Data Archive of Tracer Experiments and Meteorology (DATEM) for Dispersion Models Evaluations

An objective performance evaluation against independent measurement datasets, such as tracer experiments, is fundamental to assess the model reliability to simulate transport and dispersion features under different meteorological conditions. The data from tracer experiments are essential for model evaluation. Such experiments use passive tracer releases and employ well-designed sampling networks for meteorological variables and tracer concentration. NOAA’s Air Resources Laboratory (ARL) was involved in conducting tracer experiments in different geographic locations and a wide range of scales from sub-kilometers to thousands of km.

The Data Archive of Tracer Experiments and Meteorology, DATEM, was developed to provide access to experimental data, relevant reports, meteorological data, statistical analysis, and display software for controlled tracer experiments. For assessing the performance of ARL’s dispersion model, HYSPLIT, the DATEM archive allows users to perform analysis for many of ARL’s tracer experiments and carry out their own tests if they choose. DATEM first became available to the public in the early 2000s. Recently, we added tracer experiments in finer spatial and temporal scales and one more meteorological dataset. DATEM has standardized programs and formatted data so that it is easy to conduct sensitivity tests and compare model results with each other. We commonly use “Rank” that combines four metrics of the model errors to indicate the overall model performance.

The available meteorological datasets for DATEM initially include the Global Reanalysis (GBL) and the North American Regional Reanalysis (NARR). The GBL dataset has coarse spatial resolution and low temporal frequency, while NARR data do not have momentum flux. We have added another dataset generated using the WRF model with 27-km horizontal grid spacing and hourly temporal frequency from 1980 to the present year. The WRF data provides additional variables relevant to atmospheric dispersion, including friction velocity, turbulence kinetic energy (TKE), and time-averaged wind fields. The WRF dataset lets DATEM expand its capabilities for using different meteorological inputs and providing various options to compute the HYSPLIT mixing parameters.

We conducted HYSPLIT simulations using the NARR and WRF datasets and their corresponding mixing configurations for nine controlled tracer experiments in North America. In addition to using different meteorological datasets, we conducted dispersion ensembles (15 members) by varying physics settings in HYSPLIT and utilizing the WRF meteorological dataset. The dispersion results were evaluated against measured tracer concentrations and compared with each other on a common basis in the DATEM archive. Through the evaluation, we better understand the sensitivity of the model due to the use of different meteorological datasets and the change of the HYSPLIT model’s parameters.