Near-real-time satellite-based bias corrections for ocean model forecastsJackie C. May and Clark RowleyNaval Research Laboratory, Code 7321, Stennis Space Center, MS

BackgroundAtmospheric heat fluxes at the ocean surface are used to forceoperational ocean models. The required heat fluxes include solarradiation, longwave radiation, sensible heat flux, and latent heatflux. These forcing fields come from atmospheric numericalweather prediction (NWP) models, which often have biases.Monthly or yearly bias correction can be identified and applied tothese fields prior to starting the model forecast. However, whenchanges occur in the atmospheric model, i.e. model versionupdates, new bias corrections must be calculated which is often atime intensive process. The NWP model of interest in this study isthe Navy Global Environmental Model (NAVGEM).

NFLUXAn alternative to NWP models is the Naval Research Laboratory(NRL) ocean surface flux system (NFLUX). NFLUX is a complete end-to-end data processing, automated quality control, and 2Dassimilation system. This system provides near-real-time satellite-based 3-hourly gridded analysis fields over the global ocean for thefollowing near-surface parameters:

NFLUX and NAVGEM Performance

MethodNAVGEM provides the surface forcing fields to several operationalocean model systems, including the Global Ocean Forecast System(GOFS). The satellite-based NFLUX analysis fields are used todetermine the NAVGEM bias corrections in near-real-time. NFLUXminus NAVGEM model differences provide bias corrections overthe hindcast period. The recent history (days to weeks) of thesedifferences are then used to determine long term and short termbias corrections, which are applied to the forcing over the forecastperiod.

Forecast Bias CorrectionsLong term (persistent) and a short term (weather-dependent) biascorrections are applied to the forecast period using uniqueaveraging windows and decorrelation times for each parameter.These elements are determined from the recent history of NFLUXminus NAVGEM daily average difference fields.

Bias Averaging WindowsThe averaging windows for the long term and short term biascorrections are determined from the NFLUX minus NAVGEM dailyaverage time series for one year. For each ocean point, the varianceof the n-day average difference time series is calculated, where nranges from 1 to 30 days. The long term averaging window isidentified when the calculated n-day variance becomes less thanhalf of the original variance calculated from the 1-day average.

• 10 days – specific humidity• 11 days – air temperature• 5 days – wind speed, solar radiation, longwave radiation

The short term bias correction is determined by subtracting thelong term bias from the NFLUX minus NAVGEM daily average timeseries.

Publications• May, J. C., C. Rowley, and N. Van de Voorde, 2016: The Naval Research Laboratory ocean surface flux (NFLUX) system:

Satellite-based turbulent heat flux products. J. Appl. Meteor. Climatol., 55, 1221-1237, doi:10.1175/JAMC-D-15-0187.1.• May, J. C., C. Rowley, and C. N. Barron, accepted: NFLUX satellite-based radiative heat fluxes Part I: Swath-level products. J.

Appl. Meteor. Climatol., accepted.• May, J. C., C. Rowley, and C. N. Barron, accepted: NFLUX satellite-based radiative heat fluxes Part 2: Gridded products. J.

Appl. Meteor. Climatol., accepted.

AcknowledgementsThis work was funded of the Office of Naval Research under the Naval Research Laboratory project Calibration of Ocean Forcing with Satellite Flux Estimates.

Bias Decorrelation Time ScalesThe decorrelation time scales for the long term and short term biascorrections are also determined from the NFLUX minus NAVGEM.

N Bias RMSE R2

NFLUX specific humidity (g kg-1) 128,086 0.33 1.25 0.96NAVGEM specific humidity (g kg-1) 128,086 -0.49 1.28 0.96

NFLUX air temperature (°C) 199,944 0.24 1.25 0.98NAVGEM air temperature (°C) 199,944 -0.30 1.25 0.98

NFLUX 10m wind speed (m s-2) 194,649 0.21 2.07 0.64NAVGEM 10m wind speed (m s-2) 194,649 -0.33 2.17 0.63

NFLUX solar radiation (W m-2) 34,495 12.56 140.80 0.89NAVGEM solar radiation (W m-2) 34,495 2.66 160.59 0.85

NFLUX longwave radiation (W m-2) 42,130 -2.94 23.00 0.87NAVGEM longwave radiation (W m-2) 42,130 -3.77 27.59 0.85

difference time series. Allocean points are averagedtogether to determine asingle global long term biascorrection averaging windowfor each parameter: .

• Specific humidity• Air temperature• 10-meter wind speed

• Solar radiation• Longwave radiation

daily average time series for one year. Foreach ocean point in the time series,decorrelation times are calculated using thelong term and short term biases from theprevious 30 days. The decorrelation times areaveraged together to produce a single globallyaveraged long term and short term temporaldecorrelation for each parameter. The longterm bias correction is applied throughout theentire forecast. The short term bias correctionis applied with a decreasing weighting. . .. .function from the analysis time out to:

• 2 days – specific humidity and air temperature• 1 day – wind speed, solar radiation, longwave radiation

Long Term vs Short Term Biases

638

NFLUX specific humidity daily average01 November 2013

NAVGEM specific humidity daily average01 November 2013

NFLUX minus NAVGEM specific humidity difference01 November 201325.0

22.520.017.515.012.510.07.55.02.50.0

25.022.520.017.515.012.510.07.55.02.50.0

3.02.41.81.20.60.0-0.6-1.2-1.8-2.4-3.0

Global averaged NFLUX minus NAVGEM specific humidity variance

Global averaged NFLUX minus NAVGEM longwave radiation variance

Global averaged NFLUX minus NAVGEM specific humidity temporal decorrelation

Global averaged NFLUX minus NAVGEM longwave radiation temporal decorrelation

25.020.015.010.05.00.0

-25.0-20.0-15.0-10.0-5.0

400

360

320

280

240

200

160

400

360

320

280

240

200

160

NFLUX minus NAVGEM longwave difference01 November 2013

NAVGEM longwave daily average01 November 2013

NFLUX longwave daily average01 November 2013

25.020.0

15.0

10.0

5.0

0.0

-25.0-20.0

-5.0

-10.0

-15.0

10.0

5.0

0.0

-25.0-20.0

-5.0

-10.0

-15.0

3.02.4

1.8

1.2

0.6

0.0-0.6

-1.2

-1.8

-2.4-3.0

3.02.41.8

1.2

0.6

0.0

-0.6

-1.2

-1.8

-2.4-3.0

3.02.4

1.8

1.2

0.6

0.0-0.6

-1.2

-1.8

-2.4-3.0

3.02.4

1.8

1.2

0.6

0.0-0.6

-1.2

-1.8

-2.4-3.0

25.020.0

15.0

10.0

5.0

0.0

-25.0

-20.0

-15.0

-5.0

-10.0

25.020.0

15.0

10.0

5.0

-25.0

-20.0

-15.0

-10.0

25.020.0

15.0

10.0

5.0

0.0

-25.0

-20.0

-5.0

-10.0

-15.0

25.020.0

15.0

10.0

5.0

0.0

-25.0

-20.0

-5.0

-10.0

-15.0

01 Ju

ne 2

013

01 N

ovem

ber 2

013

01 Ju

ne 2

013

01 N

ovem

ber 2

013

Specific humidity long term (10 days) bias correction

Specific humidity long term (10 days) bias correction Specific humidity short term bias correction

Specific humidity short term bias correction

Longwave short term bias correction

Longwave short term bias correctionLongwave long term (5 days) bias correction

Longwave long term (5 days) bias correction

AMS 97th Annual Meeting | Seattle, WA | 22 – 26 January 2017