Meteorological
and Climate
Modeling
Seasonal Weather Forecasts: long-term predictions

Regional Dynamical Downscaling from the global seasonal forecast systems

The NCEP Climate Forecast System (CFS) (http://cfs.ncep.noaa.gov/) provides operational seasonal forecast data with daily and monthly output for up to nine months.

The Model: There are four members per day from initial conditions for the atmosphere and ocean, which are 1 day old. The atmospheric initial conditions are obtained from NCEP Reanalysis-2 and the ocean initial condition is obtained from NCEP GODAS (Global Ocean Data Assimilation). The integrations are complete for the first partial month + 9 full months into the future.

The raw daily CFS output forecast data is available on a local NCEP ftp server: ftp://ftpprd.ncep.noaa.gov/pub/data1/nccf/com/cfs/prod/forecast.yyyymmdd/daily_grib_memb documentation at: http://cfs.ncep.noaa.gov/cfs_data.pdf

It is highly recommended to calibrate (bias correction) the operational forecasts (using the retrospective data) before they are used in any predictive environment. For this purpose are available forecast and observed daily climatologies for selected variables, for both the mean and standard deviation.


    East Asia, in the center the Korean peninsula,
    110°E-140°E, 20°N-55°N)
    May 2009 - November 2009

Atmospheric variables for which daily forecast climatologies have been calculated:

    Surface fields:
    • temperature,
    • pressure,
    • zonal and meridional component of momentum flux,
    • precipitation rate,
    • precipitable water,
    • sensibel heat flux,
    • latent heat flux,
    • upward long wave radiation flux,
    • upward short wave radiation flux,
    • upward long wave radiation flux,
    • downward long wave radiation flux,
    • upward short wave radiation flux,
    • downward short wave radiation flux
    Vertical pressure level fields
    at 1000, 850, 700, 500, 200 hPa:
    • geopotential height,
    • streamfunction,
    • velocity potential,
    • zonal velocity,
    • meridional velocity

    The daily data are available at 00:00 and 12:00 model time. A detailed description of the variables can be found at: http://cfs.ncep.noaa.gov/cfs.daily.climatology.doc

ECMWF ENSEMBLES seasonal-to-decadal experiments
(http://www.ecmwf.int/research/EU_projects/ENSEMBLES/data/index.html)

The Model: The system consists of an ocean analysis to estimate the initial state of the ocean, a global coupled ocean-atmosphere general circulation model to calculate the evolution of the ocean and atmosphere, and a post-processing suite to create forecast products from the raw numerical output. The ocean model used is based on HOPE (Hamburg Ocean Primitive Equation model) version 2 (Latif et al. 1994, Wolff et al. 1997). The atmospheric component of the coupled model is the ECMWF IFS (Integrated Forecast System) model version 31r1.

A set of daily and monthly variables is available. The daily variables are either instantaneous at 00 GMT or accumulated daily over 24 hours starting at 00 GMT. Maximum and minimum temperature at 2 meters are computed using four values at 6, 12, 18 and 24 GMT.

    Surface fields:
    • surface temperature (temperature of the first soil layer over land, sea surface temperature over open waters and ice temperature over ice),
    • snow depth,
    • surface sensible heat flux,
    • surface latent heat flux,
    • mean sea level pressure,
    • total cloud cover,
    • zonal component of 10m wind,
    • meridional component of 10m wind,
    • 2m temperature,
    • 2m dewpoint temperature,
    • surface downward solar radiation,
    • surface downward longwave radiation,
    • surface net solar radiation,
    • surface net longwave radiation,
    • top net solar radiation,
    • top net longwave radiation,
    • moisture flux from the surface into the atmosphere or evaporation,
    • 2m Tmax (computed from 6-hourly data),
    • 2m Tmin (computed from 6-hourly data),
    • total precipitation,
    • vertically integrated volumetric soil water
    Vertical pressure level fields
    at 850, 500, 200 and 50 hPa:
    • geopotential,
    • temperature,
    • zonal wind,
    • meridional wind,
    • specific humidity

Dynamical downscaling

Downscaling of global coarse resolution model output with regional dynamical models have shown to provide more skillful forecasts than statistical downscaling especially for surface conditions and patterns of precipitation anomalies. However, there are intrinsic problems to achieve reasonable good predictions with the regional models using forcing from the global models:

  1. Inaccuracy of the long-term weather predictions:   It must be remembered, that there are tight limits on what it is physically possible to achieve with a seasonal forecast system. It is only possible to predict a range of likely outcomes. Some seasonal forecasts available today are issued with probabilities (or error bars) which have been properly calibrated against past cases. A proper calibration of a forecast system against data is not always easy to do. This is primarily because of the limited availability of past data. The problem is especially severe when the level of predictability is low so that many years of data are needed.

    Strategy:

    • calibrating direct model output (raw forecast) with the climatologies whenever available
    • using ensemble predictions

  2. Interpolation from coarse to fine resolution grid, limited number of global model output suitable for downscaling:   Operational seasonal forecast systems provide daily output of only 5 levels of height and 2 times a day. In comparison, short-term forecast data are available at minimum 17 standard levels with a more complete number of surface variables and 3 or 6 hourly output. It has been shown (Kanamaru and Kanamitsu, 2007) that a fairly large number of levels, more than 9 levels, are required to produce reasonable downscaling.

    Strategy:

    • there is a reasonable alternative to downscaling global simulations with a relatively small number of outputs. One way to extract the best information from global outputs is using the incremental interpolation method (http://g-rsm.wikispaces.com/file/view/Increment_rev7.pdf) to perform dynamical downscaling.


© Copyright 1995-2016 by:   ESS   Environmental Software and Services GmbH AUSTRIA | print page