Simulation Experiments for GEO-CAPE Regional Air Quality

GEO-CAPE WorkshopSeptember 22, 2009Peter Zoogman, Daniel J. Jacob, Kelly Chance, Lin Zhang, Philippe Le Sager, Arlene Fiore

This work is supported by the NASA Atmospheric Composition Modeling and Analysis Program

Observing System Simulation Experiment

a priori = GEOS-Chem model

“Truth” = MOZART model

a posteriori = GEOS-Chem + assimilation of synthetic observations attempt to reproduce true atmosphere

Developed OSSE capability for regional scale problems

First application is to determine GEO-CAPE measurement requirements for surface ozone

Simulation ModelsSevere test due to opposite biases at 700 hPa and surface

MOZART (1.9o x1.9o , NCEP) GEOS-Chem (1o x1o , GEOS-3) “Truth” a priori

JULY 2001

700 hPa

Below 1 km

ppbv

MethodsSynthetic Observations – apply averaging kernel to “truth”Remove cloudy scenes

Data Assimilation – Kalman FilterIncludes spatial covariance of 1 km in vertical, 300 km in horizontal (will be informed by sensitivity team)

Error is evolved and transported as a passive tracer

Model Error computed by comparison to ozonesondeGEOS-Chem error found to be 25%

Instrument smoothing error assumed to be diagonal, 10%

TES O3 AK North American 1o x1o

Nested Grid

Results at 700 hPaGEOS-Chem mean absolute bias: 7.4 + 4.1 ppbv

GEOS-Chem + Hourly Obs: 2.7 + 2.0 ppbv

Hourly observations (daytime clear-sky only) provide successful correction, much better than daily observations

Mean July 2001 Ozone

GEOS-Chem + Daily Obs: 5.6 + 3.5 ppbv

ppbv

Results below 1 kmGEOS-Chem mean absolute bias: 4.3 + 3.3 ppbv

Mean July 2001 Ozone

GEOS-Chem + Hourly Obs: 3.4 + 2.7 ppbv

GEOS-Chem + Daily Obs: 4.1 + 3.0 ppbv

TES sensitivity to boundary layer is insufficient to correct model significantly at the surface --need to do better to inform ozone surface air quality

ppbv

Sensitivity by Adjoint Adjoint model – receptor based rather than source based approach

Sensitivity of surface ozone to ozone produced at each vertical layer – tool to understand OSSE results

Surface ozone primarily sensitive to production below 2 km; this supports simulation results

Accumulated surface O3 sensitivity for Boston

Summary and Future WorkTIR measurements not sufficient for boundary layer ozoneMultispectral is needed!

Developed a framework for testing prospective instrumentation – can be applied as new sensor configurations are proposedFrequency of observations, observation domain, nighttime observations

GEOS-Chem resolution will increase – 10 km by end 2010

Use CMAQ ozone for higher resolution/different “truth”

Integration of surface sitesPerform forecasting OSSEsExpand to combine observations of other species

NOx, CO, HCHORequired frequency may be different for different species

Define requirements for climate forcing and intercontinental transport