David H. Bromwich 1 Ola Persson 2 , and John J. Cassano 3 1 Polar Meteorology Group

1. Atmosphere Observational Needs for Model Validation David H. Bromwich1 Ola Persson2, and John J. Cassano3 1Polar Meteorology Group Byrd Polar Research Center The Ohio State University Columbus, Ohio, USA 2National Oceanic and Atmospheric Administration Earth System Research Laboratory Cooperative Institute for Research in Environmental Sciences 3University of Colorado Cooperative Institute for Research in Environmental Sciences Department of Atmospheric and Oceanic Sciences

2. Just What Is “Arctic”? We Select • Arctic Ocean • Greenland Ice Sheet …and … • The rivers which empty into the Arctic Ocean and maintain its near surface stratification Thus our Arctic extends from ~45°N to the North Pole

3. Selected Weather and Climate Features and their Resolvability with Current Arctic Observations • Climate Modes • e.g., Arctic Oscillation/NAO, PNA • Yes, we can resolve these large-scale features with current observations • Weather • e.g., Synoptic-scale cyclones • Reasonably, but some shortcomings – see next slide • Mesoscale Phenomena • Polar lows, sea breezes, barrier winds, katabatic winds • Insufficiently resolved • Sea Ice • Extent, fractional coverage, thickness, albedo, snow cover, melt ponds • Yes Yes No ? ? ?? • Land • Snow cover, SWE, permafrost, vegetation, lakes • Yes ? ? Yes ?

4. The storm of 19 October 2004 as depicted by the NCEP/NCAR global reanalysis. Contours represent isobars of sea level pressure at increments of 3 hPa. [from visualization package of NOAA Climate Diagnostics Center] The Figure shows an intense storm depicted in the NCEP/NCAR reanalysis for 19 October 2004. This storm, which led to flooding of downtown Nome, Alaska, has a central pressure of 949 hPa in the reanalysis. The actual central pressure deduced by the National Weather Service was as low as 941 hPa.

5. 3 Critical Components for an Integrated Arctic Observing System • Remote Sensing Observations • only way to obtain comprehensive regional coverage • Numerical Modeling • Fills gaps in the system • Maintains physical consistency in the system • In-situ Observations • Provides the ground truth to calibrate the system

6. Typical distribution of COSMIC GPS radio occultation soundings (green dots) over a 24-hour period over the Arctic.

7. Greenland Climate Network (GC-Net) HUM TUN GIT NGP NAE NAU SUM CP1 KAR CP2 JAR1,2,3,SWC NSE DY2 SDL SDM Steffen and Box (2001), JGR

8. IASOA Observatories Data of interest to the IASOA consortium include measurements of standard meteorology, greenhouse gases, atmospheric radiation, clouds, pollutants, chemistry, aerosols, and surface energy balances. 

9. Tara Ice Station Tara Arctic 2007-2008 is a specific project for IPY. The boat will undertake a Nansen-like crossing of the Arctic Ocean, drifting from the north of Siberia to the Fram Straight during almost two year trapped in the ice. The Tara Arctic ice station will provide permanent facilities for science fieldwork, in-situ observations and maintenance possibilities of probes and automated buoys.

10. Arctic System Reanalysis (ASR)an NSF-Funded IPY Project • Rapid climate change appears to be happening in the Arctic. A more comprehensive picture of the coupled atmosphere/land surface/ ocean interactions is needed. • 2. Global reanalyses encounter many problems at high latitudes. The ASR would use the best available description for Arctic processes and would enhance the existing database of Arctic observations. The ASR will be produced at improved temporal resolution and much higher spatial resolution. • 3. The ASR would provide fields for which direct observation are sparse or problematic (precipitation, radiation, cloud, ...) at higher resolution than from existing reanalyses. • 4. The system-oriented approachwould provide a community focus including the atmosphere, land surface and sea ice communities. • 5. The ASR would provide aconvenient synthesis of Arctic field programs (SHEBA, LAII/ATLAS, ARM, ...)

11. Optimizing the Arctic Observing Network Using the ASR Framework • Observing System Experiments (OSEs) are numerical model-based experiments to test the impact of existing observations. Sometimes called “data denial” experiments. • Observing System Simulation Experiments (OSSEs) are numerical experiments that test impacts of future observing systems, e.g., new satellite sensors and AWS. • Determine the observations needed to optimize the observing system.

12. Key Points for SAON • New Data Sources • Weather and Climate Applications • Combining Remote Sensing, Modeling and In-situ Observations • Data Assimilation • Bringing Observations, Modeling and Data Users together • Arctic System Reanalysis as a Synthesis • Better Integrated Use of Resources • User Friendly Data Handling

13. Atmospheric Model Evaluation • Evaluate over a variety of polar surface types • Ice sheet • Sea ice / ocean • Non-ice covered land • Evaluate atmospheric state • Temperature, pressure, winds, humidity,… • Evaluate atmospheric processes and relationships • Surface energy budget • Cloud processes • Are we getting the right answer for the right reasons? • Do the relationships occurring in the data also occur in the models?

14. Example: ARCMIP Evaluation Comparison to ERA40 • Need to evaluate models on several scales • - At largest scales need to compare model to reanalyses • - At smaller scales can compare model to point observations, although care is needed Comparison with SHEBA surface observations

15. Comparison to SHEBA sensible heat flux Comparison to SHEBA latent heat flux - Large accumulated errors in almost all models are of concern in coupled simulations

16. ARCMIP comparisons of sensible heat flux relationship - Only one model is able to decrease the magnitude of the Hs for very stable conditions as in the observations

17. Analysis of relationship between variables: SWD and CWP It is important to not only evaluate the model state but to evaluate if the model reproduces observed relationships between variables

18. Conclusions • Care needs to be taken when evaluating variables that are the result of many interacting, complex processes • It is useful to evaluate the different processes that are responsible for the final state • Evaluation of processes and relationships between variables can provide additional insight into model performance • Useful to highlight aspects of the model that need improvement