Page tree
Skip to end of metadata
Go to start of metadata

Re-forecasts are a fundamental component of all seasonal forecasting system; they have two applications:

  1. Seasonal forecast verification metrics are based on the re-forecasts
  2. The re-forecasts allow us to compute the S-M-climate, which in turn allow the actual forecasts to be converted into an anomaly format. Expressing forecasts in terms of anomalies relative to a model climate (rather than relative to the observed climatology) means that we are incorporating some calibration for model bias and drift into the products.

Seasonal Re-Forecasts for Verification

To evaluate our seasonal forecast system, a re-forecast set has been used.  It has been created using the same calendar start date (the 1st of each month) during a fixed period (36 years: Jan 1981 to Dec 2016).  Currently, this is not updated.  It consists of a 25-member ensemble (all members are perturbed) run over a 7-month forecast period for monthly runs and a 13-month forecast period for quarterly runs.  Therefore altogether 36 years x 25 ENS members = 900 re-forecast values are available for each forecast parameter, forecast lead-time, calendar start date (the 1st of each month) and location.  These are used in their entirety for skill and reliability assessment, and verification maps are accordingly provided on the web.

Whilst there is of course some merit in examining the real-time performance of a seasonal forecasting system, the sample sizes created for one system are far too small to conclude anything about the true performance levels of that system. Hence we use the re-forecasts.

The S-M-climate

For actual forecast products, where one uses the re-forecasts to try to cancel out drift etc., the re-forecast period utilised is a subset of that used for verification. Currently it is the 24 year period Jan 1993-Dec 2016.  Therefore altogether 24 years x 25 ENS members = 600 re-forecast values are available. This defines the S-M-climate for each forecast parameter, forecast lead-time, calendar start date (1st of each month) and location.  This is shorter than the full period available from the re-forecasts, and closer overall to the current date, for two reasons: (i) to be compatible with other seasonal forecast products provided through the auspices of Copernicus (this is their standard re-forecast period), and (ii) to be more compatible with "current climate" (which is changing primarily due to global warming).

  • No labels