UK Met Office MJO Forecast Methodology


A fifteen day forecast is first carried out using the UK Met Office Global and Regional Ensemble Prediction System. Then the MJO real time multivariate index is calculated from the model results using the following method:

The MJO index from UK Met Office Unified Model Analyses are calculated first.

This follows all the same steps as outlined in Wheeler and Hendon and uses the same codes as used by them to calculate the index from observations. These steps are:

  1. remove the first 3 harmonics of the seasonal cycle to get an anomaly.
  2. remove the effect of the ENSO signal using the SST1 index calculated by the Bureau of Meteorology.
  3. remove a 120 day running mean using the last 120 days of the analysis. Any missing times are filled in with linear interpolation in time.

The climatologies used to calculate the seasonal cycle and the ENSO influence are also those used at the Bureau of Meteorology. This should not be a problem for analysed winds but might be more of an issue with OLR. However some of the systematic differences between observed and model OLR will be removed by the 120 day mean and also I have found OLR contributes less to the final index than the winds.

The resulting MJO analysis is in good agreement with the ones calculated from NCEP winds and observed OLR.

Then the same steps are applied to each ensemble forecast. The only difference is that the day 1 forecast removes a 120 day mean of the forecast from day 1 and the 119 previous analyses. Day 2 uses 118 analyses and 2 forecasts etc.