Verification of Thermal Forecasts with Glider Flight Data

Abstract

The soaring flight planning and analysis algorithm TopTask was used for multi-class national and international gliding championships in Switzerland and Sweden.  The routinely operated meteorological models LME-TOPTHERM and HIRLAM provided the regional thermal forecasts required by the TopTask algorithm: the depth of the convective boundary layer, the lift rate and the horizontal wind at flight altitude.  TopTask predicted speeds for the set tasks were compared to the scored speeds of the first and second place finishers for all days and classes of these competitions.  For Swiss Glide 2004 the predicted task speeds matched the scored speeds with no bias and a standard deviation of 10%.  For Viking Glide 2005 the predicted task speeds (108+/-8 kph) matched the scored speeds (104+/-15 kph) with a bias of 4 kph.  The standard deviation for Viking Glide 2005, however, was larger than for Swiss Glide 2004 with the appearance of three classes of accuracy: accurate predictions (speed within +/-10%), underestimated predictions (scored speed up to 22% higher than predicted), and overestimated predictions (scored speed up to 39% lower than predicted).  TopTask also was used for a model intercomparison between the HIRLAM and the LME-TOPTHERM models.  TopTask predicted flight speeds based on the regional forecasts from both meteorological models were compared to the scored speeds for Viking Glide 2005.  The task speeds obtained from HIRLAM (118+/-5 kph) were higher than those from LME-TOPTHERM (108+/-8 kph) and indicated a bias in the predicted lift rates.  The width of the HIRLAM speed forecasts of +/-5 kph was smaller than the +/-8 kph of the LME-TOPTHERM speed forecasts and the ranges of both speed predictions were smaller than the range of the scored speeds (104+/-15 kph).  Thus, both meteorological models do not yet catch the full range of variation found in the scored speeds of the winning gliders.  Nevertheless, routine numerical thermal forecasts have reached a quality that is useful for soaring practice.  Pilot briefing systems make such forecasts available and provide the possibility of thermal flight planning to the gliding community.  Glider flight data can be used to verify the performance of numerical weather prediction for thermals.