die im Zeitraum von 1948 bis 2007 Niederschlag auf das US-amerikanische Festland gebracht haben. Obwohl das Model nicht zur Vorhersage einzelner St¨ urme erstellt wurde, wurde es als unabh¨ angiger Test f¨ ur die Vorhersage des Niederschlags von sechs tropischen Wirbelst¨ urmen verwendet, die im Jahr 2008 das US-amerikanische Festland heimsuchten. Ein 100.000 Jahre umfassender statistischer Datensatz von Zugbahnen tropischer Wirbelst¨ urme im Nordatlantik wurde verwen- det, um lokale Wiederkehrzeiten extremer Niederschl¨ age in den USA zu sch¨ atzen. 1 Tropical-cyclone rain Rain is one of the three major perils of tropical cyclones (TC), together with wind and storm surge. For example, the highest ever observed rain amounts on timescales be- tween several hours and several days have occurred dur- ing TCs in La Reunion, where tropical storms approach over particularly warm ocean water and hit an island with high mountains and steep slopes (BURT, 2007). As a result rain amounts of up to 1870 mm within one day have been observed. High rain rates threaten human life. In the US 59% of the fatalities caused by tropical storms during the 30- year period before the year 2000 were due to freshwa- ter flooding caused by TC rain (RAPPAPORT, 2000). A particularly devastating example is TC Camille (1969), the remnants of which caused 113 deaths, mainly due to ∗ Corresponding author: J¨ urgen Grieser, Risk Management Solutions Ltd., Peninsular House, 30 Monument Street, London EC3R 8NB, United King- dom, e-mail: juergen.grieser@rms.com landslides and flash floods resulting from over 680mm rain within only 8 hours. These torrential rains resulted from a weakened tropical storm hitting the Appalachi- ans. According to SCHWARTZ (1970) this was the worst natural disaster in the history of Virginia. Financial losses due to TC rain can also be ex- tremely high. Some examples of recent TCs with high rain-related financial losses in the US are Fran (1996), Georges (1998), Floyd (1999), and Allison (2001). The latter was a particularly weak tropical storm without any wind losses. The total loss, however, was US$ 6.77bn (2010USD) and was solely rain driven. Given this impact, it would be useful to estimate the risk of inland flooding due to TC rain more accurately than is presently possible. The goal of this work is to describe a TC-rain model which, on the one hand, is simple enough to be run several hundred thousand times, and can hence be used to estimate extreme risks, and, on the other hand, is based on the main physics of the underlying processes. We first review some existing rules and simple pa- 0941-2948/2011/0265 $ 0.00 DOI 10.1127/0941-2948/2011/0265 c ⃝ Gebr¨ uder Borntraeger, Berlin, Stuttgart 2011