Jump to content
Sign in to follow this  
Blessed Weather

Seasonal Forecasting – An Overview

Recommended Posts

This article contains an overview of Seasonal Forecasting and its challenges with a focus on the UK Met Office seasonal model GloSea and its sub-models.

First some background: The Met Office has a range of models, each with a particular focus and ranging from daily weather forecasts to climate forecasts for the next century. Since 1990 their strategy has been to integrate the individual models into a Unified Model system that can be used for prediction across a range of timescales. This from their website:

“The Unified Model applies this seamless modelling approach, which means that the same dynamical core and, where possible, the same parameterization schemes are used across a broad range of spatial and temporal scales on a traceable frame work. The model is suitable for numerical weather prediction (NWP), seasonal forecasting and climate modelling with forecast times ranging from a few days to hundreds of years. Furthermore, the Unified Model can be used both as a global and a regional model.”

The Met Office’s Seasonal model, GloSea, sits within the Unified Model structure and is described as the main prediction system currently used for seasonal timescales (up to 6 months) and sub-seasonal timescales (between 2 to 6 weeks). It is an ensemble prediction system using a coupled ocean-atmosphere model (a variant of Met Office climate prediction model: HadGEM3 family), to generate probabilistic forecasts up to six months ahead.

The model family is shown in this diagram: 191643901_MetOffUnifiedModelDiagram.thumb.jpg.51d8857c25e4c920f751340845e8300b.jpg

The latest iteration - GloSea5 - runs daily with 28 ensemble members for sub-seasonal forecasts and weekly with 42 members for seasonal forecasts. It has two components: the real-time forecast and a companion set of hindcasts, also called historical reforecasts, used for post-processing (bias correction and calibration) and skill assessment. In the forecast suite, initial states (start conditions) for the atmosphere, land surface and ocean are calculated daily.

The atmosphere and land surface resolution (grid spacing) is 1.875° × 1.25° to 0.833° × 0.556°. The grid spacing in the ocean and sea‐ice models is 0.25°.

System initialisation includes the following data:

  • atmosphere and land surface components from the Met Office operational numerical weather prediction (NWP) model;
  • fully resolved stratosphere as part of 38 vertical levels of the atmosphere;
  • ocean surface temps (SST), ocean vertical temperature and salinity profiles, ocean currents, AMOC strength;
  • atmosphere–ocean coupling;
  • sea ice cover and thickness, ice-berg calving;
  • solar forcing;
  • climate forcings (e.g., aerosols, methane, CO2, Ozone, etc.).

Specific teleconnections and other key components modeled include:

  • ESNO (El Nino Southern Oscillation)
  • MJO (Madden-Julian Oscillation)
  • QBO (Quasi-Biennial Oscillation)
  • NAO (North Atlantic Oscillation)
  • AO (Arctic Oscillation)
  • WNPSH (Western North Pacific Subtropical High)
  • Tropical Storms (tropical storm number, spatial distribution, accumulated cyclone energy index, and landfall frequency)
  • Global 1.5m Temps
  • Global Precipitation
  • Global SSTs (sea surface temps)
  • Total cloud amount
  • Pressure at mean sea level
  • Insolation (solar radiation)
  • Outgoing longwave and shortwave radiation
  • Clear-sky outgoing shortwave and longwave radiation
  • Shortwave and longwave cloud forcing
  • Geopotential height at 200, 500, 850 hPa
  • Temperature at 200, 500, 850 hPa
  • Zonal wind at 200, 500, 850 hPa
  • Meridional wind at 200, 500, 850 hPa
  • Relative humidity at 200, 500, 850 hPa
  • Specific humidity at 200, 500, 850 hPa

The latest forecast accuracy of GloSea5 forecasts versus actual observations (measured as the correlation coefficient) for key teleconnections are:

  • ESNO correlation of 0.80
  • MJO correlation of approx. 0.78 for a 5-day forecast and 0.70 for 10-day.
  • NAO correlation of 0.62
  • AO correlation of 0.63
  • WNPSH correlation of 0.80

Below are GloSea5 overall skill maps for Precipitation, SSTs and 1.5m Temps. Values above 0.5 denote useful skill compared to climatology.

1799565939_GloSea5SkillMapPrecipitation.thumb.jpg.b42d5c1188e8a3dd2c079fde8338b97b.jpg 1148801278_GloSea5SkillMapSSTs.thumb.jpg.82a9d7aed1e9218feada373ebdb94309.jpg1764512378_GloSea5SkillMap1.5mtemps.thumb.jpg.eab20b1201ac1d6696ed0a0407bb7d7b.jpg

Whilst forecast accuracy continues to improve, seasonal forecasting still presents many challenges, particularly:

  • the very complex interaction of teleconnections,
  • the unfolding impact of rapid Arctic Amplification (warming) on previous understanding and teleconnections impact;
  • the challenge of getting initialisation data for the entire globe absolutely correct so that the “butterfly effect” of a small error in starting data doesn’t rapidly escalate to forecast errors.

Further reading on the challenges facing seasonal forecasting can be found in the following short paper from the Imperial College London titled: The challenge of seasonal weather prediction. https://wwwf.imperial.ac.uk/blog/climate-at-imperial/2014/04/14/the-challenge-of-seasonal-weather-prediction/

Despite these challenges, good progress is being made, with the performance of GloSea5 notably improved from its predecessor GloSea4.

Sources of information and further reading:

Met Office Links:

Share this post


Link to post
Share on other sites
15 hours ago, Blessed Weather said:

This article contains an overview of Seasonal Forecasting and its challenges with a focus on the UK Met Office seasonal model GloSea and its sub-models.

First some background: The Met Office has a range of models, each with a particular focus and ranging from daily weather forecasts to climate forecasts for the next century. Since 1990 their strategy has been to integrate the individual models into a Unified Model system that can be used for prediction across a range of timescales. This from their website:

“The Unified Model applies this seamless modelling approach, which means that the same dynamical core and, where possible, the same parameterization schemes are used across a broad range of spatial and temporal scales on a traceable frame work. The model is suitable for numerical weather prediction (NWP), seasonal forecasting and climate modelling with forecast times ranging from a few days to hundreds of years. Furthermore, the Unified Model can be used both as a global and a regional model.”

The Met Office’s Seasonal model, GloSea, sits within the Unified Model structure and is described as the main prediction system currently used for seasonal timescales (up to 6 months) and sub-seasonal timescales (between 2 to 6 weeks). It is an ensemble prediction system using a coupled ocean-atmosphere model (a variant of Met Office climate prediction model: HadGEM3 family), to generate probabilistic forecasts up to six months ahead.

The model family is shown in this diagram: 191643901_MetOffUnifiedModelDiagram.thumb.jpg.51d8857c25e4c920f751340845e8300b.jpg

The latest iteration - GloSea5 - runs daily with 28 ensemble members for sub-seasonal forecasts and weekly with 42 members for seasonal forecasts. It has two components: the real-time forecast and a companion set of hindcasts, also called historical reforecasts, used for post-processing (bias correction and calibration) and skill assessment. In the forecast suite, initial states (start conditions) for the atmosphere, land surface and ocean are calculated daily.

The atmosphere and land surface resolution (grid spacing) is 1.875° × 1.25° to 0.833° × 0.556°. The grid spacing in the ocean and sea‐ice models is 0.25°.

System initialisation includes the following data:

  • atmosphere and land surface components from the Met Office operational numerical weather prediction (NWP) model;
  • fully resolved stratosphere as part of 38 vertical levels of the atmosphere;
  • ocean surface temps (SST), ocean vertical temperature and salinity profiles, ocean currents, AMOC strength;
  • atmosphere–ocean coupling;
  • sea ice cover and thickness, ice-berg calving;
  • solar forcing;
  • climate forcings (e.g., aerosols, methane, CO2, Ozone, etc.).

Specific teleconnections and other key components modeled include:

  • ESNO (El Nino Southern Oscillation)
  • MJO (Madden-Julian Oscillation)
  • QBO (Quasi-Biennial Oscillation)
  • NAO (North Atlantic Oscillation)
  • AO (Arctic Oscillation)
  • WNPSH (Western North Pacific Subtropical High)
  • Tropical Storms (tropical storm number, spatial distribution, accumulated cyclone energy index, and landfall frequency)
  • Global 1.5m Temps
  • Global Precipitation
  • Global SSTs (sea surface temps)
  • Total cloud amount
  • Pressure at mean sea level
  • Insolation (solar radiation)
  • Outgoing longwave and shortwave radiation
  • Clear-sky outgoing shortwave and longwave radiation
  • Shortwave and longwave cloud forcing
  • Geopotential height at 200, 500, 850 hPa
  • Temperature at 200, 500, 850 hPa
  • Zonal wind at 200, 500, 850 hPa
  • Meridional wind at 200, 500, 850 hPa
  • Relative humidity at 200, 500, 850 hPa
  • Specific humidity at 200, 500, 850 hPa

The latest forecast accuracy of GloSea5 forecasts versus actual observations (measured as the correlation coefficient) for key teleconnections are:

  • ESNO correlation of 0.80
  • MJO correlation of approx. 0.78 for a 5-day forecast and 0.70 for 10-day.
  • NAO correlation of 0.62
  • AO correlation of 0.63
  • WNPSH correlation of 0.80

Below are GloSea5 overall skill maps for Precipitation, SSTs and 1.5m Temps. Values above 0.5 denote useful skill compared to climatology.

1799565939_GloSea5SkillMapPrecipitation.thumb.jpg.b42d5c1188e8a3dd2c079fde8338b97b.jpg 1148801278_GloSea5SkillMapSSTs.thumb.jpg.82a9d7aed1e9218feada373ebdb94309.jpg1764512378_GloSea5SkillMap1.5mtemps.thumb.jpg.eab20b1201ac1d6696ed0a0407bb7d7b.jpg

Whilst forecast accuracy continues to improve, seasonal forecasting still presents many challenges, particularly:

  • the very complex interaction of teleconnections,
  • the unfolding impact of rapid Arctic Amplification (warming) on previous understanding and teleconnections impact;
  • the challenge of getting initialisation data for the entire globe absolutely correct so that the “butterfly effect” of a small error in starting data doesn’t rapidly escalate to forecast errors.

Further reading on the challenges facing seasonal forecasting can be found in the following short paper from the Imperial College London titled: The challenge of seasonal weather prediction. https://wwwf.imperial.ac.uk/blog/climate-at-imperial/2014/04/14/the-challenge-of-seasonal-weather-prediction/

Despite these challenges, good progress is being made, with the performance of GloSea5 notably improved from its predecessor GloSea4.

Sources of information and further reading:

Met Office Links:

Thanks for that BW. The data shown there is a very big upgrade from 10 years ago and even the last 3 or 4 years. Some of the teleconnections were ignored or not though worthy of use even less than 10 years ago by the Met Office.

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Sign in to follow this  

×
×
  • Create New...