Short-term air quality forecast system for AIR TRITIA region (PWS)

1. Introduction

One of the main objectives of AIR TRITIA project "Uniform approach to the air pollution management system for functional urban area in Tritia Region" was to develop an air quality prediction system in the cities of Tritia region.

According to the provisions of the project, the objective of the system was to inform about predicted (one hour) concentration of air pollutants. This was due to the fact that the Directive 2008/50/EC of the European Parliament and of the Council of May 21st 2008 on ambient air quality and cleaner air for Europe implemented in the legislation of the EU member states requires that the member states draw up short-term action plans in order to reduce the risk or to decrease the duration of such occurrence if the specific region (area of agglomeration) is at risk of air pollution level exceeding one or several alarm thresholds listed in the annexes to the aforementioned Directive.

While the directive does not list air quality forecast as an instrument to implement short-term plans, use of such forecasts is increasingly becoming a good practice at the age of dynamic development of weather forecasts and due to the fact that informing the public of any threats resulting from health risks associated with air pollution.

2. Air quality forecast for TRITIA region – general assumptions

2.1 Selection of forecast model

A detailed analysis of forecasting opportunities (see: report) for TRITIA region has resulted in an air quality forecast model based on advanced statistical methods (data mining) and on numerical weather prediction.

Data mining is an analytical process, the purpose of which is to examine large data pools to find regularities and systematic correlations between the variables and to evaluate the results by applying the detected models to the new data subsets. The ultimate objective of data mining is to forecast e.g. customer behaviour, sales volume or certain physical phenomena.

The data mining process includes three fundamental stages:

  • preliminary exploration,
  • building a model, which includes evaluation and verification,
  • implementation and use of the model for new data to acquire the predicted values or classifications.

Long-term studies of meteorological conditions and concentration of pollutants, mostly particulates, have shown high correlation between air quality and values of certain meteorological elements. The most important ones include air temperature, wind speed and direction, precipitation, relative and absolute air humidity, atmospheric pressure, cloud cover, fog occurrence, vertical air temperature stratification, inversion occurrence (see: chapter 2). Weather, and temperature in particular, controls emission and the conditions of pollution dispersion.

The elements listed above can be identified in COSMO LM numerical forecast. The forecast covers three subsequent days and is updated every six hours.

A high number of available forecasts, for at least 5 years, allows to find a sufficient number of accurate forecasts.

2.1.1 Forecast assumptions

Air quality forecasts are based on the following assumptions:

  • Purpose of prediction (air quality forecast) – air pollution level over 72 hours (3 subsequent days);
  • The location of prediction is a specific point (city).
    Five cities have been selected for the forecast: Opava, Opole, Ostrava, Rybnik, Žilina;
  • Predictive data necessary to draw up an exploration forecast;
    • current numerical weather forecast for the prediction location;
    • historical weather forecast set over the course of several years (at least 5) of the prediction location;
    • a sequence of air pollution measurements at the dates corresponding to the historical weather forecasts, preferably close to the prediction locations.

This stage of drawing up weather forecasts includes generating a set of output data in the form of forecast PM10 and PM2.5 particle concentration over time as well as forecast hourly and daily air quality index over time. Ultimately the index shall be based on the European Air Quality Index (EAQI). However, as the index currently does not include hourly forecasts, CAQI index is used.

2.2 Air quality index as the basic information on concentration forecast

So-called air quality index is used to quickly analyse air quality conditions in addition to the measured or forecast absolute values. It is an alternative to using absolute pollutant concentration values. This globally used method is becoming increasingly important in Europe as well. Its popularity stems from the people's limited awareness of air pollution and their habit of using pictograms in public communication. Green, yellow and red colour are intuitively perceived as signalling good or negative conditions. This has been used by attributing absolute values of the indexes to different colours. AQI is usually calculated for several pollutants at a time. Air quality level is defined by the specific partial AQI, which adopted the highest (most unfavourable) value.

There are many air quality indexes in the world. They have been developed since the first automatic monitoring networks were created. Currently the aim is to standardize the approach in the areas using uniform standards.

Air quality is classified in the European Union by using the Common Air Quality Index (CAQI). Its formula has been defined for the purpose of comparing air quality in various European cities. The index is used by the European Environmental Agency [EEA 2010] as part of Eye On Earth system sharing information on air quality.

CAQI index is calculated for NO2, PM10, O3, CO, SO2 at the same time. Table 1 lists the pollutant concentration ranges attributed to individual CAQI index classes. Classes 1-3 refer to satisfactory quality of air, that is, pollution, which does not pose a threat to human health. Classes 4 and 5 indicate poor air quality - the level of pollution poses a threat to health of particularly sensitive groups (class 4) or to everyone (class 5).

Three different indexes have been drawn up for comparison at three different time scales:

  • Hour index: describes the current (today's) air quality based on hourly values, which are updated on an hourly basis,

  • Daily index: means the previous day's general air quality based on daily values - it is updated once in a day,

  • Annual index: it represents general air quality in the city for an entire year, compared to the European air quality standards. This index is based on average annual values compared to the annual permissible pollution levels and is updated once in a year.

Return to main page