31
Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic Research Institute, St. Petersburg Tel.: (812) 352 03 07; E-mail:svkl @ aari . nw . ru The fourth meeting of the International Ice Charting Working Group The fourth meeting of the International Ice Charting Working Group (IICWG (IICWG ) ) St.Petersburg, Russian Federation, April 7-11, 2003 St.Petersburg, Russian Federation, April 7-11, 2003

Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Embed Size (px)

Citation preview

Page 1: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Experience of short-range (1-5 days) numerical ice

forecasts for the freezing seas.

Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko

Arctic and Antarctic Research Institute, St. Petersburg

Tel.: (812) 352 03 07;

E-mail:[email protected]

The fourth meeting of the International Ice Charting Working Group The fourth meeting of the International Ice Charting Working Group (IICWG(IICWG) )

St.Petersburg, Russian Federation, April 7-11, 2003St.Petersburg, Russian Federation, April 7-11, 2003

Page 2: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Regions General view of model regions

Page 3: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Barents and Kara Seas

Page 4: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Bathymetry

Grid mesh of the model

(cell dimension is

25×25 km)

Page 5: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Initial data

Page 6: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Principle of GRID data interpolation

(atmospheric pressure distribution)

Page 7: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Interpolated atmospheric pressure and

calculated surface wind (forecast)

Page 8: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Interpolated air temperature (forecast)

Page 9: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Satellite images often do not cover the entire model area.

In this case the initial data for new forecast are prepared

by composing the new image data and results of previous forecast.

Page 10: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Actual distribution on the new image…

Page 11: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Previous forecast…

Page 12: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Composite ice chart used as initial data for new forecast

Page 13: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Sea temperature is prepared by correcting the climatic distribution in accordance with actual location of ice edge

Page 14: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

The model consists of four principal components:

1) thermal evolution of the sea water (based on the equations

of heat and salinity budget);

2) sea water dynamics (based on the equations of hydrodynamics);

3) thermal evolution of ice cover (based on the heat budget equation);

4) ice cover dynamics (based on the non-stationary equations of ice dynamics with viscous-plastic rheology).

MODEL

Page 15: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Results

Page 16: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Forecast of ice drift and weighted-mean thickness

Page 17: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Forecasted and actual distribution

Page 18: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Examples for other regions Pechora Sea

Page 19: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Laptev Sea

Page 20: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

East-Siberian Sea

Page 21: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

1) Skill score of the model forecast P;

Ncorrect – number of cells in which forecasted and actual values are close (difference is not more than permissible error);

Ntotal – total number of cells2) Skill score of the inertial forecast I;

Ninert – number of cells in which the initial and final actual values are close (difference is not more than permissible error);

Ntotal – total number of cells3) Efficiency E

E = P - I

Criteria of quality

total

inert

NN

I

Page 22: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

The sense of these formulas is as follows:

The model forecast affirms: “Ice conditions will change in accordance with the model results”.

The inertial forecast affirms: “The changes of ice conditions will not be significant, and we may accept them constant.”

The forecast efficiency shows: “which of these two hypotheses is closer to reality”.

If:

1) efficiency is positive (the model forecast has higher skill score than the inertial forecast): the changes of ice conditions are significant, hence, we may not accept them constant and it is more reasonable to employ the model forecast;

2) efficiency is negative or zero (the model forecast has lower or equal skill score than the inertial forecast): it is more reasonable to assume the ice conditions constant than to employ the model forecast.

Page 23: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

PE 68.0max

Typical formula for maximum permissible error:

where Emax – maximum permissible error, - standard error, P - natural variability of forecasted parameter for the temporal scale equal to prognostic period

As for ice concentration, maximum permissible errors coincide with standard concentration gradations defined in the “International Symbols for Sea Ice Charts and Sea Ice Nomenclature” as follows:open water – 0 tenths (0%); very open ice – 1-3 tenths (less than 35 %);open ice – 4-6 tenths (36-65 %);close ice – 7-8 tenths (66-85 %);very close ice – 9-10 tenths (more than 85%)

Page 24: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Algebraic error

Absolute error

Skill score, %

Efficiency, %

Ice concentration, tenths

0.03 1.75 85.6 5.0

Ice drift velocity, cm/s

2 5 79.9 26.0

Ice drift direction, degrees

-13 35 84.0 48.4

Ice thickness, cm 1 8 88.9

Equivalent thickness of ridges, cm

-3 3 85.9

Ice pressure, points (3 point scale)

0.08 0.19 7.8

Verification Generalized results of ice forecasts

Page 25: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Statistical distribution

of ice concentration

forecast errors

Algebraic errors

Absolute errors

Page 26: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Interface General view of interface panel

Page 27: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Entering the forecast parameters

Page 28: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Entering initial data

Page 29: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Running

Page 30: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Results demonstration

Page 31: Experience of short-range (1-5 days) numerical ice forecasts for the freezing seas. Sergey Klyachkin, Zalman Gudkovich, Roman Guzenko Arctic and Antarctic

Conclusions

The main directions of development.•Improvement of methodology of initial ice chart composing;

•Improvement of methodology of initial water temperature correction;

•More detailed simulations of the sea currents (including tides);

•More accurate estimate of horizontal heat fluxes in the near-edge zones;

•Elaboration of fast ice boundary forecasting.