W1: Workshop on Environmental Monitoring & Adaptive Management

organised by John Norton and Kenneth Reckhow

Abstracts

Title: Modeling in Environmental Sciences: Evaluation of the Current State and Future Perspectives

Authors: George Arhonditsis

Abstract: The veracity of the scientific methodology of numerical models and their adequacy for forming the basis of public policy decisions has frequently been challenged. A recent evaluation of the current state of mechanistic aquatic biogeochemical modeling indicated that there is still considerable controversy amongst model developers and resource managers about how to develop, evaluate, and interpret numerical models. The same analysis provided overwhelming evidence that ocean modeling is the most vibrant area of the current modeling practice. Models developed to gain insight into the ocean carbon cycle/marine biogeochemistry are most highly cited, the produced knowledge is exported to other cognitive disciplines, oceanic modelers are less reluctant to embrace technical advances (e.g., assimilation schemes) and test new ecological theories. Contrary to our predictions, model application for environmental management issues seems to have languished; the pertinent papers comprise a smaller portion of the published modeling literature and receive lower citations. Given the critical planning information that these models aim to provide, we hypothesize that the latter finding probably stems from conceptual weaknesses, methodological omissions, and an evident lack of haste from modelers to adopt new ideas in their repertoire when addressing environmental management issues. Striving for novel modeling tools, Bayesian calibration of process-based models is a methodological advancement that warrants consideration in environmental research. This modeling framework combines the advantageous features of both mechanistic and statistical approaches, i.e., mechanistic understanding that remains within the bounds of data-based parameter estimation. The incorporation of mechanism improves confidence in predictions made for a variety of conditions, while the statistical methods provide an empirical basis for parameter selection and allow for realistic estimates of predictive uncertainty. Other benefits of the Bayesian approach is the ability to sequentially update beliefs as new knowledge is available, and the consistency with the scientific process of progressive learning and the policy practice of adaptive management.


Title: Village Level Tactics for the Control of Parasitic Disease: An Example at the Intersection of Control Engineering and Bayesian Analysis

Authors: Robert Spear

Abstract: For the last decade our group has pioneered the application of dynamic models of disease transmission to design village-specific controls for the suppression of the parasitic disease schistosomiasis in the mountainous regions of China. The transmission model has served as a platform for the integration of knowledge of the biological mechanisms involved, data from the literature on parameter values controlling the rates of these processes, and a range of site-specific information. We have conducted large-scale field studies to obtain this site-specific data. A constant theme has been the reduction of the uncertainty of the model predictions over time. In that context, our general approach has been that of Regional Sensitivity Analysis that can be considered a specific example of the more general statistical approach termed Bayesian Melding. To date our experience has shown that the modeling approach, as an adjunct to ongoing and extensive field data collection efforts, has been very productive. It has demonstrated: the limitations of the nature and frequency of collection of field data normally collected in rural China, the need for new methods of environmental monitoring for the presence of the parasite the need for new analytical methods for separating the effects of site-specific parameters from those that can be assumed to be at least regionally invariant, and the need for sustainable environmental modifications in these villages in addition to the usual modalities of suppressing transmission advocated by the World Health Organization and currently followed by the Chinese authorities. In a more general context, our work has motivated a new conceptualization for incorporating the impact of environmental variables on the transmission processes of this class of infectious diseases.


Title: Developing Monitoring Plans and Forecasting Models for Adaptive Implementation of Shellfishing Resource Areas

Authors: Andrew Gronewold, Kenneth Reckhow, Robert Wolpert

Abstract: Water quality monitoring programs which support shellfishing resource area management decisions focus on preventing human consumption of contaminated shellfish and minimizing associated disease outbreaks. Management decisions primarily include either opening or closing resource areas, and closure decisions are often based on the historic relationship between rainfall events and subsequent increased pathogen concentrations in receiving waters. Water quality monitoring program results are therefore used mainly as a confirmatory rather than a predictive tool, and shellfishing resource area management decisions are limited to a temporal scale no greater than the time period between rain events. In addition, shellfishing resource area monitoring plans, while providing short-term protection of human health, are not designed, either as a stand-alone tool or in combination with predictive models, to evaluate the effectiveness of closure decision criteria. EPA's total maximum daily load (TMDL) program, in contrast, focuses on long-term restoration of impaired water bodies and requires the development of monitoring plans and forecasting tools which can simultaneously and iteratively achieve multiple management objectives. This paper explores the development and implementation of adaptive management plans for shellfishing resource areas through a two-fold approach addressing both short-term protection of human health and long-term restoration of contaminated natural resource areas. Resulting plans and modelling tools attempt to integrate monitoring data from a broad range of spatial and temporal scales and not only improve short-term and long-term resource management capabilities, but also provide suggestions for monitoring plan and model improvements.


Title: Managing a Saline Catchment Exhibiting Multiple Stable States and Thresholds

Authors: Tim Peterson

Abstract: The Murray-Darling Basin Commission, Australia, recently introduced a framework for capping the export of salt from dryland catchments into the Murray River. The framework set the establishment of End-of-Valley Salinity Targets (EoVT) for each of the major catchments within the Basin, thus establishing accountability to a local region for its in-stream salt load exports. Effectively it is a negative feedback control process of observation of stream salinity, prediction of current and future loads and control, via within region landuse change. This discussion focuses on the viability of this command-and-control framework for the Goulburn catchment in the context of recent work suggesting the catchment has multiple-coexistent stable states and thus questionable observability. The Goulburn catchment (excluding the Goulburn Plains), situated in north central Victoria, is 13,161 km2 of which 62% is cleared and utilised for grazing. The mean stream salt load is approximately 206,000 tonnes/year or 18 tonnes/km2. The EoVT for the catchment has been set to 100% of this current load. The regions salinity management plan estimates this requires 2,100 km2 of currently grazed land (i.e. 32% grazing area) to be revegetated with high density trees, totally AU$ 540 M (Goulburn Broken Catchment Management Authority 2002). The adopted policy of the management authority is to achieve only 50% of the target through revegetation of 50% of the 2,100 km2. The management strategy states that it is founded upon the principles of Adaptive Management (AM) in that it believes it acknowledges the uncertainties in such goals and thus proposes a five plus year cycle of implementation and monitoring followed by a review of the plan. In a review of salinity management plans in the region Allan et al. (2002) concluded that the proposed adaptive management is only passive AM (pAM), rather than active AM in which policies are designed and viewed as experiments to accelerate learning (Holling et al. 1978; Walters 1986). The salinity plans do not implement such structured experiments and are very much an ad-hoc sequential learning and decision making process. They also found that management assumes rational decisions will lead to expect outcomes. This somewhat command-and-control structure is very questionable in light of the limited observability of the catchment salinity process and recently proposed positive biophysical feedbacks producing multiple co-existent states. The concept of coexistent states within biophysical systems emerged from the ideas of ecosystem resilience. Often such states are characterised as socially desirable and undesirable, with, for instance high/low fish stocks or oligotrophic/eutrophic lake states. Their significance is that small stochastic forces may force the system over a threshold toward an alternative, potentially undesirable, state. Simple ordinary differential equations models have been developed to numerically predict the existence of such multiple stable states and quantify resilience. The first catchment-resilience model was recently developed by Anderies (2005) and expanded by others (Peterson et al. 2005). At each location it quantifies the cumulative disturbance required to shift Goulburn catchment to a near zero watertable depth as a function of the percentage land cleared. That is, it provides a means to estimate the sensitivity of outcomes to disturbances through quantification of the disturbance required to produce a state change. Once a threshold has been crossed different positive feedbacks dominate, making a return to the previous state questionable. The observation of this threshold crossing in the catchment data will also become apparent only once the system has approached the new stable state. Current salinity management in light of these proposed coexistent states is considered and found to be questionable. Most notable is the observability of a threshold and the uncertain controllability even if such a threshold was foreseen. Lags in either decision making or implementation of mitigation options would further confound the outcome and potentially induce long term cyclical biophysical. In light of this and the numerous well documented challenges to active AM, options for a more active form of passive AM which focuses on management of resilience is also considered.


Title: Adaptive Management in Complex Systems

Authors: Craig Stow

Abstract: The idea of adaptive management arose in the 1970s when directed institutional decision-making to achieve specific environmental outcomes was a relatively new endeavor. The concept is premised on the recognition that that ecosystems behave idiosyncratically, are imperfect replicates of one-another, and are not well-understood by reductionist approaches. Consequently, forecasting the outcome of alternative management actions would carry considerable uncertainty. The adaptive management concept emphasizes that environmental management is a process and that management actions should be regarded as ecosystem-scale experiments providing an opportunity to learn more about ecosystem behavior and refine ongoing management actions to incorporate accumulating knowledge. A concurrent outlook in the 1970s was that mathematical models would eventually be capable of delivering precise ecological forecasts as our increasing understanding of ecosystem processes was reflected in progressively more accurate models. This belief was implicit in many environmental laws and regulations crafted during that time. However, in the intervening years, considerable experience and an increasing appreciation of ecosystems as complex adaptive systems has reinforced the notion that the precise prediction is infeasible. Thus, adaptive management has assumed new relevance and the development of fast, cheap computers is providing a new opportunity to merge the principles of adaptive management with the development of mathematical models for improved forecasting and management. Exploiting this opportunity requires embracing the concept that environmental management is a continuing process - more closely analogous to maintenance than repair. It also invites the recognition that disturbed ecosystems are more likely to experience reorganization rather than recovery, when anthropogenic stressors are reduced, with consequences that are unlikely to be fully anticipated.


Title: Updating of lake respiration rate and adapting of artificial oxygenation efficiency that comply with the oxygen standard using Markov chain Monte Carlo (MCMC) method

Authors: Olli Malve, Marko Laine, Heikki Haario

Abstract: Respiration rate in a hyper eutrophic lake has been estimated using non-linear dynamic model and Markov chain Monte Carlo (MCMC) method (Adaptive Metropolis-Hastings algorithm). Oxygen regime in the future winters have been predicted using the posterior information of lake respiration coming from the past observations. Using new observations the respiration rate has been updated and needed artificial oxygenation efficiency that comply with the oxygen standard with the given margin of safety has been adapted accordingly. Benefits of adaptive decision strategy in water quality management has been highlighted and discussed in details. Presented example reveals clearly the usefulness and the flexibility of MCMC method in adaptive management of environmental systems. MCMC allows for analysis and predictions that take into account all the uncertainties in the model and in the data, pools information from different sources, and quantifies the uncertainties using full statistical approach.


Title: Modelling and monitoring environmental outcomes in adaptive management

Authors: John Norton, Kenneth Reckhow

Abstract: The main aim of this paper is to stimulate questions about the future of adaptive management (AM) of natural resources, and more specifically about what approaches may be feasible which have not yet been explored well. The method will be to compare the histories, strengths and limitations of AM, control engineering and Bayesian analysis, which have superficial similarities, significant differences and perhaps lessons for each other. Some difficulties encountered by control engineering and complex environmental models are pointed out.