Mesoscale Forecasting System with Commercial Aircraft (TAMDAR) Observations

Figure 1. Comparison of precipitation forecasts from the RTFDDA, RUC and NAM models with the Stage IV analysis (upper-left panel, based on a merger of radar and rain-gage data).  The dashed lines indicate the axes of precipitation features in the Stage IV analysis.

Recently a private firm, AirDat, LLC, launched a new generation atmospheric observing system utilizing the TAMDAR (Tropospheric Airborne Meteorological Data Reporting) atmospheric sensor. AirDat is deploying the TAMDAR sensors on aircraft nationwide, building a completely new atmospheric data collection network.. The sensors measure a complete set of upper-air weather variables including winds, temperature, pressure, humidity, icing, and turbulence, and report in near-real-time as aircraft climb, cruise and descend. In collaboration with AirDat, RAL has developed a mesoscale forecasting system, based on its real-time four-dimensional data assimilation and forecast (RTFDDA) system, which incorporates TAMDAR data. This modeling system is being used to conduct research on the impact of TAMDAR data on mesoscale data assimilation and forecasting and on new methods for optimizing TAMDAR data usage in numerical weather prediction (NWP).  Figure 1 illustrates the results of a test where the RTFDDA system used TAMDAR data for precipitation forecasts, and the results were compared with National Weather Service forecasts from the RUC and NAM models.

FY06 Accomplishments:

Statistical verification of the precipitation forecasts from the TAMDAR-based RTFDDA model, and the RUC and NAM models was performed by comparing the model forecasts to the NCEP Stage IV hourly precipitation analyses. The verification results show a significant advantage of the RTFDDA forecasts in terms of the prediction of the total number of precipitation objects, their size, and their intensity distributions (See figure below). The model worked particularly well in predicting strong mesoscale rain features.  TAMDAR data’s impact on forecasting a wintertime massive cold-air outbreak was also evaluated, as was the value of the data in predicting severe convection. A key element of all TAMDAR-related research efforts is the need to identify and correct bad data and systematic biases; quality control is an on-going concern if TAMDAR data is to be useful in NWP models. An Observation System Simulations Experiments (OSSE) testbed was developed and used to study the potential impact of the anticipated full-fleet deployment of TAMDAR sensors on mesoscale NWP. The WRF-based-RTFDDA system was used to generate a nature run and to conduct data impact experiments. Experiments with a “perfect data” assumption suggest a significant improvement of the model's 0–48 h forecasts, using TAMDAR instead of standard soundings.

FY07 Plans:

While it was expected that TAMDAR data would have a significant positive impact on NWP model performance, efforts to date have not clearly demonstrated this. It remains a challenging task to further examine the data quality and properly incorporate the observation-error statistics into the data-assimilation weighting scheme. At issue is the fact that the lower troposphere, especially during the daytime when most TAMDAR flights occur, is very turbulent. The representativeness errors of in-situ observations from TAMDAR, relative to the operational model grids and the TAMDAR sampling density, can be very large. Another difficulty is that in the summer aircraft typically avoid convective regions, and thus the collective TAMDAR observations are biased to clear air.  Therefore, if the data are used with horizontally isotropic weighting functions and invariant influence radii, the analyses with the data will lead to a dry bias. Further studies of TAMDAR data quality, representativeness errors, and data assimilation weighting functions are, therefore, necessary. The OSSE testbed tool, in conjunction with real-data, will continue to play a significant role in these research activities.