Estimating uncertainty of reanalyses using energy budget diagnostics Michael Mayer Michael Mayer - April 2012

Estimating uncertainty of reanalyses using energy budget diagnostics

Michael Mayer

Michael Mayer - April 2012

Aim of this work


Evaluation of climatology, anomalies and trends of the atmospheric energy budget 1979-2011

Data:Reanalyses: ERA-Interim, MERRA, CFSRAdditionally: CERES, ISCCP, GPCP

Changes in the budgets:True climate signal?Changes in the observing system?

Divergence of horizontal energy transport

• Calculation of the divergence from u, v, q, T

• Analysed fields with 6-hourly resolution

• Problems with mass consistency


Direct methodIndirect method

Calculation of the divergence as a residualUse of 12-hourly accumulated vertical fluxesEnergy tendency from 12-hourly forecastsNo problems with time resolutionMass inconsistencies as well

Mass budget in ERA-InterimSystematically deficient

surface pressure forecasts (Up to -600hPa/year!)Instantaneous mass flux

divergence also erroneousMass correction is

necessary!

Climatology at TOA


Too little energy input in the Tropics

Too weak meridional RadTOA gradient

Mean poleward energy transport


Much too weak total poleward energy transport (atmosphere + ocean)

Atmospheric transport in better agreement with reference estimate

Variability: Zonal mean budget (20N/20S)


Tropical energy export is varying very weaklyDirect/Indirect method show different anomalies


Radiative fluxes quite stable in timeTrends in latent heat flux are compensated by trends

in energy tendency forecasts

Fluxes at TOA and at the surface

Analysis increments


Defined as the difference between the analyzed and the forecasted state of the model, using 12-hourly forecasts

Total energy tendency analysis increments:

Increments can change in time due toChanges in the observation systemImperfectly forecasted changes of the

atmospheric circulation

Analysis increments in the Tropics


Strong shifts in analysis incrementsPrecipitation and moist analysis increments highly

correlated


Strongest shifts related to SSM/I in 1992 and 2008

Apparently relatively weak anomalies related to climate anomalies

On overall, no improvement over time

Zonal mean latent heat analysis increments


Zonal structure of tropical moisture flux divergence


Use different climatologies to get rid of inhomogeneities

Simple algorithm reducing RMS-value of analysis increments

ENSO-related anomalies

Zonal structure of tropical latent heat analysis increments


2nd iteration: remove either 2008 shift or stripes

Further reduce inhomogeneities


Result: fairly homogeneous anomaly timeseries

Trends in zonal mean transport


IPCC AR4 models predict increase in total poleward energy transport + stronger Hadley cell

Indirect ERA-I estimate shows opposite trends

Held and Soden (2006):Indirect estimate

ΔF

[P

W]


Direct estimate slightly more stable in timeBoth estimates show unrealistically strong trendsReliable trend estimation of energy transports not

possible

Direct estimateHeld and Soden (2006):

ΔF

[P

W]

Conclusions


Reanalyses have too weak mean meridional RadTOA

gradient and thus too weak poleward total energy transports

Weak variation in zonal mean budgets, true climate anomalies hard to identify

Strong increase of evaporation over time -> compensated by analysis increments

For investigation of interannual variability split of climatologies is helpful

30-year trends in zonal mean poleward transport are too strong and in disagreement with climate models





Indirect method

t

eFRadF

fc

STOAindA

*

*2.

Correction methods I

Correction methods II


Direct method

Mass budget in ERA-Interim II


Time-averaged mass flux divergence unrealisticIndirect estimation from mass tendency and E-P

much better


Mass budget in ERA-Interim

Up to -600hPa/year!Forecast error:

ans

fcss

t

p

t

p

t

p

Systematically deficient surface pressure forecasts ()

Results so far – Direct method


Many important climatological features, but noisy

Indirect method


Much clearer and less noisy

Poleward energy transports– different results


FT08…Fasullo und Trenberth (2008)

Mass correction reduces cross-equatorial energy transports

Interannual variations of the transports

• Which anomalies are true climate signals (e.g. due to ENSO)?

• Which anomalies are artefacts? Michael Mayer - April 2012


• Changes in the observing system affect analysis increments

D. Dee(2009)

COSMIC

SSM/I

AMSU-A

AIRS

Next steps • Understanding the different lags of moist and dry

anomalies


Negative lag of total export anomalies

~Zero lag (moist/dry/total)

Positive Lag of total export anomaly

Next steps• Fast propagating anomalies (continous

blue) in agreement with Trenberth et al. 2002

• We find a much larger pattern (dashed blue)

• Slowly propagating anomalies (red) linked to ENSO?-> Reproducibility with other datasets?-> Objective method for determining phase speed of the transient features required (e.g. wavelets)-> propagating SLP anomalies in Peterson and White (1998) – connection? Michael Mayer - April 2012

Next steps

• Reliable ‚independent‘ datasets are required

• ISCCP is being reprocessed

• More homogeneous datasets expected


Next steps

• Prolongation of the ERA-Interim-datasets back to 1979 is being produced– extension of the study period– Long-term transport trend estimates become possible– Try to verify expected increase of poleward energy transports

(Held and Soden, 2006)

• Employment of a GCM to– Study varying partition between moist and dry anomalies

– Study effects of El Niño on energy transports without data inhomogeneities

– verify features found in the reanalysis data


Summary – Scientific goals

• A new estimate of the poleward atmospheric energy transport

• A better understanding of the interannual variability of the energy transports

• Understanding of the mechanisms leading to the observed eastward moving anomaly patterns

• Feedback regarding data homogeneity and consistency to the producers of the used datasets


Work plan and financing• 1st year: Visit CAS at NCAR (02/2011-07/2011)

-> access to new datasets (GCM output, reprocessed satellite data, ocean data) and expertise of the group

• 2nd year: access to extended ERA-Interim dataset and investigation of long-term trends

• 3rd year: Employment of GCM results and investigation of the described questions

• Thesis work is funded within FWF project “Global historic in situ upper air data for climate change research” (running at least until 06/2012)

• Stay at NCAR mainly funded by ASP


Thank you for your attention

Literature


Documents

Estimating uncertainty of reanalyses using energy budget diagnostics Michael Mayer Michael Mayer - April 2012