Operational Consensus Forecasts (OCF) and official forecasts of daily maximum and minimum temperature in 2006 are verified for 107 locations where both forecasts are available. At eight capital cities the forecasts extend from one to six days ahead but for the remaining 99 locations the official forecasts usually extend only one day ahead. Over all sites combined, the next day’s forecast from OCF is better than official forecasts for both maxima (61 per cent of sites; mean absolute error (MAE) 1.28°C compared with 1.31°C) and minima (75 per cent; MAE 1.48°C compared with 1.53°C). The comparative verification results revealed detailed systematic differences between the two forecasting schemes. For example, the official forecasts of capital city maxima were usually more accurate than OCF whereas for forecasts of capital minima OCF were more accurate. Forecasters at the three capital cities where official forecasts of maximum temperatures were clearly better than OCF provided summaries of their perceived weaknesses in OCF that enables them to improve on OCF guidance. The relative weaknesses of OCF and official forecasts are discussed with a view to improving future performance. The main suggestions arising from this study are that OCF may be improved by: (a) basing OCF bias-correction on wind analogues or developing an OCF model output statistics scheme; and (b) rationalising the number of numerical models used in OCF to reduce the over-representation of the Australian regional numerical model in its one to two days ahead forecasts and increasing the number of numerical models that contribute beyond two days ahead. Official forecasts should benefit from documentation of the meteorological situations where OCF performs both well and poorly then using that information when formulating official forecasts. Official forecasts of minimum temperature generally and maximum temperature at most locations would be more accurate if more weight were given to the OCF guidance.
Woodcock et al. (Sun,) studied this question.