Environmental Sources of Error in the Navy ESPC MJO Forecasts and MJO-Teleconnections

The boreal winter Madden-Julian Oscillation (MJO) in Navy Earth System Prediction Capability (ESPC), is known to be too strong due to biases in the vertical moisture advection and to propagate too quickly in the western Pacific due to excessive evaporation in this region, which supports moistening to the east of the MJO convective center. In this study, we seek to better understand the sources of errors in individual MJO events and identify the impact of mean state biases in forecasts where the Navy ESPC predicts the MJO well, and those where the MJO is poorly predicted. The Navy ESPC is a fully coupled global model that has been used to produce 45-day 16-member ensemble operational forecasts since August 2020.

Individual MJO events are tracked following Chikira (2014), using meridional means between 10N and 10S of MJO-filtered outgoing longwave radiation (OLR). Several MJO metrics are compared, including phase speed and amplitude, and the tracking algorithm identifies the peak amplitude, the central track, and the shape of the MJO in longitude-time space. The MJO forecasts in the Navy ESPC are split into those that are well predicted and those that are poorly predicted. A forecasted MJO event is considered well predicted if 1) an MJO exists in both the forecast period and the corresponding observational period, and 2) if the forecasted MJO event overlaps with the observed event in longitude and time by at least 30%. If the MJO exists in the observations but not in the forecasts, and vice versa, the forecast is considered a poor MJO forecast.

The structure and phase speed of both the good and the poor MJO forecasts are similar to the observed MJO characteristics. This suggests that the structure of the MJO in the Navy ESPC is fairly robust, even when the MJO is poorly forecast. There are compelling differences in the environment in which the good and poorly forecasted MJOs form. Both the good and poor MJO forecasts occur in environments that are too dry, but the good MJOs have slightly stronger moisture anomalies in the core of the MJO than the poorly forecasted MJOs. The anomalous westerlies associated with the MJO at 850hPa are substantially weaker in the poor MJO forecasts than in those that predict the MJO well, though both tend to predict too strong anomalous easterlies to the east of the MJO convective center. Additionally, we will report on the effects that these environmental forecast errors have on the MJO-teleconnection patterns in the northern hemisphere, including extratropical cyclones, surface winds, and clouds in the Navy ESPC.