Lumping, testing, tuning: The invention of an artificial chemistry in atmospheric transport modeling

https://doi.org/10.1016/j.shpsb.2010.07.002Get rights and content

Abstract

Since the late 1950s computer simulation has been used to investigate the transport of pollutants in the atmosphere. About 20 years later also the chemical transformation of atmospheric pollutants was included in computer models of photochemical smog formation. Due to limited knowledge of atmospheric chemistry and due to limited computer capacity, chemical processes in the atmosphere were modeled with the help of simplified chemical models. In these models chemical substances are lumped together forming artificial virtual compounds with virtual characteristics. The paper aims at studying the practices developed in chemical model building and the creation of confidence in these models. Core of the paper will be the analysis of the Urban Airshed Model (UAM) for the Los Angeles region, a pioneering development in the early 1970s. The construction of the UAM involved the “lumping” of chemical processes and extensive testing and tuning. These practices led to a consistent model representation, in which diverse pieces of information fitted and were mutually stabilized. The pragmatic achievement of consistency created confidence, even though empirical tests of the models remained ambiguous and problematic.

Introduction

Computer models and simulation in the atmospheric sciences involve peculiar features, which have been described as “epistemic opacity”, “semi-autonomy” and “simplification”. These features have been considered as problematic, because they cause problems of model validation and assessment and involve new forms and not quantifiable ranges of uncertainty of simulation results (e.g. Petersen, 2006). The case of atmospheric chemistry modeling shows that epistemic opacity, semi-autonomy and simplification also provided gains. These characteristics—in spite of being problematic—opened windows of opportunity and created spaces for the realisation of computer models and simulations, which otherwise would have been much more restricted. The construction of the so-called Urban Airshed Model (UAM) to simulate photochemical air pollution in the Los Angeles area provides an interesting example. This model was developed in the early 1970s by a group of researchers around the atmospheric scientist John Seinfeld at the California Institute of Technology.1 It represented a scientific response to a new type of complex air pollution problem, which was first discovered in the Los Angeles area in the early 1940s.

In this paper I will analyse the construction of the UAM and explore two main issues: first, the practices that have been developed and employed in the construction and use of the Urban Airshed Model, and second, the creation of confidence and trust which the scientists eventually put into the model. The introductory section consists, first, of an outline of epistemic opacity, semi-autonomy and simplification as significant features of computer models and, second, of an introduction into the problem of photochemical air pollution. The core section is dedicated to the analysis of scientific practices developed by Seinfeld and his research group. In the final section I will discuss the characteristics of simulation practices in atmospheric chemistry modeling and the requirements and conditions for the creation of confidence in these practices.

Section snippets

Opacity, semi-autonomy and simplification

Computer models and simulation results need to be tested by comparison with measured data like any physical theory. These model tests are usually called “model validation” and—in spite of its limitations—considered crucial for creating trust into the model (Oreskes et al., 1994, Oreskes, 1997). But the interpretation of a comparison of simulated and measured data is usually not straightforward and often ambiguous. Testable results of physical theories can be established and scrutinized step by

The problem: A new type of smog in the Los Angeles area

The majority of air pollution problems in the nineteenth and large parts of the twentieth century were related to smoke, dust and sulphur emissions from coal burning. A completely new type of air pollution was discovered in the early 1940s in the Los Angeles area. It became apparent by excessive eye irritation, reduced visibility, crop damages and rubber cracking and occurred in warm, high pressure periods (Bartel and Temple, 1952, Haagen-Smit, 1952).5

The case: Constructing the Urban Airshed Model

Seinfeld was fully aware of the difficulties he faced in his attempt to construct a computer-based simulation model for urban photochemical pollution. The involvement of hundreds of chemical substances, among them many complex organic compounds, and an even larger number of reaction pathways, the lack of empirically established rate constants for many of them, the combination of gas-phase, liquid phase and aerosol chemistry, the non-linearity of chemical processes and the additional role of

Fitting practices and the creation of confidence

The development of the Urban Airshed Model (UAM) was an influential pioneering effort. For the first time it linked a Eulerian atmospheric transport model with a chemistry model describing atmospheric photochemistry and ozone formation. With the invention of an artificial chemistry based on lumping chemical species the model builders consciously and explicitly left established chemical knowledge. This approach proved successful, both in terms of model performance and in terms of scientific

References (82)

  • R.J. Ackermann

    Data, instruments, and theory, A dialectical approach to understanding science

    (1985)
  • R. Atkinson

    Kinetics and mechanisms of the gas-phase reactions of the hydroxyl radical with organic compounds in atmospheric conditions

    Chemical Review

    (1986)
  • A.W. Bartel et al.

    Ozone in Los Angeles and surrounding areas

    Industrial and Engineering Chemistry

    (1952)
  • K.H. Becker

    Formation, transport, and control of photochemical oxidants

  • Brienes, M. (1975). The fight against smog in Los Angeles, 1943–1957. Dissertation, University of California,...
  • W.P.L Carter

    Experimental investigation of chamber-dependent radical sources

    International Journal of Chemical Kinetics

    (1982)
  • A. Dahan Dalmedico

    History and epistemology of models: Meteorology as a case-study (1946–63)

    Archive for History of Exact Sciences

    (2001)
  • K.L. Demerjian et al.

    The mechanism of photochemical smog formation

    Advances in Environmental Science and Technology

    (1974)
  • R.G. Derwent et al.

    Photochemical oxidants in the atmosphere. Status report prepared for the Nordic Council of Ministers

    (1991)
  • R.G. Derwent et al.

    Elevated ozone levels in the air of Central London

    Nature

    (1973)
  • Dodge, M. C. (1977), Combined use of modeling techniques and smog chamber data to derive ozone-precursor relationships....
  • M.C. Dodge

    A comparison of three photochemical oxidant mechanisms

    Journal of Geophysical Research

    (1989)
  • P.N. Edwards

    Global climate science, uncertainty and politics: Data-laden models, model-filtered data

    Science as Culture

    (1999)
  • P.N. Edwards

    A vast machine. Computer models, climate data, and the politics of global warming

    (2010)
  • A. Eliassen

    A Lagrangian long-range transport model with atmospheric boundary layer chemistry

    Journal of Applied Meteorology

    (1982)
  • D.M. Elsom

    Atmospheric pollution

    (1992)
  • A.Q. Eschenroeder et al.

    Concept and applications of photochemical smog models

    Advances in Chemistry Series

    (1972)
  • A.H. Falls et al.

    Continued development of a kinetic mechanism for photochemical smog

    Environmental Science and Technology

    (1978)
  • R. Feynman

    "Surely you're joking, Mr. Feynman!" Adventures of a curious character

    (1985)
  • S.K. Friedlander et al.

    A dynamical model of photochemical smog

    Environmental Science and Technology

    (1969)
  • P. Galison

    History, philosophy, and the central metaphor

    Science in Context

    (1988)
  • G. Gonzales

    Urban growth and the politics of air pollution: The establishment of California’s automobiles emission standards

    Polity

    (2002)
  • G. Gonzales

    The politics of air pollution: Urban growth, ecological modernization, and symbolic inclusion

    (2005)
  • G. Gramelsberger

    Computerexperimente, Zum Wandel der Wissenschaft im Zeitalter des Computers

    (2010)
  • Gramelsberger: Conceiving processes in atmospheric models—General equations, subscale parameterizations and...
  • W. Grant

    Autos, Smog and Pollution Control

    (1995)
  • Guillemot: Connections between simulations and observation in climate computer modeling. Scientists’ practices and...
  • A.J. Haagen-Smit

    Chemistry and physiology of Los Angeles smog

    Industrial and Engineering Chemistry

    (1952)
  • A.J. Haagen-Smit

    Atmospheric reactions of pollutants

    Industrial and Engineering Chemistry

    (1956)
  • Z. Haagen-Smit

    Interview by Shirley K. Cohen. Pasadena, California, March 16 and 20, 2000. Oral History Project, California Institute of Technology Archives

    (2003)
  • I. Hacking

    On the Stability of the Laboratory Sciences

    The Journal of Philosophy

    (1988)
  • Cited by (9)

    • Path lumping kinetic model for aqueous phase reforming of sorbitol

      2013, Applied Catalysis A: General
      Citation Excerpt :

      The practice of kinetic models that lumps together complex mechanisms with little knowledge forming artificially small representative compounds or chemical reactions with virtual characteristics dates back to the early 1950s [19,20]. This simplified approach which flourished in oil refining [21], combustion [21,22], air pollution [19], biochemical processes and polymerization [23]. In spite of being simple and associated with “pseudo-information” it was ideally suited for the modelling of fluid flow of complex reactors, real-time control prediction of atmospheric chemistry, optimization and plant monitoring, which otherwise would have been much more restricted by the detailed or molecular-based kinetics.

    • Holism, entrenchment, and the future of climate model pluralism

      2010, Studies in History and Philosophy of Science Part B - Studies in History and Philosophy of Modern Physics
    • Key characteristics of cultures of prediction

      2017, Cultures of Prediction in Atmospheric and Climate Science: Epistemic and Cultural Shifts in Computer-based Modelling and Simulation
    • On the formation-process of the Atmospheric Sciences in the FRG since 1968

      2016, NTM International Journal of History and Ethics of Natural Sciences, Technology and Medicine
    View all citing articles on Scopus
    View full text