18th Conference on Weather and Forecasting, 14th Conference on Numerical Weather Prediction, and Ninth Conference on Mesoscale Processes

Wednesday, 1 August 2001: 10:30 AM
Verification of cloud forecasts with in-situ aircraft measurements
Hong Guan, MSC, Downsview, ON, Canada; and S. G. Cober and G. A. Isaac
The Meteorological Service of Canada has conducted several winter field projects during the past decade, including the Second Canadian Atlantic Storms Program, the First and Third Canadian Freezing Drizzle Experiments, and the Alliance Icing Research Study. A common objective of each project was to improve the accuracies for cloud and aircraft icing forecasts. Since in-situ aircraft based measurements include extensive microphysical parameters that are not normally measured at meteorological stations, these measurements were used to validate the predictions of meteorological fields that were made with the High Resolution Model Application Project version of the Global Environmental Multiscale model. Specifically, forecasts of temperature, horizontal winds, dew point, cloud and supercooled cloud water (SCW) were compared with the in-situ aircraft measurements. The aircraft data were also used as a basis for intercomparisons of four cloud schemes and two SCW schemes. The tested cloud schemes included the Sundqvist scheme (SUND) (Sundqvist 1989), explicit moisture scheme (HSIE) developed by Hsie et al. (1984), mixed-phase scheme (MIX) (Temblay et al. 1996), and explicit microphysics scheme (KY) (Kong and Yau 1997).

The forecast accuracy for temperature, horizontal winds, and dew point agreed closely with the results from radiosonde-model validation experiments, which gave confidence in the aircraft-model validation methodology. The hit rate (HR), false alarm rate (FAR), and true skill statistic (TSS) for the Canadian operational cloud (icing) forecasts were 0.52 (0.37), 0.30 (0.22), and 0.22 (0.15) respectively, when the model data were inferred at a horizontal resolution of 1.5 km. The HRs (FARs) for cloud and SCW events are sensitive to horizontal resolution and increase to 0.76 (0.50) and 0.66 (0.53), respectively, when a horizontal resolution of 100 km is used.

Intercomparisons of the four explicit cloud schemes with the aircraft data gave similar TSS values, varying between 0.27 for SUND and 0.34 for KY. The TSS for SCW forecasts was 0.15 and 0.30 for the MIX and KY schemes respectively. Quantitative comparisons show that the total water content (water drops and ice crystals) obtained by the MIX scheme agreed mostly closely with the aircraft measurement, while the HSIE and KY schemes overestimated TWC. The forecast ice water contents for the HSIE, MIX and KY schemes are too large. Although the KY scheme produced a higher TSS score, the predicted supercooled liquid water content was normally substantially higher than the observations, implying that improvements to the microphysics parameterizations are necessary. Strengths and weaknesses for each cloud and SCW scheme will be discussed.

Supplementary URL: