Assessment of Medium-Range Model Forecasts over the CONUS during an Active MJO Phase

The National Centers for Environmental Prediction (NCEP) internship program focuses on enhancing the education for those seeking experience in becoming a meteorologist, forecaster, or model developer. A primary goal at NCEP’s Environmental Modeling Center (EMC) is to improve the accuracy and reliability of the Global Forecast System (GFS). Since the atmosphere is a multi-scaled system, it is important to evaluate how various phenomena, whether climate or weather, can impact the performance of a model over the duration of a forecast. Focusing on features that can easily influence the larger scale patterns, particularly those of a climate nature, and cross-examining the difference in performance between operational models is one way of isolating inherent or common problems that may exist in the model that underperforms. In this work we focus on forecasts of the Madden-Julian Oscillation (MJO) for both European Centre for Medium-range (ECM) and GFS models and how the forecast of the MJO impacts medium range forecasts of weather in the extratropics.

The MJO is an intraseasonal, eastward propagating climate pattern with a 30-90 day period along the equatorial zone, with the source of its energy located over the Indian and Western Pacific Oceans. It has both a strong convective and suppressed convective component-related to the ENSO phase-that manifests itself most clearly within the tropical region as anomalous rainfall, influencing weather and climate systems in North America and across the globe. It was reported by the Model Evaluation Group at EMC that the ECM and GFS model differed in both amplitude and phase of the MJO during December, January, and February (2017-2018). The objective of the project is to determine the performance of medium range forecasts for the ECM and GFS models, and whether the differences between the models can be linked to differences in how the MJO was represented.

In order to identify cases where the European and GFS models differed significantly, calculations of the Anomaly Correlation Coefficient (ACC) for the geopotential height field at 500-mb were examined. This required an analysis of data from the standard operational verification statistics database (VSDB) system between December, 2017 through February, 2018 to determine which days/forecast hours resulted in significant degradation of one model relative to the other. An error tracking approach using the difference in geopotential heights at 200-mb was then employed to determine the origin of the error for selected cases, where a more detailed analysis was conducted for days that clearly showed the origin of error over the Pacific region. Details of the analysis include an evaluation of how this error (and growth of this error) impacted the evolution of flow patterns in the extratropics, the transport of moisture from the equator towards the extratropics, and the impact of this transport on precipitation totals over the CONUS region between models. Further analyses include an evaluation of the vertical structure at selected latitudes near the equator, with an emphasis on circulation patterns intersecting the MJO and the meridional moisture fluxes exiting the MJO.