Modelling long-range dispersion1 May 2000
Chernobyl, and more recently the Gulf war, have acted as catalysts for improvements both to early warning systems and the complex computer models on which they rely.
It took the Chernobyl accident in Spring 1986 to highlight how ineffective most nations were at passing comprehensible radiological information between one and other. Up to that point the International Atomic Energy Agency’s (IAEA’s) role in co-ordinating emergency planning had been growing steadily, following Three Mile Island in 1979. However, the Ukrainian disaster acted as a catalyst for both the re-evaluation of the IAEA’s emergency response system and the issuing of guidelines for information exchange and co-operation among member states.
The IAEA’s member states supported the setting up of two international conventions. One focused on early notification, the other on assistance in case of a nuclear accident. Both now form the basic framework of the Agency’s emergency response system.
As part of the early notification convention, the World Meteorological Organisation (WMO) was asked by the IAEA and other international organisations to arrange for early warning messages about nuclear accidents to be transmitted over the global telecommunications system (GTS).
In the late 1980s the IAEA and WMO presided over the setting up of regional specialised meteorological centres (RSMCs) at Toulouse in France, Bracknell in the UK and Montreal in Canada in order to improve early warning systems. To begin with, Meteo-France in Toulouse provided global coverage until each WMO region was able to set up at least two RSMCs for transport model products.
However, it then took the Gulf war to highlight the fact that many of the models being used to predict smoke plume behaviour were, at best, producing misleading and confusing information. It was a point of particular concern to the US.
As a result, in November 1992 the US National Oceanic and Atmospheric Administration demonstrated its RSMC capabilities to the WMO and was formally accepted the following July. Today there are also RSMCs at Beijing, Tokyo, Melbourne and Obninsk in the Russian Federation.
Under the IAEA convention on early notification, member states agree to notify the Agency of any nuclear incident which could affect neighbouring countries. The IAEA’s role is to record this incident notification, inform other states that may be affected and also to give details which may be useful in helping other member states or international organisations to respond.
If the Agency verifies that the emergency is likely to affect neighbouring countries, it then contacts member states in a number of ways, including the WMO’s GTS.
In terms of assistance, the international convention calls for swift help to be given to any member state in an effort to minimise any radiological impact, and to protect life, property and the environment. Specifically, the convention states: “If a state party needs assistance in the event of a nuclear accident or radiological emergency, whether or not such accident or emergency originated within its territory ... it may call for such assistance from the Agency.”
Under the same convention, the Agency is also committed to respond to requests for support from the state and, where necessary, act as a broker between states which provide and receive the assistance.
However, as the Gulf war showed most recently, predicting what help will be needed – and where – depends on the accuracy of the particular dispersion models used.
Looking for a super model
In the US, the Washington RSMC is a joint venture which combines the forecasting skills and operational capabilities of the National Centres for Environmental Prediction (NCEP) with the analysis abilities of the Air Resources Laboratory (ARL).
NCEP provides a 24 hour per day initial contact point for assistance requests and its staff connect to ARL’s computer system in the event of an accident. Customised transport and dispersion models are then run with outputs being distributed automatically to predesignated country representatives.
Such outputs would include forecasts of trajectories, exposures and deposition using the Hysplit (Hybrid Single-Particle Lagrangian Integrated Trajectory) transport and dispersion model. Hysplit is a complete system for computing simple air parcel trajectories to complex dispersion and deposition simulations. Following a joint effort between the US National Oceanic and Atmospheric Administration (NOAA) and Australia’s Bureau of Meteorology the model has recently been upgraded. Version 42, released in December, includes improved advection algorithms, updated stability and dispersion equations, a new graphical user interface and the option to include modules for chemical transformations.
The model can be run either interactively on the Web through the Ready system on the NOAA’s site, or the code and executable and meteorological data can be downloaded to a PC using Windows 95 or later.
Recently a joint ARL and NCEP project began in an effort to develop a more operational-coupled meteorological dispersion model. Modifications to NCEP’s regional spectral model (RSM) mean that it can be applied over any region of the world. RSM model outputs are then linked directly with the ARL dispersion models.
In a further effort to simplify the whole process, software has been developed for MS-DOS PCs to provide a basic level of training for operations and research staff that might be involved with atmospheric emergency response functions of the RSMC. The software is provided on the internet in a self-extracting .EXE file and is available in English, French and Spanish.
In addition, there is an ongoing programme at the ARL aimed at identifying and explaining why occasional differences in RSMC predictions arise. These can be due to differences in model spatial resolution and in methods used to compute pollutant dispersion. In the meantime, regional RSMCs such as Washington and Montreal will issue joint statements on differences between their products.
The NAME of the game
One of the best known models is the UK Meteorological Office’s Nuclear Accident Model, or NAME. It was developed as part of the National Response Plan (NRP) established by the UK government as a result of the Chernobyl accident. Today it has a worldwide reputation as a leading atmospheric dispersion model and as a versatile tool for accident and episode analysis and pollution forecasting.
It was also used in the Gulf war when concern was growing that smoke from burning oil wells could reach the stratosphere, where it could stay for decades and possibly have an impact on climate. However, NAME was successfully able to show that such a risk was minimal.
“All of the modern long-range dispersion models since Chernobyl are concerned with transboundary pollution movements and different RSMCs have evolved their own different models,” said UK Meteorological Office environment manager Karl Kitchen. “However, ours is generally regarded as being one of the best: in a recent test, called ETEX, against 50 or 60 others which involved modelling the release of a trace gas from Brittany in France, NAME came first or second in almost every category.”
NAME follows the 3D trajectories of air parcels and computes air concentrations and ground depositions. It also runs simulations from a few kilometres to tens of thousands of kilometres from the site of an accident.
The model’s strength lies in its ability to use full 3D global weather data, something which few other models can do. In addition, it takes into account dry and wet deposition of airborne material on the underlying surface, a factor which was later found to be crucial to the understanding of how Cs-137 from Chernobyl moved through the environment. NAME’s key features include: plume rise; realistic boundary layer simulation; upper-level transport; all spatial scales catered for; fully flexible emissions; and a suite of diagnostic tools.
Because any nuclear incident is likely to be complex and involve other countries, the Met Office has assisted in the development of PACRAM – ‘procedures and communications in the event of a release of radioactive material’ – as part of the NRP. A major component of these communications is a set of national gamma radiation monitoring sites and a nuclear emergency response system known as the Radioactive Incident Monitoring Network, or RIMNET. Together with NAME, this system provides the main warning mechanism for the UK Government against any future problems.
If, as happened with Chernobyl, an overseas incident threatens the UK, the Department for the Environment, Transport and the Regions (DETR) takes the lead and the NRP would be implemented and RIMNET activated. Then NAME would be run and its results used to issue information to government agencies and local authorities.
To make sure that the PACRAM procedures do in fact work, regular accident exercise are carried out involving simulated nuclear accidents.
The most recent was called Best Endeavours 1999 and involved a simulated nuclear release from Finland. If such an accident occurs at an overseas location and a radioactive cloud appears across the UK, then part of the response is to monitor the gamma radiation using RIMNET and then to combine this data with the output of NAME. In this simulation a historic weather sequence was used for the calculations. The radioactive cloud appeared over Scotland a few days later and proceeded to move south across the UK. NAME was used to provide simulated estimates of the radioactivity from the gamma radiation monitoring sites and to supply forecasts of the movement of the cloud and deposition of radioactivity on to the ground. The exercise lasted five days, during which the Met Office supplied NAME output in real time to government departments and local authorities.
However, these warning systems are always being refined and the latest development is the Environment Monitoring and Response Centre (EMARC). Opened in January at the Met Office’s Bracknell site, it provides co-ordinated and rapid meteorological responses to environmental incidents. Specifically, it will work very closely with other government departments, the IAEA and other overseas met services, such as Meteo-France, to respond to the threat of nuclear accidents worldwide.
Because of its flexibility, NAME is already helping local authorities, industry and the public to prepare for increases in levels of pollutants such as carbon monoxide, NOx and SOx. It also predicts levels of particulate matter (PM10) that form in the atmosphere under certain meteorological conditions.
Factors such as variations in traffic flow and fluctuations in the power supply demand are also brought into the equation. These elements are fed into the system to forecast the most accurate picture of pollution according to the day of the week and the time of the year.
The data will be delivered to local authorities via email to allow air quality warnings to be issued as rapidly as possible to the local media in potential pollution ‘hot spots’. A customised internet web site will show pollution levels throughout the area, with links to public transport information timetables and other useful environmental information.
The aim is that these warnings will persuade industry to curb industrial processes that contribute to pollution at times of poor air quality. The information can also be used to promote sustainable methods of transport and encourage people to travel by public transport.
“Our ultimate aim is that this information will actually influence behaviour – encouraging people to use alternative methods of transport and persuading industry to take measures to ensure it doesn’t contribute further to pollution levels,” said Kitchen.
While a PC version of NAME for medium to long ranges should be available later this year, one of the biggest modelling challenges remains the ability to handle the more short range aspects of atmospheric dispersion in the same manner as long range modelling is currently handled. Topographical problems such as near-building effects and small hills have long caused problems for developers of these particle-based models.
“For the nuclear industry, it would be a benefit because they could look at the effect of incidents over any range. This level of modelling is not required under EC law at the moment, but it could change in the future. It should also replace a lot of the very outdated models, such as R91, which are still in use out there,” noted Kitchen.
The holy grail as far as all modellers are concerned is to be able to integrate every dispersion aspect, from near source and very long range, with all of the complex chemistry needed to understand and study dispersion, chemistry and depositions within a single model framework.
“Currently a number of models are needed to study the range of questions asked. Separate models are needed for short range, long range, atmospheric chemistry, dense gases, deposition and so on. To bring these together into a single model would be a great leap forward. Our three-year research project beginning in April is targeted at doing just that,” said Kitchen.