Ability of a Poor Man's Ensemble to Predict the Probability and Distribution of Precipitation

A poor man's ensemble is a set of independent numerical weather prediction (NWP) model forecasts from several operational centers. Because it samples uncertainties in both the initial conditions and model formulation through the variation of input data, analysis, and forecast methodologies of its component members, it is less prone to systematic biases and errors that cause underdispersive behavior in single-model ensemble prediction systems (EPSs). It is also essentially cost-free. Its main disadvantage is its relatively small size. This paper investigates the ability of a poor man's ensemble to provide forecasts of the probability and distribution of rainfall in the short range, 1–2 days. The poor man's ensemble described here consists of 24- and 48-h daily quantitative precipitation forecasts (QPFs) from seven operational NWP models. The ensemble forecasts were verified for a 28-month period over Australia using gridded daily rain gauge analyses. Forecasts of the probability of precipitation (POP) were skillful for rain rates up to 50 mm day−1 for the first 24-h period, exceeding the skill of the European Centre for Medium-Range Weather Forecasts EPS. Probabilistic skill was limited to lower rain rates during the second 24 h. The skill and accuracy of the ensemble mean QPF far exceeded that of the individual models for both forecast periods when standard measures such as the root-mean-square error and equitable threat score were used. Additional measures based on the forecast location and intensity of individual rain events substantiated the improvements associated with the ensemble mean QPF. The greatest improvement was seen in the location of the forecast rain pattern, as the mean displacement from the observations was reduced by 30%. As a result the number of event forecasts that could be considered “hits” (forecast rain location and maximum intensity close to the observed) improved markedly. Averaging to produce the ensemble mean caused a large bias in rain area and a corresponding reduction in mean and maximum rain intensity. Several alternative deterministic ensemble forecasts were tested, with the most successful using probability matching to reassign the ensemble mean rain rates using the rain rate distribution of the component QPFs. This eliminated most of the excess rain area and increased the maximum rain rates, improving the event hit rate. The dependence of the POP and ensemble mean results on the number of members included in the ensemble was investigated using the 24-h model QPFs. When ensemble members were selected randomly the performance improved monotonically with increasing ensemble size, with verification statistics approaching their asymptotic limits for an ensemble size of seven. When the members were chosen according to greatest overall skill the ensemble performance peaked when only five or six members were used. This suggests that the addition of ensemble members with lower skill can degrade the overall product. Low values of the spread–skill correlation indicate that it is not possible to predict the forecast skill from the spread of the ensemble alone. However, the number of models predicting a particular rain event gives a good indication of the likelihood of the ensemble to envelop the location and magnitude of that event.