In This Issue
The Vertiginous March of Technology
March 1, 2000 Volume 30 Issue 1

Earth Systems Engineering: The World As Human Artifact

Wednesday, March 1, 2000

Author: Brad R. Allenby

    Managing the Earth’s complex systems and their dynamics is the next great challenge for the engineering profession.

The Earth is increasingly a product of human engineering. Up until very recently, however, this engineering process has occurred without conscious recognition; it consists of the sum of human activities, grown to scales unprecedented in the history of the globe. A myriad of economic and engineering decisions, evaluated and taken as if independent, are in reality tightly coupled both to each other and to underlying natural systems. Anthropogenic climate change; loss of biodiversity and critical habitat; degradation of soil, water, and air resources; dispersion of toxic metals and organics -- these are the fruits of human engineering just as surely as the computer, the automobile, and the highway infrastructure.

In many cases, these effects are unintended, but they are, nevertheless, the direct results of human activity and design. For example, virtually all island communities suffer from invasive species, introduced by humans, and the resulting extinction of indigenous species -- a process that resulted directly from the growth of transportation technologies that accompanied the industrial revolution, the growth of capitalism and global markets, and the expansion of the European culture around the world.

Developing the capability to engineer at the level of global systems -- from energy, transportation, and information systems to the carbon and nitrogen cycles -- is the next great challenge for engineering. This "Earth systems engineering" (ESE) capability will not replace traditional engineering disciplines, but will augment them. However, to address the complexities and dynamics of global systems, it will require a fundamentally different way of engineering (Allenby, 1999a).

We must begin by recognizing a basic, if disconcerting, truth: the Earth, as it now exists, is a human artifact. It reflects the (frequently unintended and unconscious, but nonetheless real) design of a single species. Although this process has been accelerated by the industrial revolution, "natural" and human systems at all scales have in fact been impacting each other and co-evolving together for millennia, and they are now more tightly coupled than ever.

Copper production during the Sung Dynasty, as well as in Athens and the Roman Republic and Empire, is reflected in deposition levels in Greenland ice (Hong et al., 1996). And lead production in ancient Athens, Rome, and medieval Europe is reflected in increases in lead concentration in the sediments of Swedish lakes (Renberg et al., 1994). The buildup of carbon dioxide in the atmosphere began not with the post-World War II growth in consumption of fossil fuel, but with the deforestation of Europe, Africa, and Asia over the past centuries and millennia (Jager and Barry, 1990). Humanity’s impacts on biota, both directly through predation and indirectly through the introduction of new species to indigenous habitats, has been going on for centuries as well (Jablonski, 1991).

What is striking today is the discontinuity between the relatively minor and localized impacts of human activity which predominated before the industrial revolution, and the global, systemic impacts which now characterize the interrelationships between human activity and fundamental biological, physical and chemical systems (Ehrlich and Wilson, 1991; Turner et al., 1990; Science, 1997).

The issue, then, is not whether we should begin ESE, because we have been doing it for a long time, albeit unintentionally. The issue is whether we will assume the ethical responsibility to do ESE rationally and responsibly. The challenges are substantial, and the intellectual tools that conscious ESE will require are not yet available. Some of the requisite capabilities are beginning to develop in the new multidisciplinary field of industrial ecology and the scientific work on global climate change issues, but there must be parallel progress in relevant social sciences, as well as the development of appropriate institutional, ethical, and policy structures. To a large degree it is in these latter areas, heavily imbued with religious and cultural values, where awareness of the challenge is lowest and the barriers to change are most substantial (Allenby, 1999b).

Conceptually, Earth systems engineering may be defined as the study and practice of engineering human technology systems, and related elements of natural systems, in such a way as to provide the required functionality while facilitating the active management of the dynamics of strongly coupled fundamental natural systems. Such fundamental natural systems might include, for example, the grand elemental cycles (e.g., the carbon, nitrogen, and sulfur cycles), critical habitats, and atmospheric or oceanic systems. Minimizing the risk and scale of unplanned or undesirable perturbations in such systems is an obvious ESE objective.

Engineering Human and Natural Systems

It is worth noting that, given the tight coupling between human and natural systems, engineering and managing elements of both might well be required. Thus, for example, response to droughts might include not just hydrological projects, but also bioengineering of salt-tolerant agricultural species. But an important caution must also be borne in mind -- hubris and concomitant premature manipulation of critical natural systems are real potential dangers. If we were not already damaging these complex systems unintentionally, we should not dare to try to engineer them intentionally. But existing impacts and extinctions are real, and the domination of critical systems’ dynamics by humans is real, if unintended. Under these circumstances, it would be irresponsible, verging on unethical, not to begin developing the capability for sophisticated ESE.

Examples of Earth Systems Engineering

Once we recognize that human and natural systems have been coupled in a meaningful way for centuries, it is relatively easy to see examples of Earth systems engineering in the past, and potential examples for the future. The principle differences between the historical and the proposed are two: one of scale and one of intent. In the past, Earth system engineering projects were in general more local, and had less global impact. This, of course, reflects the more local scale of human population levels and activity -- only loosely coupled over regional areas, and uncoupled at the global scale -- that tended to dominate at the time (Turner et al., 1990). Second, to the extent that a set of independently initiated human activities gave rise to "emergent characteristics" at regional or global levels, they tended to be unanticipated and unintended. It is the scale and intent that differentiate ESE from existing engineering disciplines.

Consider, for example, the case of the Aral Sea, a classic example of unintended impacts. Actually a lake, the Aral Sea straddles the borders of Kazakhstan and Uzbekistan. Only decades ago it was the fourth largest lake in the world, but in a few short years it has lost about half its area and some three-fourths of its volume because of the diversion of 94 percent of the flow of two of its feeder rivers, the Amu Darya and the Syr Darya. The purpose of the diversion was primarily to grow cotton, but ironically, the resulting irrigation system is extremely inefficient. Some estimates are that only 30 percent of the diverted water, carried in unlined canals through sandy desert soils, reaches its destination.

The unintended results of this engineering project are staggering -- the desertification of the region (the resulting Ak-kum desert, expected to reach 3 million hectares this year, did not even exist 35 years ago); the generation of some 40 to 150 million tons of toxic dust per year with substantial detrimental impacts on regional agriculture and human health; potential impacts on the climate regimes of China, India, and southeastern Europe; the increased salinization of the Aral Sea which resulted in the loss of 20 of its 24 fish species and a drop in the fish catch from 44,000 tons in the 1950s to zero today (with a concomitant loss of 60,000 jobs); the reduction in nesting bird species in the area from 173 to 38; and possibly the release of biological warfare agents previously contained because they were quarantined on an island (Voskreseniye Island) which is now becoming a peninsula (Postel, 1996; Feshbach, 1996).

The most obvious initial observation is that this case study is archetypal. Such water management activities have been going on for centuries, frequently with unanticipated or undesirable impacts, and often with significant political implications. For example, in ancient China, massive irrigation projects were a signature of, and a major co-evolved institution supporting, the development of highly organized feudal power as far back as the Chin state in the Wei River valley some 2,300 years ago (Needham, 1954). Indeed, such impacts arise not just from individual projects, but from water management regimes taken as a whole, often combined with other anthropogenic activities.

To take an example from North America, data from the Nature Conservancy indicate that, apart from extinctions that have already occurred, 36 percent of freshwater fish species, 38 percent of amphibians, 50 percent of crayfish, and 56 percent of mussel species are in jeopardy in the United States, not usually considered a center of extinction activity. The three main forcing functions for this impact on aquatic biodiversity are agriculture, dams, and exotic species -- all anthropogenic causes (Doyle, 1997).

One area where ESE approaches have begun to evolve, albeit in an ad hoc manner, is in the area of global climate change research and the development of potential geoengineering responses. An example of this genre is the idea of shifting fossil fuel power plants, the perennial "bad guy" contributing to climate change, to become an important component of an engineered carbon cycle management system. The technology involved is conceptually simple: carbon dioxide resulting from fossil fuel combustion in power plants is captured and sequestered for centuries in deep aquifers, the ocean, geologic formations, or other reasonably long-term sinks.

Many of the technologies currently exist to capture the carbon dioxide emissions and inject them into various sinks, and, especially if carbon capture is implemented at the initial design stage rather than retrofitted, such systems appear to be technologically and economically feasible (Socolow, 1997). For example, Norway’s state-owned petroleum company, Statoil, is currently sequestering the carbon dioxide content of the natural gas it is extracting from the Sleipner gas field off the coast of Norway back into an aquifer about 1,000 meters below the seabed. Statoil finds this economically preferable to paying the $55-per-ton tax that would apply if the carbon dioxide were simply vented. Thus, there is already proof of concept, even though a number of issues (e.g., environmental impacts, technological and economic feasibility, liability) remain to be resolved regarding each potential technological option. When combined with an economic structure where end-user energy needs in transportation and buildings are met through the use of hydrogen technologies, such carbon sequestration raises the possibility of being able to exploit fossil fuel reserves without substantial increases in carbon dioxide emissions.

An ESE approach would push further, though, and suggest that carbon sequestration, in combination with hydrogen end-use technologies, could become an important control mechanism in a deliberately engineered human carbon cycle governance system. Here, the global set of fossil fuel plants is tuned to produce over time the desired atmospheric concentration of carbon dioxide, given other variables (e.g., impacts on vegetation, desired degree of global climate change, lag times of various components of the systems involved, changes in solar insulation, other carbon dioxide emissions, concentrations of other greenhouse gases, and the use of other mitigation technologies such as energy efficiency and biomass sequestration).

The control functions of such a system at the facility level are two-fold: the ratio of biomass and municipal waste to fossil fuel input into the system, and the amount of carbon dioxide emitted to that sequestered. At the systems level, the goal would be to move towards, and then maintain, a target atmospheric concentration. Of course, the fossil fuel plant control system would be combined with the entire suite of carbon control methods to accomplish this, and it must not be thought that each component would by any means be as susceptible to engineering control as the fossil fuel plant component. In fact, the system viewed as a whole has critical dynamics scaling over many magnitudes of temporal and spatial systems, with a daunting static and dynamic complexity. Indeed, many components of the carbon cycle system are not susceptible to direct management at all.

Ethical Considerations

ESE requires that human institutions not only accept moral responsibility for human-natural systems, but move beyond to assume an active management role for most global systems. Consider, for example, the approach adopted in the Kyoto negotiations, which relies on reducing global emissions of anthropogenic carbon dioxide and other greenhouse gas emissions to "safe" levels. This is a belated recognition of the ethical responsibility of humans for their effects on global climate, and, given the pervasive interpenetration of human and natural systems, is entirely inadequate from an engineering and management perspective. Accordingly, an ESE approach would require development of an institutional ability to deliberately modulate the carbon cycle within specified and desirable domains. One of these domains could well involve the maintenance of specified atmospheric concentrations of carbon dioxide and other greenhouse gases. However, it would not be thought of in isolation from other elements of the carbon cycle and coupled systems, nor would responses be focused on end-of-pipe emissions controls as they are now; rather, a suite of potential levers to affect systems change would be developed (Keith and Dowlatabadi, 1992).

This raises a fundamental issue: there is an ethical dimension to any consideration of Earth systems engineering, and that is the definition of desired endpoints. ESE is a means to an end which can only be defined in ethical terms. Simply put, the question "To what end are humans engineering, or should humans engineer, the Earth?" is a moral matter, not a technical one. Moreover, as with ESE itself, it is not hypothetical: human institutions are implicitly answering that question every day, and thus positing an answer. To date, however, failure to recognize the strong coupling between human activity and the state of environmental systems has permitted the engineering process to proceed without explicit consideration of the ethical content of the results. At some point this veil of ignorance will be pierced, perhaps in addressing the complex issues raised by global climate change and possible mitigation. But this has not occurred yet.

Some might argue that the idea of sustainable development -- "development that meets the needs of the present without compromising the ability of future generations to meet their own needs" (WCED, 1987) -- establishes that ethical framework, but it may be more likely that it has had the effect of lulling people into a false complacency. There are, in fact, a myriad of potential "sustainable" worlds, depending on the choices made regarding a number of dimensions of global civilization -- that is, choices of a world sustainable with how many individuals, at what level of material well-being, with what level of equity, with what institutional systems, with what variability or stability, for how long, embodying what values (Cohen 1995). In fact, a cynic might argue that a more likely "sustainable world," given human history, is one characterized by economic and political elites protected from environmental and aesthetic insult by wealth and military power, with the political, economic, and cultural sustainability of such a structure maintained by high mortality rates among the poor and low levels of biodiversity.

We must recognize that a significant barrier to any substantial implementation of ESE is the lack of an adequately sophisticated ethical framework. In its absence, social acceptance of ESE and the institutional power it assumes is likely to be minimal, and justifiably so, until the necessary ethical competencies are developed. Unfortunately, it is not apparent where leadership in this dimension will come from, as there are currently no institutions which society has empowered to respond to this difficult challenge. This does not mean that individual ESE projects cannot go forward, particularly where they can be done in line with the guiding principles discussed below, and the perturbation to be addressed is immediate, difficult to reverse, and extensive in time and space. This does mean that the evolution of appropriate ethical and institutional structures is desirable, and should be actively encouraged by those who become interested in ESE. In this instance, failure to be sensitive to the ethical dimensions of ESE projects is, quite simply, bad engineering.

Principles of Earth Systems Engineering

Even given the nascent state of ESE, and the lack of supporting institutional and ethical frameworks, it is possible to generate some preliminary principles that should govern ESE. In doing so, for example, we can draw on previous experience with large engineering projects (especially hydrologic and civil) and complex technological systems such as the U.S. space shuttle program, global air transport control and safety programs, and nuclear power systems in many countries. Without pretending to comprehensiveness, then, here are some obvious ESE principles:

1. Only intervene when required, and to the extent required. The traditional medical axiom, "first, do no harm," is a reflection of humility in the face of complexity which is equally appropriate for Earth systems engineering. In this sense, ESE more reflects a medical diagnosis model than a traditional engineering model.

2. Know what the objectives of any intervention are from the beginning, and establish metrics which can (a) track progress towards satisfying the objectives and (b) provide early warning of unanticipated or problematic system responses. Requiring such preparation for ESE activities is useful not just in itself, but because it implicitly requires that the boundaries of the systems involved be defined, and at least a hypothesized model of systems behavior be developed.

3. Engineering such systems must not be based on implicit or explicit models of centralized control in the traditional rigid sense. Such an approach is appropriate for simple, well-known systems, but not for the complex, unpredictable, and contingent systems involved here. In many cases, these projects will require integrated management of coupled biological, physical, and traditional engineered systems with high levels of uncertainty, and control and feedback mechanisms will be widely distributed along many temporal and spatial scales (NASA has faced some of these issues in its attempts to build space colonies; see NASA, 1979). Rather than attempting to dominate a defined system, the Earth systems engineer will have to see herself or himself as an integral component of the system itself, closely coupled with its evolution and subject to many of its dynamics, with all the self-referential implications. This self-referential pattern is already evident in the global climate change arena, where human decisions about acceptable levels of global climate change forcing affect atmospheric concentrations of carbon dioxide, which in turn feed back into responsive engineering activities (like designing fossil fuel plants to sequester carbon). This will require an entirely different psychology of engineering.

4. Whenever possible, engineered changes should be incremental and reversible, rather than fundamental and irreversible. There must be room for the continuous learning and feedback that incremental engineering interventions support. In all cases, scale-up should allow for the fact that, especially in complex systems, discontinuities and emergent characteristics are the rule, not the exception, as scales change.

5. In a similar vein, experience with existing systems engineering projects demonstrates that the focus of the Earth systems engineer will be on the characteristics and dynamics of the system qua system -- the interfaces, links, and feedback loops among systems components -- rather than just on the constituent artifacts. The focus is on system state, rather than on artifact construction.

6. Continual learning at the personal and institutional level must be built into the process, as is the case now in "high reliability organizations," or HROs, such as aircraft carrier operations or well-run nuclear power plants (Pool, 1997). This learning process is messy and highly multidisciplinary, and accordingly difficult to maintain even in the best of circumstances. It is also problematic because the learning will probably have to occur at an institutional rather than personal level because of the complexity of the systems involved and the inability of any single person, no matter how qualified, to understand them in their entirety.

7. ESE must explicitly accept high levels of uncertainty as endogenous to the engineering function, rather than thinking of engineering as an effort to create a system certain. The mental model must be one of working within complex systems where uncertainty and variability are endemic, not simple systems where it is possible to define system outputs from known inputs unambiguously (Allenby, 1999b). For example, as Morgan and Dowlatabadi (1997) point out, "nonlinear processes that determine climate span roughly 12 orders of magnitude, from microscopic to planetary scales, and it is doubtful whether even future supercomputers will be able to model processes across as much as half that range." These are the types of systems with which Earth systems engineering will have to deal.

8. Similarly, because of the complexity of Earth systems engineering projects, management and organizational skills will be as important to success as traditional engineering skills. Stakeholder management, transparent processes for defining and implementing projects, and managing to social and cultural objectives (as well as scientific and technological ones) will be necessary.

9. An important goal in Earth systems engineering projects should be to support the evolution of resiliency, not just redundancy, in the system. The two are different: a redundant system may have a backup mechanism for a particular subsystem, yet still be subject to difficult-to-predict catastrophic failure; a resilient system will resist degradation and, when necessary, will degrade gracefully, even under unanticipated assaults. Thus, for example, a resilient strategy for global climate change and carbon cycle management would include not just carbon cycle management utilizing a variety of mechanisms (e.g., iron fertilization, carbon sequestration, reforestation), but a broader range of mitigation options as well (Rubin et al., 1992). Should any single option fail or prove to be too expensive, the engineering and management program could adjust gracefully elsewhere using other mechanisms (e.g., energy efficiency and demand side management). That is, even though each engineered option might be subject to failure, the system itself would be engineered and managed to be resilient.

10. Analogously, it is preferable to design (or encourage the evolution of) inherently safe systems, rather than engineered safe systems. An inherently safe system, when it fails, fails in a noncatastrophic way. An engineered safe system is designed to reduce the risk of catastrophic failure, but there is still a finite probability that such a failure may occur. Light-water nuclear power plants, for example, are engineered to be safe, but, as Three Mile Island demonstrated, are not inherently safe. Thus, for example, if one were concerned about the possible failure of underground carbon dioxide sinks and the potential release of substantial amounts of the gas in a lethal manner, one would preferentially use deep geologic aquifers under the ocean, rather than under populated areas. If the former failed, the carbon dioxide would (most probably) be contained in the deep ocean; if the latter failed, the escaping gas might impact human populations.

11. There must be adequate resources available to support both the project and the science and technology research and development which will be necessary to ensure that the responses of the relevant systems are understood. Financial pressures can be particularly insidious with complex engineering technologies even today (Pool [1997] cites the Bhopal Union Carbide chemical plant and the Challenger incident as examples where pressures generated in part as a result of chronic underfunding resulted in catastrophic failure of such systems). Earth systems engineering projects are likely to be at least as complex as existing technological systems, and to last over longer time periods than the usual budgetary cycles, meaning that they may be particularly prone to financial fluctuations.

12. If any Earth systems engineering project is to achieve public acceptance and social legitimacy, it must at all stages be characterized by an inclusive dialogue among all stakeholders. Not all of them will agree, for a number of reasons, but to be successful, a project requires broad public support. The most obvious example of a complex technology system where these mechanisms have proven inadequate is, of course, civilian nuclear power. In the United States, for example, the secrecy and technological hubris which grew out of the nuclear weapons program and the nuclear Navy meant that both the regulators and the experts were culturally adverse to open communication and dialogue, with eventual results for the industry that were both predictable and disastrous (Pool, 1997).


Earth systems engineering may be defined as the study and practice of engineering human technology systems and related elements of natural systems in such a way as to provide the required functionality while facilitating the active management of the dynamics of, and minimizing the risk and scale of unplanned or undesirable perturbations in, strongly coupled fundamental natural systems. ESE involves developing technologies and strategies for managing complex coupled human-natural systems at very different scales, from highly granular to broadly integrated, in a comprehensive manner. This new area of engineering can draw on significant experience from past practices and results, particularly derived from projects involving complex technology systems or large natural systems (e.g., irrigation), and can draw much support from the evolving field of industrial ecology. Even so, it is apparent that the science and technology and the institutional and ethical infrastructures to support such approaches do not yet exist, although progress in relevant disciplines and technologies has been made on a piecemeal basis. This, and the fact that unanticipated consequences of such major programs could be quite significant, argues for a cautious and balanced effort to begin such a program. Fundamentally, however, the question is not whether humans are engineering Earth systems, but whether humans will do so rationally, intelligently, and ethically.


Allenby, B. R. 1999a. Earth systems engineering: The role of industrial ecology in an engineered world. Journal of Industrial Ecology 2(3):73-93.

Allenby, B. R. 1999b. Industrial Ecology: Policy Framework and Implementation. Upper Saddle River, N.J.: Prentice-Hall.

Cohen, J. E. 1995. How Many People Can the Earth Support? New York: W. W. Norton & Company.

Doyle, R. 1997. Freshwater fish at risk in the U.S. Scientific American (December):32.

Ehrlich, P. R., and E. O. Wilson. 1991. Biodiversity studies: Science and policy. Science 253:758-762.

Feshbach, M. 1996. Former Soviet Union’s environmental and health problems. Pp. 173-188 in Environmental Threats and National Security: An International Challenge to Science and Technology, B. R. Allenby, T. J. Gilmartin, and R. F. Lehman II, eds. Livermore, Calif.: Lawrence Livermore National Laboratory.

Hong, S., J. Candelone, C. C. Patterson, and C. F. Boutron. 1996. History of ancient copper smelting pollution during Roman and medieval times recorded in Greenland ice. Science 272:246-249.

Jablonski, D. 1991. Extinctions: A paleontological perspective. Science 253:754-757.

Jager, J. and R. G. Barry. 1990. Climate. Pp. 335-352 in The Earth as Transformed by Human Action, B. L. Turner, W. C. Clark, R. W. Kates, J. F. Richards, J. T. Mathews, and W. B. Meyer, eds. Cambridge: Cambridge University Press.

Keith, D. W., and H. Dowlatabadi. 1992. A serious look at geoengineering. Eos 73(27):289, 292-293.

Morgan, M. G., and H. Dowlatabadi. 1997. Energy technology R&D essential to curb global warming. Environmental Science and Technology 31(12):574A-575A.

NASA. 1979. Space Resources and Space Settlements (NASA SP-428). Washington, D.C.: NASA.

Needham, J. 1954. Science and Civilization in China, Volume 1: Introductory Orientations. Cambridge: Cambridge University Press.

Pool, R. 1997. Beyond Engineering: How Society Shapes Technology. Oxford: Oxford University Press.

Postel, S. 1996. Forging a sustainable water strategy. Pp. 42-59 in State of the World 1996, L. Starke, ed. New York: W. W. Norton & Company.

Renberg, I., M. W. Persson, and O. Emteryd. 1994. Pre-industrial atmospheric lead contamination in Swedish lake sediments. Nature 363:323-326.

Rubin, E. S., R. N. Cooper, R. A. Frosch, T. H. Lee, G. Marland, A. H. Rosenfeld, D. D. Stein. 1992. Realistic mitigation options for global warming. Science 257:148-149, 261-266.

Science 277:485-525. 1997. Special report, Human-dominated ecosystems.

Socolow, R., ed. 1997. Fuel Decarbonization and Carbon Sequestration: Report of a Workshop (Report No. 302). Princeton, N.J.: Princeton University Center for Energy and Environmental Studies.

Turner, B. L., W. C. Clark, R. W. Kates, J. F. Richards, J. T. Mathews, and W. B. Meyer, eds. 1990. The Earth as Transformed by Human Action. Cambridge: Cambridge University Press.

WCED (World Commission on Environment and Development). 1987. Our Common Future. Oxford: Oxford University Press.

Note: The opinions expressed in this paper are those of the author only, and do not necessarily represent those of AT&T, Columbia University, or any other entity with which the author has been or is now affiliated.

About the Author:Brad Allenby is vice president of environment, health and safety at AT&T and adjunct professor, Columbia University School of International and Public Affairs.