Classic data sets for dispersion model evaluation

Harmonisation Home page

More information on:

Previous conferences: Information and proceedings

Conference in Bologna, Italy, October 2017

Conference in Budapest, Hungary, May 2016

Conference in Varna, Bulgaria, September 2014

Conference in Madrid, Spain 2013

Model Validation Kit

Related activities and tools

John Irwin's data archive

Steering committee


At the conferences on Harmonisation within Atmospheric Dispersion Modelling for Regulatory Purposes certain data sets have received particular attention in the past, notably in relation to work with the so-called Model Validation Kit.

In 2012, additional data for several classic tracer dispersion experiments were published as described below.

One of the concerned the data sets is the EPRI Kincaid dataset, which now comprises around the double amount of data compared to the data set previously applied generally in the modelling community.

Furthermore, other interesting data have been published such as new and additional data from EPRI Bull Run, Round Hill and Project Prairie Grass.

The person behind the efforts to make the data available was John Irwin, formerly at the US NOAA/EPA. Sadly, John passed away in August 2016.
The contents of John's website ( has been transferred to the url

Below you will find a brief description of the data sets.


  • Kincaid. During the Kincaid experimental campaign there were approximately 350 hours of tracer experiments during the experimental campaign. The data were divided (by day) into two parts a developmental data base and an evaluation data base. The package distributed for use with the Model Validation Kit ( ) is the developmental data base, and it contains only half of the SF6 tracer data collected during the study, i.e. 171 hours. In 2012, also the second half of the data have become generally available. Also added for analysis are the hourly and 5-min SO2 concentration values, which can be investigated separately or in combination with the SF6 data.
  • Project Prairie Grass. Project Prairie Grass is a classic tracer data set for near-surface releases, and has been used in past studies. But there are few people who have made use of the vertical concentration profiles that were collected as part of this experiment. The vertical sampling conducted for Project Prairie Grass has now been made available.
  • Round Hill experiment. A little known study of significance is the Round Hill Experiment. Some of the researchers working on Project Prairie Grass were graduate students at Massachusetts Institute of Technology (MIT) near South Dartmouth, Massachusetts. These students conducted a series of SO2 tracer releases in 1954 and 1955 that mimic the experimental setup planned for Project Prairie Grass (which was conducted in 1956). The interesting difference is that the experimental site at Round Hill has a 30 cm roughness in contrast to the 0.8 cm roughness for Prairie Grass. These data provide a direct challenge to any model that professes a capability to respond to changes in roughness! As if this is not enough, in 1957 they conducted a new series of releases to directly observe the effects of averaging time, by sampling the first 1-min and 3-min of each 10-min sampling period. These are unique data that few have ever worked with.
  • Bull Run. In 2012 work has been conducted to post the complete set of SF6 and SO2 data collected during the EPRI Bull Run experiments, which were conducted in moderately complex terrain in contrast to the nearly flat homogeneous terrain of EPRI Kincaid. The plume was very buoyant, and the winds were often light and variable.
  • Cabauw. The Cabauw SF6 tracer study has not been used much. It involves release of SF6 tracer from a height of 80 or 200 m, with sampling at 1.5 meters above ground along a single arc that ranged from 3 to 5 km (depending on wind direction). The conditions were neutral to somewhat unstable, but never had strong convective unstable conditions. Sampling was for two consecutive 30-minute periods. Site roughness varies from 10 to 20 cm, depending on wind direction. These data are now posted on
  • Green Glow and Hanford. Several data sets are available from tracer studies conducted at the Hanford research facility near Richland, WA USA. The good thing about these data is that there is a great deal of data. The bad thing about these data is that the tracer tends to deposit, and nobody is certain how best to characterize the deposition characteristics. Deposition does not affect use of these data to test characterization of lateral dispersion. The data sets in question are: Green Glow, Hanford-67, and Hanford-64.

hvids_sm.gif (2298 bytes) See also: Home page of the "Initiative on Harmonisation within Atmospheric Dispersion Modelling for Regulatory Purposes"

This page is maintained by Helge Rørdam Olesen
It was last modified on August 10, 2017
The "Harmonisation..." web site is hosted by the Department of Environmental Science, Aarhus University (Denmark)