FSL-RUC Experiment and analysis design for TAMDAR

Design issues having to do with our parallel runs of the dev and dev2 20km RUC models. (These are identical except that dev2 includes TAMDAR.)

• Observation error for TAMDAR RH is currently set to near value for raob RH. Never before has there been aircraft moisture obs to consider, so this setting is important. Obs errors for temperature and winds from TAMDAR are set the same as that for non-TAMDAR aircraft (fairly small, lower than that for raobs).

• AirDat flags for RH are used to screen TAMDAR RH obs (if AirDAT indicator shows uncertainty of greater than 10% the data are rejected).

• Dev2 and dev 20km tests at FSL are set to both initialize at identical time every hour to ensure that TAMDAR is the only observation difference. The dev and dev2 RUC20 runs also read Eta boundary conditions from the same directory in an attempt to ensure no difference from this factor.

• "Spamming" procedure for aircraft data in RUC analysis: This flags all obs from a given tail number if at least 25% have some problem in the buddy check QC. This check is definitely improving the QC for those TAMDAR aircraft that have continuing problems.

• In buddy check QC, obs from a given tail number cannot corroborate each other (since given aircraft can have recurring problems). Again, this makes the TAMDAR QC more robust. (See p.25 from PDF manuscript file, section 4.e - Benjamin, S.G., D. Devenyi, S.S. Weygandt, K.J. Brundage, J.M. Brown, G.A. Grell, D. Kim, B.E. Schwartz, T.G. Smirnova, T.L. Smith, and G.S. Manikin, 2004: An hourly assimilation/forecast cycle: The RUC. Mon. Wea. Rev., 132, 495-518 (Feb. issue)).

• Buddy check QC is performed also for temperature and wind obs from TAMDAR, along with all other obs in an hourly window for a given RUC cycle.

• All TAMDAR reports are treated in their 3-d location within the hourly window. RUC does not assume a sounding, like a raob.

• Major differences between dev and dev2 runs are investigated to ensure that these differences can be plausibly attributed to TAMDAR obs. We have an automated script that logs any differences that may be relevant.

• Verification of dev and dev2 RUC cycles is performed every 12h against raob data, with 3 different verification domains, over the Great Lakes area, over the eastern US, and over the entire RUC domain. Statistics from this raob verification are inspected for temporal consistency. A major outlier persisting from one time to another is investigated if a processing problem occurred rather than a TAMDAR-related difference.

Interpretation points

• Some non-local TAMDAR impacts are possible, even far from any TAMDAR reports (e.g., off the West coast or near the southern boundary of RUC domain). Assimilation of the TAMDAR winds, in particular, can modify the always-present wind/mass imbalance, which results in some gravity wave activity in the model. These gravity waves are always present to a modest extent, even despite the use of a digital filter initialization every hour in the RUC model. The exact strength and location of the waves is different with the addition of TAMDAR data, especially its winds, and this can result in some differences in temperature and moisture far from the TAMDAR data, even to the west and south.

• Some runs may be missed, leading to non-TAMDAR differences. We plan to re-synch the TAMDAR (dev2) vs. non-TAMDAR (dev) RUC runs every other night at 09z.