Environmental science involves the study of many different kinds of patterns. There are simple patterns such as the response of plant growth to seasonal cues and climate. There are more complicated patterns such as the response of fish to different levels of a toxin in water. There are also complex responses to a host of simultaneously varying factors. All of these, and more, represent patterns that we try to detect, study, understand, and modify. While humans are usually very good at sensing patterns in their environment or seeing trends in data collected by instruments, there are many types of interactions that we have trouble detecting because of either multiple scales of the problem or the complexity of the system. This is a crucial problem because it is exactly these types of patterns (multiscale and complex) that represent the current challenges in environmental science and policy. The challenge is not only how to detect patterns in the environment, but how to get other people to believe that these are significant enough to take action. One example of a crucial pattern is the global rise in temperature. Although many scientists accept as extremely probable future for earth, the challenge is to get governments and the citizens to take action.

There are two approaches to identifying general patterns in the environment. The first is to make an exhaustive list of all the categories of patterns with examples. Any environmental observations can be matched to this list to determine which patterns might be in play. Such a list should help define the differences in behavior between patterns and help the users to consider a broader range of patterns. The use of this list is algorithmic; collect observations, compare the pattern in the observation to the list, and make a selection. The second approach is to develop descriptions of these patterns as a special language. Because this approach is generative it is completely open ended. Both of these approaches are useful in different situations.

A categorical list of environmental patterns

The following is a list of patterns that might be observed in environmental data. The identifying characteristic for each category of pattern is given and, in some cases, critical elements that differentiate this pattern from other similar patterns is supplied. This list is useful when attempting to consider a broader range of possible relationships between environmental factors.

Correlations

The change in one factor in direct relationship to another factor is a common pattern. There are examples of change with distance and time. The increasing depth of a lake away from shore is an example of a correlation between distance and depth. Change with time represents the rate of a process. Cause-effect pairs are also correlations. Correlations are probably the most common pattern reported in Environmental Science data both because they are easy to identify and because the predominant paradigm for analysis is to look for cause and effect relationships.

Correlations can be described and modeled as functions. Factor Y is a function of factor X, often implying that changes in factor X cause the changes in factory Y.

Distributions

Statistical variations in data about a mean is the most recognizable form of a "distribution". This type of variation in the data can be explained with error and probabilities Error distributions and confidence intervals are often part of the process of developing and explaining a correlation between several factors.

other - clusters ??

Distributions are often modeled using probability such as, the measured value of X has a 95% probability of being within a certain range of values.

Cycles

Often factors vary within a range in a cycling pattern. Cyclic patterns are particularly common in environmental data that deal with light periods, seasons, or tides. Sometimes cycles are rapid variations that can be superimposed on the slower trends, such as the case for the atmospheric CO2 concentrations observed at Mauna Loa. This data showed seasonal cycles of CO2 superimposed upon the much longer term increase in CO2 due to anthropogenic input. Some cycles are longer, such as the interannual variability that results from the oscillations of ocean (*****).

Scale

Environmental processes act over a range of time and space scales. Some of these scales are directly sensed and processed by humans but others are not. Patterns that extend across these human scale thresholds can be particularly problematic. For example, humans are very good at telling you the weather for this season, but they are very unreliable for comparing the current season to historical weather. Slow processes that accumulate over a long period of time are difficult to distinguish from the background, however these slow changes may be more important than the season-to-season variability. Carpenter (****) provides good examples of this for lakes and discusses the environmental management consequences of this.

Some patterns are "scale independent", meaning that the processes that govern ???. The most attractive examples of this are fractal patterns. These occur in river sheds, (Bak ****).

Long time and large space scales are studied and described in different language than short term cause and effect studies. This language is considered unscientific. Many authors have noted that the time scale of crucial environmental processes is many times longer than the average Ph.D. dissertation process or research grant funding. Fractal observations are modeled as ratios, which can be visualized by log transforms of the data.

Complex and non-linear

"Linearity" is the assumption that if a 10% change in X causes a 20% change in Y that a 11% change in X will result in a 22% change in Y. Linearity is common assumption, often made with no justification or consideration for potential non-linearity. There are many sets of observations that vary widely or make seemingly unpredictable jumps. These behaviors may be explained by underlying complex of non-linear mechanism. An example of a complex system is a food web in which slight changes in the interactions results in dramatic shifts in the behavior (see ****). These shifts can be described in terms of basins around particular attractors. A slight shift in starting conditions, logic or strength of the interactions can result in a change from one sequence of processes to a totally different set of interactions. These complex systems are described with terms such as basins, point attractors, periodic attractors and chaotic attractors.

Emergent

Emergent behaviors occur when a pattern of action occurs at a higher level of organization of individual agents, and that this action couldn't be predicted or understood by only studying the individual agents. Ant and termite colonies are the archetypical example for emergent behavior; studying individual ants in isolation can't be used to predict the structure of the colony and rich sets of responses that occur with sociality. Another example is the overall patterns of behavior that can result due to extremely simple rules for interactions between spaces or "cells". The spread of a fire in a landscape with an intermediate level of connectedness is a common example. At high levels of connectivity, the whole forest burns; at low levels of connectivity, only individual parcels burn, but at a critical level of connectivity, the fire burns for a long period of time and leads to the creation of a mosaic pattern in the landscape.

Emergent behaviors in natural systems can be described by comparison to the behavior or model systems of multiple agents with simple rules. This is a very different type of description than for other models. It may not even be possible to say that a particular behavior is the consequence of one set of rules, but rather we may have to settle for a description of the process that says that this behavior is one possible instance of a class of behaviors that could be caused by a systems description that is limited to multiple agents and very simple, local rules.

Dissipative structures and self-organized complexity

In many aspects of environmental science it is crucial that we differentiate between non-specific loss of energy and material to the environment (pollution) and the flow of energy of energy and material by mechanisms that maintain the crucial structure of the system itself. An oversimplified yet illustrative example is to compare burning yard debris rather than composting that material. Burning grass and twigs leads to carbon being introduced to the atmosphere, heat being lost and some ashes (containing nutrients) being left over in the burn pit. Composting will release almost as much carbon and heat into the environment, eventually. The breakdown of organic molecules generates heat within the stack which accelerates the microbial reactions. Worms come in from the existing soil, creating a microstructure of channels. Moles and lizards come into the stack to eat the worms and bugs, further changing the microstrucutre of the stack and dispersing the organic molecules. The difference is that the burn pile results in indiscriminate broadcasting of CO2, heat, soot and ashes whereas the composting network uses these same fluxes to build a network. Heat is not just heat and CO2 is not just CO2, the pattern of their dissipation is more important than their eventual fate.

The challenge for us to learn to identify currently existing dissipative structures that are working well and to learn how to deal with systems so as to either encourage self-organization or at least not to disrupt this organization where it exists.

see dissipative-structures.html

The above list of patterns might be considered as a search process. Given a set of observations, what are underlying processes that can be used to explain the pattern in this set? Starting from the simplest approach which is to look for a correlation between a cause and effect and then progressing to explanations of variations from that simple model due to measurement or sampling error. If this still doesn't explain the data, maybe there are underlying rapid or slow cycles or maybe you need to consider different scales of processes. If the pattern is still obscure, maybe you might have to look at complex and non-linear processes. Finally, you might have to consider that a seemingly complicated behavior stems from interactions between many different components of the system rather than a complicated cause-effect relationship for the system as a whole. This isn't a valid search strategy because you may need to consider several patterns and, most importantly, there is no stopping rule for when your search would be complete.

It is better to consider this list as a checklist of patterns that should be considered (see a list of complex patterns list.html). Although most of us are able to easily identify linear correlations, cyclic patterns and understand the importance of variability, it takes different skills to identify the other patterns. These skills are probably best developed by working through the concrete examples that will be provided in the case studies, rather than from study of the underlying mathematical and theoretical formulations.

A generative grammar for environmental patterns

Although having a "check list" of patterns seems tidy and well defined, the world is often messy and ill-defined. Ill-defined in the sense that new observations may fit into multiple categories and could be modeled, with equal value, using very different approaches. For example loss of biodiversity due to habitat destruction can be modeled using linear systems dynamics that demonstrates an equilibrium between loss and immigration, or it can be modeled using cellular automata that approximates habitat fragmentation (Bak ****). Both modeling approaches are based on justifiable mechanisms and describe key features of the behavior. There are slight differences in the predicted threshold for maximum decline in biodiversity that makes this comparison interesting. But the point here is that there is not just one way to describe the behavior of the real system.

The problem we face, as Environmental Scientists, is how to describe patterns within our discipline and to others outside the discipline. This is a general problem faced by many groups, do you use a constrained vocabulary that operates within a defined set of categories, or do you use an adaptable vocabulary that can be applied to new patterns as we discover and study them? The adaptable vocabulary is the most powerful within the discipline but the categories may be the most useful for policy and education purposes. We need to have and use both approaches. In particular, as the adaptable vocabulary identifies important patterns more specifically, we need to include these new terms into the constrained vocabulary/categories.

This discussion of a language of patterns relies heavily on the work of Alexander (1979) and Alexander et al. (1977). Although his focus was architecture, the importance and process of building a language to describe patterns should be the same. Alexander defines a "pattern" as **check quote- page 253** a rule between a context, the system of forces in this context and a configuration that allows these forces to resolve themselves. Each pattern includes the elements of "context", "system of forces" and "configuration". The purpose of the pattern language is to precisely describe patterns, give them a name and share this new word in the language with others. Alexander makes the point that "the expertise in the language". Thus to develop and share this expertise, we need to develop and share a language.

A simple example of a description of the pattern would be:

This pattern could be named "soil erosion and gully formation". What is interesting about this pattern is that as gullies are formed, that causes a new steep edge to be created and further gully's leading into that gully. It has a fractal characteristic. This could lead us to create a new pattern that has to do with how these gullies are arranged on a larger scale.

We could call this pattern "fractal drainage basin" and this pattern depends on the system of forces that causes the gully's Likewise "gully formation" must be part of a larger context that collects and causes water to flow from a source. Alexander (1979) describes how each pattern is part of the pattern of a larger context. The larger context relates to each pattern in a specific manner and *** can't be that pattern without the context and the context is dependent on the component patterns ***.

These two example patterns illustrate the grammar, or structure of these relationships. Each identified pattern has the three elements; context, forces and the outcome. The context may include other patterns, resulting in nested sets of patterns. There is no reason to constrain a pattern to only exist in one context, and thus these sets of context and sub-patterns are not strictly hierarchical.

If a language of patterns were developed for environmental observations, it would improve communication and shared expertise. If no shared language develops and each group or individual even has their constructs their own language, it will loose this larger power. However, even for your individual use, describing all patterns with the same set of elements and making explicit relationships between contextual and sub-patterns should help clarify the value of describing these patterns.

Table 1: Four further examples of patterns that contain all three elements. There can be one or more forces that are resolved through this pattern. These examples are given in a markup language style to show how they could be presented uniformly.

<pattern>

<name>LD50</name>

<context>range of toxic concentrations in water</context>

<force>toxic strength of compound</force>

<force>variability in individuals</force>

<force>bell shaped distribution of variability</force>

<configuration>cumulative toxicity is predicted by the concentration with 50% of the organisms dying at or below the concentration called the "LD50"</configuration>

</pattern>

<pattern>

<name>intraspecific genetic variability</name>

<context>natural selection acting on a population</context>

<force>variable reproductive potential due to a condition or resource </force>

<force>limited population size</force>

<force>genetic processes that lead to variable genetic makeup of progeny in any generation</force>

<configuration>a crucial environmental factor leads to higher relative reproductive rates for some phenotypes/genotypes within a population of fixed size</configuration>

</pattern>

<pattern>

<name>resource limitation</name>

<context>population growth that depletes a particular resource</context>

<force>intrinsic growth rate potential</force>

<force>growth of organisms sequesters resource into unusable form </force>

<force>actual growth rate depends on available resource concentration</force>

<configuration>increased numbers of organisms decreases available resource which results in decreased growth rates as the population size is higher, the logistic growth model is one instance of this</configuration>

</pattern>

<pattern>

<name>demographic transition</name>

<context>population growth in countries</context>

<force>birth rate </force>

<force>death rate </force>

<force>increase health services</force>

<force>industrialization</force>

<configuration>In a very simple view, the demographic transition is a set of steps that countries go through with a decrease in mortality through better health care followed by a decrease in birth rate that is related to (but not cause directly by) industrialization</configuration>

</pattern>

I have started making this type of catalog for patterns that define the interface between human and nature.

please see: pattern-grammar.xml

Internet Explorer displays this xml document in a nested outline form. Safari doesn't display this very well.

Some very important patterns in the environment are difficult to detect

Clear patterns in environmental factors allow us to understand the underlying processes and guide our technological applications and policy decisions. Some of the most important problems that we face, however, aren't marked by clear signals. In fact, ambiguous or cryptic patterns may be the reason why these problems are persistent and difficult to address. This principle is similar to one familiar to aquatic ecologists, if you can measure the nutrient in the water easily, that nutrient is probably not of interest. In this case, if you can identify the factors in an environmental problem easily, it's probably not a very serious problem. A corollary to this statement is that if you think you can describe and solve a serious environmental problem in terms of a single set of factors, you are probably mistaken.The most challenging problems that we face are both complex and have poor alignment between actors values and the benefits from alternative solutions. These are classified as "wicked problems" in which neither simply more study or public awareness will be sufficient to address the problem (see the Institution viewer).

One example of a crucial process that is difficult to detect at early stages is runaway positive feedback. At low values the incremental growth is small, but as the value increases so does the increment in any time and can eventually lead to an explosive growth in the system. In the early stages the positive feedback nature can be hidden in the variability in the data or by overlapping cycles. Global warming is a good example of this type of process. If this is a positive feedback process (such as might be caused by increasing temperature releasing more CO2 from tropical soils or methane from the tundra), it will be much easier and cheaper to take preventative steps now than repairing the damage that is done later. The issue is that we (as environmental scientists) don't know if this is a simple increase or a vicious downward spiral.

Biodiversity loss is another crucial issue facing us. Currently is it generally accepted that most processes are linear, a 1% increase in the causative factor will have a proportional change in the output function. Biodiversity loss may be highly non-linear. There may be a threshold in our level of human disturbance that leads to a rapid and dramatic restructuring of ecosystems and communities to be much simpler. Complex models for this type of shift have been constructed that show there maybe crucial levels of fragmentation that happen at some threshold. Our human burden is how to detect the threshold before we cross it, especially if it is a non-linear response. We may never be able to recover what we lost. One of the favorite metaphors for biodiversity loss, is that we are going to remove some random rivets in your airplane. How many rivets can we remove with no effect and how few would we have to remove after that to have a catastrophic failure of the plane. Although very physical/mechanical, this metaphor illustrates the potential to be near failure without crossing, but that when just one more insult is added to the system there is catastrophe.

A final example of a crucial pattern is the case where the net activity of a system is actually being controlled mainly by emergent behavior of self-organizing independent agents following simple rules but our, human, perception is that the agents are dutifully obeying an externally imposed set of laws and regulation. Some of the most powerful illustrations of this delusion are in common pool resource management. In many cases, a local community has organized and monitored itself to harvest local resources equitably and sustainability (see for example the Canadian fisheries example ***). In this case the higher level government thought the the fisheries needed to be controlled by federal law and superceded local common pool resource institutions ** give some details here ** (Ostrom

Summary

One of the first steps that we take in understanding and responding to the environment is to look for patterns. Because humans are innately good at seeing useful patterns, we might take this activity for granted. Instead of limiting our abilities to primitive innate skills, we need to develop both a broader awareness of different types of patterns. In addition to the usual correlations, distributions, periodic cycles and patterns on different scales, we need to be aware of other patterns including non-linear, complex and emergent behaviors. In order to develop our appreciation of new patterns and the interconnectedness of patterns, we need to use a set of rules (a generative grammar) for describing the elements of patterns. Both of these approaches used together, a comprehensive list and a generative procedure for descriptions, will allow us to elevate the study and consideration of patterns to a more appropriate level in environmental studies.

The crucial problems that need to be addressed our environment turn out to be those that have ambiguous or hidden patterns. Clear patterns would lead provide clear signals for solutions. The three types of solutions to problems described by Wendell Berry (1981) highlight this challenge; some solutions don't solve the problem at all, some solutions just push the problem somewhere else, but the solutions that we need are those that solve the problem in the pattern of its context.