Environmental Science Computing. Important Tasks treated by the Environmental Models

Environment and Society. The environmental problems are becoming more and more important for the modern society, and their importance will certainly be increased in the near future. High pollution levels (high concentrations and/or high depositions of certain harmful chemical species) may cause damage to plants, animals, and humans. Moreover, some ecosystems can also be damaged (or even destroyed) when the pollution levels become very high. This explains why the pollution levels must be studied carefully in the efforts

- to predict the appearance of high pollution levels, which may cause different damages in our environment and/or
- to decide what must be done to keep the harmful concentrations and/or depositions under prescribed acceptable limits.

The control of the pollution levels in highly developed and densely populated regions in the world is an important task that has to be handled systematically. This statement is true for many regions in Europe and North America but also other parts of the world are under economic development currently and urgent solutions of certain environmental problems either are already necessary or will soon become necessary. The importance of this task has been increasing steadily in the beginning of the new millennium. The need to develop reliable and easily applicable control strategies for keeping harmful pollution levels under prescribed limits will become even more important in the next decades.

Climate changes are causing another challenging problem for the modern society. The quick changes have many different consequences. The impact of the climatic changes on the pollution levels is one of the consequences, and this consequence must be investigated carefully by studying the relationship between climatic changes and high pollution levels. It should also be mentioned here that there is a feedback: The pollution levels influence the climatic-changes. It is necessary to couple environmental models with climatic models to study fully the interrelations between climatic changes and pollution levels. This task is very challenging.

Mathematical models are powerful tools when the trends in the development of the pollution levels and the measures which the society must take to ensure a sustainable development are studied. These models are often very complex and lead to huge computational tasks. Some tasks cannot be treated even if powerful modern computers are used.

Important Tasks treated by the Environmental Models. Advanced mathematical models for studying environmental phenomena can be used successfully to design control strategies for keeping the pollution under critical levels under the assumption that these models produce reliable results. The application of comprehensive environmental models in sensitivity tests is important in the efforts

- to understand better the physical and chemical processes involved in the environmental phenomena or to treat efficiently the tasks proposed by policy makers and
- to ensure that the control strategies for keeping the pollution under the prescribed acceptable limits are reliable.

Sensitivity tests can be applied to resolve these two tasks. It is important to study the sensitivity of concentrations and deposition of harmful pollutants caused by variations of:
- anthropogenic emissions,
- biogenic emissions,

- meteorological conditions,
- velocity rates of chemical reactions,
- boundary conditions,
- initial conditions, and
- numerical algorithms.

This list is certainly not complete and can be continued. It is even more important to emphasize the fact that this list tells us that the task of performing complete sensitivity analysis by applying large-scale environmental models is extremely large and very difficult. Finally, many terms, in which the parameters from the above list are involved, are nonlinear. The nonlinearity causes great difficulties, which can be resolved only by conducting many experiments with different scenarios and studying carefully the results to find typical trends and relationships.

The difficulties are increased because there are interconnections of the effects because of variation of different parameters. For example, the variation of both the anthropogenic emissions and the biogenic emissions may lead to some effects, which are not observed when only the anthropogenic emissions or only the biogenic emissions are varied. The effects caused by simultaneous variations of several key parameters can only be studied by increasing the number of scenarios used in the experiments. Thus, the tasks become larger and more difficult.

The necessity of validating the results is an additional source for difficulties. The problem of designing a completely reliable routine for validating the results of the sensitivity tests is still open. Two approaches can be used (and, in fact, are commonly used) in the attempts to validate the results from sensitivity analysis tests:

- comparisons with measurements and
- comparisons with results obtained by other models.

The objections, which can be raised against the complete reliability of the comparisons with measurements for the validation of the model results, are many and serious. The most important objection is the well-known fact that two different quantities are compared when such a procedure is used.

The measurement is a quantity, either concentration or deposition, which is obtained at a given geographical point (the location of the measurement station). The corresponding quantity, which is the quantity calculated by the model, is a representative mean value averaged in some surrounding (determined by the spatial discretization chosen) of the point in which the measurement station is located.

This fact implies that even if both the measurement and the corresponding calculated result are exact (which will never happen in practice), they will in general be different. Another implication, which is even more important from a practical point of view, is the following: We should expect to improve the possibility (the potential possibility, at least) of getting better validation results by using comparisons with measurements when we increase the spatial resolution of the model, but the computational tasks become larger and much more difficult when the spatial resolution is increased.

It may become necessary to replace some physical and chemical mechanisms used in the model with coarse resolution with more accurate mechanisms when the resolution is refined. Finally, the need for accurate input data for large-scale models defined on refined grids also causes great difficulties.

The objections, which can be made in the case where the results obtained by two or more models are compared, are also both many and serious. It is extremely difficult to determine in a reliable manner the precise reason for differences of results produced by different models.

The answer of the following question is interesting when a long sequence of sensitivity tests is run: What is the relationship between the parameter that is varied and the studied quantity (the concentration or the deposition of a certain harmful pollutant)? If two models are run with the same sequence of sensitivity tests and if the relationship between the parameter that is varied and the model results is different for the two models, then the difference may, partially or totally, be caused by several reasons.

The differences may, for example, be caused (or, at least, be influenced to some degree) by the use of different projections and/or resolutions in the different models, by the application of different sets of input data, by the fact that the numerical methods used are different, by the fact that the chemical schemes are not the same, and so on. It is not clear how these unwanted reasons for differences of the results from two or more models can be eliminated to study only the relationship between the parameter we are varying and the studied quantity.

It should be emphasized here that the objections against the two commonly used procedures for validating results from sensitivity tests indicate that it is necessary to be cautious. It should also be emphasized, however, that it is absolutely necessary to do such comparisons. Many sound conclusions can be drawn after such comparisons, but one should not fully rely on the results of the comparisons. One should continue the search for better and more reliable validation tests.