Detailed Analysis of ECMWF Surface Pressure Data

E. Fagiolini1, T. Schmidt1, G. Schwarz2, L. Zenner3

(1) GFZ Department 1, Potsdam, Germany(2) DLR IMF, Oberpfaffenhofen, Germany

(3) TUM IAPG, Munich, Germany

EGU 2012, Vienna, Austria

• Project Background

• Selected Surface Pressure Data

• Basic Data Characteristics

• Spatial Neighbourhoods

• Temporal Neighbourhoods

• Outlook

Summary of this Presentation

• Priority Programme ”Mass transport and Mass distribution in the System Earth” of the German Research Foundation (DFG).

• Project IDEAL-GRACE: ”Improved De-Aliasing for Gravity Field Modelling with GRACE”Partners: TUM IAPG, GFZ, DLR IMF, UHH IfM.

• An improved gravity field modelling requires computation of atmospheric contributions (mainly surface pressure).

• How good are available surface pressure data?

Project Background

Selected Surface Pressure Data

• We selected ECMWF surface pressure data from operational analysis,, global coverage, every 6 hours, N48 Gauss grid (192 columns, 96 lines), 29% land / 71% water.

• ECMWF surface pressure = combination of predictive modelling and measurements.

• Time period: 1 year from Sep. 1, 2007 to Aug. 31, 2008366 days, every 6 hours => 1464 data sets.

• Footprint per sample: about 200 * 200 km.

• A few outliers had to be removed by interpolation.

Median Values of 00, 06, 12 and 18 Hour Data

Small differences around Indonesia.

Europe Asia Pacific Ocean N. America Atlantic Ocean

Pressure Profile Along +45 deg. N

Histogram of Global 1 Year Surface Pressure Levels

Mountain areas

Contrast Enhanced Dynamic Pressure Range

Equatorial Regions are quiet

Diurnal Variation in the Tropics

Surface Pressure Time Series Data

Horizontal axis: transect along -30°S

Vertical axis: time (1 year, midnight data)

Principal motion of the pressure areas.

Sept.

Febr.

Aug.

Time Series Data

Horizontal axis: transect along -60°S

Vertical axis: time (1 year, midnight data)

Pressure data Low pass data High pass data

Horizontal axis: time (1 year, midnight data)

Vertical axis: transect along the zero meridian

Zero Meridian Transects versus Time (Contrast Enhanced)

North Pole

South Pole

Equator

Total Variation of 6 Hour Pressure Levels

Longitude-dependent total variation: interpolated time step effects?

Contrast Enhanced Entropy of Pressure Levels

The information content is not constant.

Histogram of East/West 1 Sample Differences

Back-to-back Laplacians visible small fluctuations and noise.

Histogram of North/South 1 Sample Differences

Broader the surface pressure differences are direction-dependent.

Sum of 2 Gaussians two overlapping effects.

Histogram of East/West 2 Sample Differences

Histogram of North/South 2 Sample Differences

Again, direction dependence.

Histogram of East/West 16 Sample Differences

Single Gaussian we see different effects.

Histogram of North/South 16 Sample Differences

the longitude dependence becomes smaller.

Pressure < 990 hPa above Water

The east coast shores of the continents show rather low pressure values. Why?

Differences Between 6 h Time Steps above Land

Single Gaussian very regular distributions above land.

Differences Between 6 h Time Steps above Water

Double peak distribution above water. Due to the microwave instruments?

Distribution of 6 h Biases above Water at 00h

Distribution of 6 h Biases above Water at 06h

The bias locations have changed. Is it a sensor effect, or is it physical reality?

Conclusion and Outlook

• ECMWF data describe a large variety of atmospheric effects in the spatial and temporal domain and at various scales.

• Outlier correction and removal of topographic background allows detailed studies of temporal phenomena (e.g., transects).

• Many open questions.

• A further topic to be discussed is the analysis of estimated errors provided by ECMWF (current activities).

Thank you!

Backslides

Contrast Enhanced Positive Skewness of Pressure Levels

Contrast Enhanced Negative Skewness of Pressure Levels