Experimental Determination of Forecast Sensitivity and the Degradation of Forecasts through the Assimilation of Good Quality Data

Abstract

he case of a small vigorous cyclone crossing the United Kingdom on 1 November 2009 is investigated. Met Office Global Model forecasts at the time displayed a marked change in solutions at a forecast range of 72 h, with those at longer ranges being more representative of the correct solution and those at shorter ranges only gradually migrating toward it. The strong bimodal nature of the Global Model forecasts is enough to overwhelmingly dominate the solutions from the Met Office Global Ensemble on which it is based. An investigation into the case is used as a vehicle for developing an experimental method determining the critical location of assimilated data leading to the largest impact on forecast consistency and the origins of the bimodal solutions. It allows the identification of one global positioning system radio occultation (GPSRO) and three surface observations located around the developing low that have conclusively led to the degradation in forecast skill. An assessment of these observations concludes that they are of relatively good quality and correctly assimilated. The case is suggested to be an example of forecast degradation as a result of the addition of growing errors by the data assimilation scheme.