Title of Example |
||
|
Short term AQ forecast methods in Seville |
||
Example |
|||||||||||
|
Introduction Ozone (O3) is a powerful oxidant that forms in trace amounts in two parts of the atmosphere: the stratosphere (the layer between 20-30 km above the earth’s surface, also known as “the ozone layer”) and the troposphere (ground-level to 15 km). Stratospheric ozone is formed naturally and shields life on earth from the harmful effects of the sun’s ultraviolet radiation. However, ground-level ozone can be harmful to human health and ecological receptors, the extent of which depends on the ozone concentration, exposure duration, exposure pattern and ventilation. Ozone is a secondary pollutant, in that is not emitted directly from tailpipes or smokestacks, but instead is formed in the atmosphere as a result of reactions between other directly emitted pollutants (ozone precursors). Ozone is formed by a complicated non linear series of chemical reactions, initiated by sunlight, in which nitrogen oxides (NOx) and volatile organic compounds (VOCs) react to form ozone. VOCs are produced mainly by road traffic and the use of products containing organic solvents. NOx is mostly emitted from transport and combustion processes. Although NOx and VOCs are the most important precursors of elevated levels of ozone, production of ozone can be also stimulated by carbon monoxide, methane, or other VOCs produced by plants, trees and other natural sources. Apart from ozone precursors emissions it is found that additional factors are directly implicated in the ground level ozone concentration in metropolitan areas: weather conditions, which cause the precursors to interact photochemically and to disperse in the atmosphere, and features of the area like street width and building height. The
potential for ozone damage has been known for decades but it is only in the
most recent years that its impact has become of concern in Article 7 of Directive 2002/3/EC sets out the requirements for short-term actions plans. It is for Member States to identify whether there is significant potential for reducing the risk, duration or severity of any ozone exceedance, taking account of the national geographical, meteorological and economic conditions. Dissemination of information on ozone concentration combined with adequate forecasting may reduce the exposure duration or exposure intensity of the population to the high ozone values. In major cities and regions in Mediterranean countries a proper meteorological forecast-analysis is always required and it should be very highly tuned to the local-regional meteorological process. Systems for forecasting and information of ozone episodes are usually based on statistical relationships between weather conditions and ambient air pollution concentrations. The most widespread technique used for this purpose is the multivariate statistical approach. However, pollution-weather relationships imply complex and non linear properties, especially for ozone. In this sense, the problem of ozone forecasting can be well-suited by neural networks technology, which allows to incorporate nonlinear relationships to make somewhat more accurate predictions of ozone than regression models using the same set of input data. Ozone forecasting in the urban area of Sevilla In the urban area of Sevilla, like in most Mediterranean cities, road traffic is the main responsible for the observed ozone levels since it constitutes the major source of VOCs and NOx. At present, Sevilla is carrying out an action plan in order to elaborate an ozone forecasting model using neural network technology (see the Annex below for more details on the method). By the application of this model, it will be possible to characterize the spatial distribution of ozone concentrations and therefore it will constitute a basis for short-term action plans related to the traffic management in the city. The
first phase in the design of a neural network model is to obtain a great number
of data from past and current measurements. To evaluate ozone pollution in the
city of The following phase is the identification of the VOCs/NOx ratio in episodes of high ozone concentrations. This ratio is in fact one of the main aspects to be taken into account when studying ozone concentrations in ambient air. It is known that a decrease in NOx can lead to an increase in O3 at low VOCs/NOx ratios under specific conditions. It is often called the VOC-limited regime and hence, emission control of organic compounds is more effective for reducing peaks values of ozone pollution locally. On the contrary, at high VOC/NOx ratios, the chemistry tends towards the NOx-limited regime and NOx reductions are considered more effective for reducing ozone. The value of this ratio is highly influenced by geographical and meteorological conditions. The identification of the VOCs/NOx ratio is to be made according to measurements. An exhaustive campaign is carried out, using online monitors equipped with O3, NOx and VOCs sensors in measuring stations allocated in strategic places over the city. In parallel, the identification of the meteorological conditions over the same period of time is to be completed. Equally important is the identification of the most relevant VOC precursors and their reactivity since this factor determines the level of ozone formation as well. A further analysis of the influence of VOCs in ozone generation is also to be performed. The
third phase of the action plan consists on the neural network approach to
estimate and predict ozone concentration levels. So as to do this approach, a
preliminary neural network model, developed by the Chemical and Environmental
Engineering Department of the A description of the modelling techniques by neural networks, pointed out the main advantages in comparison with other conventional techniques, is given in the annexe of this example. For the elaboration of the neural network model, the following input variables (1-hour average) were chosen as the most relevant variables affecting ozone concentration (Figure 1): ▪ Weather Conditions: - Temperature - Relative moisture - Wind speed - Wind direction - Lagged UV radiation (2 hours) ▪ Lagged traffic flow (4 hours) ▪ Lagged ozone concentration (ozone maximum level from the previous day) ▪ Features of the area: - Street width - Building height - Street orientation - Boundary of the area The model output variable is the estimated ozone concentration.
Figure 1: A schematic of the neural network for ozone forecasting The regression analysis of the neural network model results obtained provides accurate ozone forecasting, as it is shown in Figure 2.
Figure 2: Regression analysis of the neural network for ozone forecasting Activities to be performed within this phase include the following: - Validation
of the neural network in the measuring stations previously selected to
elaborate and train the neural network developed by the Chemical and
Environmental Engineering Department of the - Selection of additional measuring stations according to historic ozone levels and input variables availability. - Evaluation and validation of the neural network in the measuring stations. - Comparison of neural network O3 estimations and monitoring network O3 measurements. The confirmation that the O3 concentrations estimated by the neural network can be assumed as the O3 concentrations measured in ambient air will lead to take appropriate steps to limit and prevent exposure and to warn authorities, industry and public to adopt emission reduction measures. In this sense, the capabilities of the model perfectly match the requirements set up in the European Directive relating to ozone in ambient air. Bibliography - Directive
2002/3/EC of the European Parliament and of the Council of - Guidance for implementing Directive 2002/3/EC of the European Parliament and of the Council of 12 February 2002 relating to ozone in ambient air. June 2002 - Ad-Hoc working group on ozone directive and reduction strategy development. Ozone position paper. Julio 1999. - Guideline for developing an ozone
forecasting program. - A.C. Comrie. Comparing Neural Networks and Regression Models for Ozone Forecasting. Journal of the Air & Waste Management Association. June 1997. - G. Reyes; V.J. Cortés. Ozone forecasting in the urban area of Annex
Artificial neural network technology for ozone forecasting 1. Fundamentals Artificial Neural Network (ANN) technology is an approach to describe physical system behaviour from process data, using mathematical algorithms and statistical techniques. ANNs simulate biological neural systems, in that they are made up of an interconnected system of nodes (neurons) and in terms of learning and pattern recognition. These nodes are operating in parallel and inspired by biological nervous systems. A neural network can be trained to identify patterns and extract trends in imprecise and complicated non-linear data. A particular function can be performed by adjusting the values of the connections (weights) between elements following a determined training algorithm. Neural networks have been under development for many years in a variety of disciplines to derive meaning from complicated data and to make predictions. In recent years, neural networks have been investigated for the use in pollution forecasting. Because ozone formation is a complex non-linear process, neural networks, which allow incorporating nonlinear relationships, are well suited for ozone forecasting. 2. Strengths of artificial neural networks Many methods exist for predicting ozone concentration. Table 1 summarises the most commonly used forecasting methods. Strengths of ANNs include the following: - ANNs allow for non-linear relationships between variables. The method can weight relationships that are difficult to subjectively quantify. - Neural networks have the potential to predict extreme values more effectively than regression. - Once the neural network is developed, forecasters do not need specific expertise to operate the ANN. - Neural networks can be used to complement other forecasting methods, or used as the primary forecasting method. On the other hand, neural networks are complex and not commonly understood and hence the technology can be inappropriately applied.
Table 1: Comparison of forecasting methods.
3. Neural Network architecture The basic structure of an ANN involves a system of layered, interconnected neurons. The neurons are arranged to form an input layer, one or more “hidden” layers and an output layer, with nodes in each layer connected to all nodes in neighbouring layers (Figure 1).
Figure 1: The architecture of a multi-layered feed forward neural network. The layer of input neurons receives the data either from input files or directly from electronic sensors in real-time applications. The output layer sends information directly to the outside, to a secondary computer process, or to other devices such as a mechanical control system. The internal or hidden layers contain many of the neurons in various interconnected structures. The inputs and outputs of each of these hidden neurons go to other neurons. In most networks each neuron in a hidden layer receives the signals from all of the neurons in a layer above it. After a neuron performs its function it passes its output to all of the neurons in the layer below it, providing a feed forward path to the output. Artificial neurons comprise seven major components, which are valid whether the neuron is used for input, output or hidden layers: 1) Weighting factors, which are adaptive coefficients within the network determine the intensity of the input signal. These input connection strengths can be modified in response to various training sets and according to a network specific topology or through its learning rules. 2) Summation function, which transforms the weighted inputs in to a single number. The summation function can be complex as the input and weighting coefficients can be combined in many different ways before passing on to the transfer function. The summation function can select the minimum, maximum, majority, product or several normalizing algorithms depending on the specific algorithm for combining neural inputs selected. 3) Transfer function, which transforms the result of the summation function to a working output. In the transfer function the summation total can be compared with some threshold to determine the neural output. If the sum is greater than the threshold value, the processing element generates a signal. If the sum of the input and weight products is less than the threshold, no signal (or some inhibitory signal) is generated. 4) Scaling and limiting. This scaling multiplies a scale factor times the transfer value, and then adds an offset. Limiting mechanism insures that the scaled result does not exceed an upper or lower bound. 5) Output Function (competition). Neurons are allowed to compete with each other, inhibiting processing elements. Competitive inputs help determine which processing element will participate in the learning or adaptation process. 6) Error function and back-propagated value. The difference between the current output and the expected output is calculated and transformed by the error function to match particular network architecture. This artificial neuron error is generally propagated backwards to a previous layer in order to modify the incoming connection weights before the next learning cycle. 7) Learning function, which modifies the variable connection weights on the inputs of each processing element according to some neural based algorithm. The software first adjusts the weights between the output layer and the hidden layer and then adjusts the weights between the hidden layer and the input layer. In each iteration, the software adjusts the weights to produce the lowest amount of error in the output data. This process “trains” the network. 4. Neural networks training
Training and production are essential for the neural network application (Figure 2).
Figure 2: Essential phases of the neural network application: training and production The development of ANNs comprises the performance of a series of consecutive steps. In addition, a thorough knowledge of the process to be modelled is also required. The general steps to develop neural networks for ozone forecasting are the following: - Complete historical data analysis and/or literature reviews to establish the air quality and meteorological phenomena that influence ozone concentrations in the area under study. - Select parameters that accurately represent these phenomena. This is a critical aspect in developing the neural network since an appropriate selection improves significantly the results obtained by the ANN. - Confirm the importance of each meteorological and air quality parameter using statistical analysis techniques (Cluster analysis, correlation analysis, step-wise regression, human selection). - Create three data sets: a data set to train the network, a data set to validate the network general performance and a data set to evaluate the trained network.
- Train the data using neural network software. It is important not to over train the neural network on the developmental data set because an over trained network would predict ozone concentrations based on random noise associated with the developmental data set. When presented with a new data set the network will likely give incorrect output since the new data random noise will be different than the random noise of the developmental data set: the network memorized the training examples but it did not learn to generalize to new situations. One of the most commonly used method for improving generalization is called “early stopping”. In this technique, when the validation error increases for a specified number of iterations, the training is stopped, and the weights and biases at the minimum of the validation error are fixed.
- Test the generally trained network on a test data set to evaluate the performance. If the results are satisfactory, the network is ready to use for forecasting.
5. Neural networks operation The operation of an ANN is simple and requires little expertise. Although use of the network does not require an understanding of meteorology and air quality processes, it is advisable that someone with meteorological experience be involved in the development of the method and evaluate the ozone prediction for reasonableness.
As part of a forecasting program forecasters should regularly evaluate the forecast quality. The verification process can be complex since there are many ways to evaluate a forecast including accuracy, bias and skill. Many verification statistics is needed to compute in order to evaluate completely the quality of the forecast program. References - Guideline for developing an ozone
forecasting program. - Artificial Neural Networks Technology. - Ad-Hoc working group on ozone directive and reduction strategy development. Ozone position paper. July 1999. - A.C. Comrie. Comparing Neural Networks and Regression Models for Ozone Forecasting. Journal of the Air & Waste Management Association. June 1997. - G. Reyes; V.J. Cortés. Ozone forecasting in the urban area of - S. Amoroso; M. Migliore. Neural networks to estimate pollutant levels in canyon roads. Urban Transport VII. WITPRESS. 2001.
|
|||||||||||
%20%20-%20Sevilla_files/image002.gif)
%20%20-%20Sevilla_files/image004.gif)
%20%20-%20Sevilla_files/image006.gif)
%20%20-%20Sevilla_files/image008.gif)
%20%20-%20Sevilla_files/image010.gif)
Last Updated |
||
| |
13th January 2005 |
|