Ch1 p.1

Environmental Modeling and Decision Support Systems

WS 2014/15

Prof. Dr. Struss, Dr. Dressler

Chapter 1. The Topic

The Topic

What is Eco-Informatics?

In simple terms, Eco-informatics can be defined as an application of computer science to the particular domain, namely ecology, likewise bio-informatics, medicine informatics etc. are also considered to be an application of computer science to the respective domains. It uses the same methods and techniques, such as DB technology, simulation, image analysis, and problem solving techniques for analysis, prediction, decision making, control and management

Now the question is that why is it justified to call Eco-informatics a field in itself and teach a specific course? While for applications in ecology, standard IT techniques are used and most importantly ecology implies some specific challenges to IT because of its complex interactions, and dynamic nature and processes,

The goal of the first chapter is to highlight these special challenges, based on a discussion of a number of case studies.

Ecology in Practice

To discover the real problems faced by ecologists and how they try to solve them, it is best to consider some real problems in a little detail. While reading the following examples you should focus on how they illuminate three main points (i) ecological phenomena occur at a variety of scales; (ii) ecological evidence comes from a variety of different sources; and (iii) it relies on truly scientific evidence 2

Example: Impact of Introduction of Trout in New Zealand

Example 1: Impact of the introduction of trout in New Zealand

Trout have been introduced to New Zealand rivers long time ago. Trout compete with native fish “Galaxias”. Both of them feed on invertebrates. After some observations, we classify river sections into 4 types: sections with no fish at all, sections with trout only,

Ch1 p.2

sections with Galaxia only and sections with both species. A field study has been carried out to investigate the details and impact of the situation.

Counting visible invertebrates

Mayfly nymphs are invertebrates, which commonly graze microscopic algae growing on the beds of New Zealand streams, but there are significant differences in their activity rhythms depending on whether they are in Galaxias or trout streams. In one experiment, nymphs collected from a trout stream and placed in artificial laboratory channels were less active during the day than the night, where are those collected from Galaxias streams were active both day and night, as shown in Figure1.1.

In another experiment with another invertebrate Deleatidium mayfly records were made of individuals visible in daylight in artificial channels. Daytime activity was significantly reduced in the presence of either of species, but to a greater extent when trout were present. Figure 1.2 shows different ways of locating prey. The explanation for the different patterns of activity: trout locates prey visually and Galaxias locate prey mechanically.

Figure 1.2 Deleatidium visible of locating prey

Figure 1.1 Nesameletus visible of Galaxias and Trout in stream

Ch1 p.3

Abundance of the fish species

When trout try to migrate upstream, they will be prevented by waterfalls. There exists some correlation with elevation. Fig. 1.3 shows means and, in brackets, standard errors for important discriminating variables for site classes. In particular, compare the 'Galaxias only' and 'brown trout only' classes.

Figure 1.3 Migration correlated with elevation

Impact on invertebrates and algae

Trout reduces population of invertebrates and, thus, causes an increase in biomass of algae. Figure 1.4, shows biomass of invertebrates and algae for stream sections with trout only (T), Galaxias only (G) or no fish (N).

Figure 1.4 Biomass of invertebrates and algae

Ch1 p.4

Biomass- production and demand

Figure1.5 shows the number of production and demand w.r.t. algae, invertebrates and fish for trout streams and Galaxias streams.

Figure 1.5 Production and demand varies in different environment

The cite case study focused on understanding the impact of the two fish species and their interaction w.r.t. invertebrates and the algae biomass. However, this is not the end of the actual impact. For instance, algae feed on nitrate, ammonium and sulfate. Hence, an increase of algae will reduce the concentration of nitrate, ammonium and sulfate in the water and may affect the concentration not only locally, but also downstream. The reduced concentration of these substances can have a non-negligible effect on organisms downstream. And this could have further consequences…

Conclusion: The chain of causal effects is unbounded in principle, and any limit is not intrinsic but determined by the focus of our interest and/or our available knowledge and information.

Ecology : Definitions and Basic Concepts

One definition of Ecology

“The scientific study of the distribution and abundance of organisms and the interactions that determine distribution and abundance”- (Townsend et al.08)

Ecology was first defined in 1866 by Ernst Haeckel, a German biologist as 'the comprehensive science of the relationship of the organism to the environment'.

Ch1 p.5

Scales of operation

Ecology operates at a range of scales:

- Different levels of Interactions - time scales, and - spatial scales

It is important to appreciate the breadth of these and how they relate to each other.

The Level of Interactions

. Ecology deals with the four levels of interactions:

individual organisms: how do individuals behave, feed, suffer from diseases, etc.;

populations (consisting of individual of the same species): how do individuals of a species interact with each other, mate, compete, etc.;

Communities (consisting of a greater or lesser number of populations of different species): their structure, composition, and interdependencies of different species, such as competing for resources, predation, etc.

ecosystems (comprising the community together with its physical environment)

The following Figures show these levels examples.

3Levels of interaction - Individual

Individuals identified by their common genetic makeup, behavior, physical characteristics and ability to breed with one another.

Figure 1.7a Individual

Levels of interaction - Population

A population is the set of individuals of a particular species within a given area, which will be defined within some kind of physio-graphic region such as a mountain range or contiguous habitat, and not by political boundaries.

Ch1 p.6

Figure 1.7b Population

Levels of interaction - Community

It consists of all the living organisms (the biota) in a given area and their interrelationships. Understanding the relationships between competitors, predators, prey, diseases, food supply etc. can shed light on the cause-effect relationships influencing population distribution and abundance.

Figure 1.7c Community

Levels of interaction - Ecosystem

It consists of both the biotic (living) and abiotic (non-living) components of a habitat and their interrelationships. Ecosystems form the working units of nature in which populations and communities work in balance with one another and with the non-living environment.

Ch1 p.7

Figure 1.7d Ecosystem

Different spatial scales

Within the living world, there is no arena too small nor one so large that it does not have an ecology. Even the popular media talk increasingly about the 'global ecosystem', and there is no question that several ecological problems can only be examined at this very large scale. These include the relationships between ocean currents and fisheries, or between climate patterns and the distribution of deserts and tropical rain forests, or between elevated carbon dioxide in the atmosphere and global climate change.

At the opposite extreme, an individual cell may be the stage on which two populations of pathogens compete with one another for the resources that the cell provides. At slightly larger spatial scale, a termite's gut is the habitat for bacteria, protozoans and other species, a community whose diversity is comparable to that of a tropical rain forest in terms of the richness of organisms living there.

Different temporal scales

One may study the deposition of a lump of sheep dung to its decomposition (a matter of weeks). Or the ecological succession - the successive and continuous colonization of a site by certain populations, accompanied by the extinction of others change – caused by climate changes from the end from the last ice age to the present day and beyond (around 14,000 years and still counting). Migration may be studied for butterflies over the course of days, or in the forest trees that are still (slowly) migrating into de-glaciated areas following that last ice age.

Ch1 p.8

Different sciences and knowledge sources

Ecological evidence comes from a variety of different sources. Observations and field experiment, laboratory experiments, simple laboratory systems and mathematical models requires different sciences and knowledge sources in the field study, like biology, chemistry, physics geophysics and hydrology.

Trout field study – aspects

As an exercise: to understand concepts explained above, consider trout field study and try to find out which levels do we need to consider regarding spatial aspects, temporal aspect, and which disciplines are involved?

Ecology: Tasks and Goals

Ecology - goals

Keep in mind the definition of ecology as “The scientific study of the distribution and abundance of organisms and the interactions that determine distribution and abundance”. Which steps does this comprise?

Firstly, we need to describe the relevant phenomena; it is required to characterize a problem in terms of observations and measurements of certain properties and quantities of the subject of inquiry, provide definitions of them etc. Based on this, we try to suggest an explanation of a phenomenon, by proposing a possible relationship among a set of known phenomena that would entail the phenomenon in question, i.e. a model. There could be different model hypotheses.

Such a model is useful only if it enables predictions, by reasoning including deductive reasoning. It might predict the outcome of an experiment in a laboratory setting or the observation of a phenomenon in nature. Prediction does not only mean inferring future development of a system, but also the restriction of unobserved (or even unobservable) properties, hidden states, etc.

Figure 1.8

Ch1 p.9

Understand in order to influence

But even understanding is not a purpose in itself. The important motivation for this study is to influence a system pursuing a particular objective, e.g. to limit a bad impact of human activity, secure continued exploitation and solve environmental problems.

In the overall process, we get basic description at the very beginning and try to explain why the situation occurs, then try to make predictions for further study and verify or refute certain hypotheses. Based on a hypothesis that has some evidence, we can try to manage or control a situation by prediction and again describe new situation.

Figure 1.9

Example: Degradation of Mangroves in India

Optimism – “we will preserve local flora and fauna”

Figure 1.10

The following example stems from a case study in South India, attempting to counteract the degradation of mangrove forests in the delta of the Kavery River. The sign reads “In

Ch1 p.10

this area the forest department of the Pichavaram Mangroves has started management activities in 1995 in order to preserve the local flora and fauna”, but the environment shows that the initial attempts to simply plant new mangroves were not very successful. Why? Because the reasons for dying mangrove forests was not understood.

Meanwhile, Upstream...

The investigations identified the ultimate reason: the building of dams far upstream the river. They caused a change in the flooding and sedimentation processes, which resulted in reduced deposition of sediments in the delta. This in turn modified the profile of the delta in a way that turned the slopes into an area of basins, in which water would be captured even during low tide (rather drained to the sea). Evaporation caused an increase of salinity of the water beyond the level tolerable to the mangroves, which started to die. It is important to notice, that the mangrove forests not only provide a significant contribution to local life and economy (fishing, fire wood), but also an effective shelter against the impact of severe storms and tsunamis. Indeed, fatal consequences of such incidents to a significant degree are due to the elimination of mangrove forests.

Only after identifying this causal explanation, management efforts, based on creating channels to drain the area during low tide and, thus, decrease the salinity, became effective.

Figure 1.11 chain reaction

“Side-effects”…

The lesson to be learned from this example: human activity besides its immediate objectives (building dams in order to control flooding and generate electricity) has further impact, which we to call “side-effects”, because they are not the intended ones

Ch1 p.11

and perhaps not anticipated. This term is misleading, because it refers only to human ignorance. In nature, there are no “main effects” and “side effects”, but only causes and effects. We need to understand as much as possible of the natural cause-effect network in order to avoid negative consequences of human intervention in eco-systems or, in case they already occurred to counteract and mitigate them.

Figure 1.12 Side-effects

Humans and “Environment”

“Environment”??...

This term is already flawed, because it characterizes the world only from the perspective of human activity, i.e. from an anthropocentric perspective.

Figure 1.13 Environment

Ch1 p.12

Anthropocentric perspective

Anthropocentrism is either the belief that humans are the central and most significant entities in the universe, or the assessment of reality through an exclusively human perspective. Anthropocentrism, or human-centeredness, is believed by some to be the central problematic concept in environmental philosophy, where it is used to draw attention to a systematic bias in traditional Western attitudes to the non-human world.

The anthropocentric view suggests that humans have greater intrinsic value than other species. A result of this attitude is that any species that are of potential use to humans can be a "resource" to be exploited. This use often occurs in an unsustainable fashion that results in degradation, sometimes to the point of extinction of the biological resource, as has occurred with the dodo, great auk, and other animals4.

Figure 1.14 Anthropocentric perspective

Ch1 p.13

“Side-effects”…

Figure 1.15 Side-effects

What we call “environment” is actually the world, which includes humans. We need to understand the complex interactions of organisms and natural phenomena and systems in the world and to view Human activities as additional influences in this network of interactions.

Causal chains – distant and paradox effects

Exercise: Consider some links between causes and (distant) effects and try to refine them:

- Building dams result in more tsunami victims. - Introduce trout will produce more algae in the environment. - Extinguishing forest fires cause more trees and homes destroyed by fire. - Extinguishing fires in Sequoia forest might cause the extinction of Sequoias. - Treating cattle with Diclophenac (medicine against inflammation) will result in

more diseases of people and difficult burial of dead Parsis.

Exercise

Trout field study – exercise

Exercise: Design a semi-formal or diagrammatic representation that describes and explains the impact of introduction of trout in New Zealand. The solution presented for this exercise was essentially a graph that contained, in particular, the relevant species as nodes, the links between them labeled with "feeds on", "reduces", or "requires". Other interactions would represent activities, such as migration of trout and the impact

Ch1 p.14

of obstacles (waterfalls) etc. For explanatory purposes, there will usually be no equations or numbers required (and, often, not possible) but links (arrows) that represents and directs a causal relationship.

Conclusion: Numerical information and formal mathematical expressions play no or only a secondary role in a model to support our understanding. The crucial aspect of such a model lies at a conceptual level and describes the involved entities and their interrelationships. This layer is not present in almost all existing modeling formalisms and tools. The task requires methods from knowledge representation.

Challenges for IT

How can IT help?

Ecological modeling is concerned with the development of models of the relationships among members of communities and between these communities and their abiotic environment. These models can then be used to better understand the domain at hand or to predict the behavior of the studied communities, thus, support decision making for environmental management. Typical modeling topics are population dynamics of several interacting species and habitat suitability for a given species (or higher taxonomic unit).5

In order to understand the role of IT and assess its current contribution, we extend our view of the basic tasks: Beyond the describe-explain-predict-manage-describe cycle, we need to plan observation and collecting data. We can use IT to plan experiments and do field studies in order to gain information that helps to refute hypotheses, establish new concepts and models that help to pursue the efforts based on a better foundation.

Figure 1.16 Help with IT

Ch1 p.15

How can IT help? -1 Observation

Figure 1.17 Observe

IT support: Collecting Data

There are different relevant and established ways of collecting data; such as remote sensing is one of the popular way for collectingsatellite data.

The spatial dimension of the data is often crucial, and the volume of spatial data is huge. Another potential source of problem is the necessity for data with long-time ranges.

Before you start collecting environmental data determine the following:

1. What kind of environmental issues are you interested in? E.g. traffic, water, noise or light pollution.

2. Which measurements or data you need to collect? E.g. Counting the number of cars in your area; Using the logbooks to sense the amount carbon monoxide in your school car park Or Measuring the levels of noise in your local area.

3. How to use the equipment that will help to collect your data.7

Some examples of Environmental and Ecological Data can be found here

IT Support: Storing and Retrieving Data

Database technology is used for storing and retrieving data. For instance, use of a GIS (geographical information system). Representation of spatial information is a challenge. Spatial requires formal techniques representing topological, geometric, or geographic properties. A geographic information system (GIS) is a system that captures, stores, analyzes, manages, and presents data that are linked to location. It is

Ch1 p.16

the merging of cartography and database technology. Common errors often arise in spatial analysis, some due to the mathematics of space, some due to the particular ways data are presented spatially. Data is distributed in non-homogeneous database systems, to integrate them or process to make a single source, is a challenge.

How can IT help? – 2

Figure 1.18 Describe and Explain

IT support: Analyzing and Interpreting Data

Data analysis and interpretation is the process of assigning meaning to the collected information and determining the conclusions, significance, and implications of the findings.8

We can use statistical analysis and image processing and analysis (for example, to determine vegetation coverage from satellite data). There are some challenges when analyzing and interpreting data:

9Plenty of data.

Spatial relationships are important but difficult to measure.

Inherent uncertainty due to scale.

Difficult to make data sources compatible.

Quantity vs. Quality Questions.

Firstly, huge data volumes need to be analyzed and interpreted. But the essential difficulty is grasping the meaning of data. E.g. image processing is not enough; image understanding is required to serve the purpose.

Ch1 p.17

How can IT help? – 3

Figure 1.19 Predict

IT Support: Prediction

Today, mainly numerical models and simulations are used in the prediction process. But this faces faces some challenges, especially of the complexity due to various levels of interactions and many different aspects (partial models) under consideration. In ecology, often our understanding is only partial; it lacks precise parameters, and involves non-numerical data, information, and knowledge. For instance, it is hard to represent a phenomenon like degradation of mangrove forests. Therefore, what we need is a conceptual modeling, which includes concepts like causality, and provides explanation and supports causal understanding.

Numerical models usually do not include a description of the model boundaries (because they are tied to conceptual aspects) and cannot characterize their scope of applicability.

Better support to model development is necessary, but requires explicit representation of the conceptual layer of models, that is usually only present in the heads of the modelers. In a nutshell, modeling serves as a knowledge representation.

How can IT help? – 4

Figure 1.20 Manage, control and plan

Ch1 p.18

IT Support: Drawing conclusions and taking decisions

Environmental decision support systems are introduced for drawing conclusions and taking decisions based on ecological data and knowledge. A DSS (Decision Support System) is an interactive software-based system intended to help decision makers compile useful information from different sources, solve problems and make decisions.

DSS are (or, should be) more than information retrieval systems, they need to exploit automated problem solving. A challenge is to represent and combine many different aspects, for instance to integrate ecological knowledge with social, economic, political aspects.

Currently, IT is mainly supporting processing data and performing numerical simulation, while the tasks that involve knowledge processing and problem solving is not well supported. On the other hand, these tasks reflect the distinctive nature of this application domain. In addressing these issues, we will actually contribute to turn eco-informatics from a simple application of standard data storage, retrieval and processing techniques into a special field of IT to also incorporate automated reasoning

Figure 1.21

The role of information technology

In summary, current use of IT provides good support to acquiring, storing, processing, generating data, it still leaves the crucial activities, reasoning, and decisions to humans: gaining an initial understanding, deciding which quantities to measure, structuring a model, interpret data (obtained by observation or simulation) and deciding on the proper actions to take. What enables humans to perform these tasks is a conceptual model of the domain. The challenge to IT is to also support these activities, which requires a formal representation of such conceptual models in computer systems and the exploitation of inference systems on this basis. This means, it is a challenge to knowledge representation and automated reasoning and, hence, to Artificial Intelligence(AI). This will be the focus of this lecture.

Ch1 p.19

Figure 1.22 Role of IT

The challenge for knowledge representation and reasoning

Figure 1.23 Challenge for knowledge representation and reasoning

Challenges for knowledge representation and reasoning is to develop a model based system that represents ‘a domain of discourse’ and helps in interpretation of the objects and its relations as well as conceptual modeling which contributes in modeling and problem solving tasks.

Ch1 p.20

Focus of the lecture: the weak parts

Figure 1.24 Weak parts

Challenge for IT in ecology

In summary, there are fundamental challenges for IT in the domain of ecology.

Supporting a deeper understanding of the relevant phenomena requires better support of the modeling process and an explicit representation of and reasoning about essential concepts of the domain (e.g. population, predation, migration).

Providing common ontology for modeling. Ontology defines a common vocabulary for researchers who need to share information in a domain. It includes machine-interpretable definitions of basic concepts in the domain and relations among them.

Ontologies are required:

To share common understanding of the structure of information among people or software agents

To enable reuse of domain knowledge

To make domain assumptions explicit

To separate domain knowledge from the operational knowledge

To analyze domain knowledge

Generating explanations and supporting educational tasks. This involves automated causal reasoning (which is beyond what numerical simulation systems can contribute).

Powerful decision support systems need to be based on knowledge representation and automated reasoning, rather than on data analysis, case bases, and rules of advice.

Example: Drinking Water Treatment

Ch1 p.21

In Porto Alegre, Brazil, at some point, customers complained about the bad taste of the drinking water. While it was easily identified that the taste was a result of the high iron concentration in the drinking water, and, hence, the appropriate counteraction was reducing the concentration by adding (or increasing the amount of) an oxidation agent (chlorine, ozone), the experts had also to find out to ultimate reason for the unusually high iron concentration. The explanation they finally found was the following causal chain: “metallic taste” results from the human perception of iron dissolved in the water. The iron had been transported in the water that was pumped from the reservoir (more precisely, the upper layer of the stratified water body) into the treatment plant. It reached the upper water layer by ascending from the lower one. And, ultimately, the iron concentration in the lower layer was caused by the process of dissolving solid iron, which is contained in the sediment in the reservoir. The reason why this process was unexpected was that it only happens when the pH is in a particular range: it requires acidic conditions. The example illustrates what would be expected from an IT tool that supports the assessment of such unexplained or unexpected scenarios: from some initial facts and observations, an explanation should be generated in terms of a causal chain (or network) combining a number of known physical, biological, chemical, etc. phenomena and processes. Constructing a model that implies the given observations solves the problem.

Goal: Detecting the causes of problems automatically

The water can illustrate the objectives of this work. If the system contains a representation of the relevant, potentially present processes, such as redissolving and ascending of iron, it should be able to propose a model including these phenomena and a revision of unsupported assumptions (e.g. about the pH) as an assessment of the currently observed situation.

Figure 1.25 Water treatment domain

Ch1 p.22

Goal: Find potential remedies automatically

Figure 1.26 Final potential remedies automatically

Based on such a situation assessment, and given models of the possible human interventions, it should be able to propose a model including interventions (such as adding an oxidation agent) that transform the system towards a given goal (reducing the iron concentration in the drinking water).

What is the expected use and benefit of such collections of models and model-based decision support systems?

Benefits Expected from Use of IT

Supporting Experts

If experts (of the same or different domains and scientific areas) could turn their scientific results into models that capture their knowledge about certain phenomena independently of each other, but based on a shared ontology and can be represented and combined in a computer system, they can more easily share, compare, and combine their knowledge in order to achieve synergy, produce more powerful models and do so more easily.

Ch1 p.23

Figure 1.27 Expert supporting

Supporting non-experts

Once expert knowledge is represented in a knowledge-based system, it can be made available to non-expert, who are unaware of the details of this knowledge, but can enter a superficial description of a particular domain problem and obtain proposals for proper decisions and actions based on the expert knowledge.

Figure 1.28 Non-experts Supporting

The vision: research and decision making

The grand vision for knowledge-based environmental decision making is that turning research results into model fragments that can be represented, combined, and executed in a computer helps to overcome the barriers between different disciplines and

Ch1 p.24

results from different areas and supports the integration of scientific insight. Combining it with model elements that represent the human activities derives an integrated model which helps to identify possible interventions and their impact on the environment or particular eco-systems. In a nutshell, these efforts will bridge a gap between the decision makers and research communities to not only analyze and address the current environmental problems but also to predict the future implications.

Figure 1.29 Research and decision marking

Notes

1 http://en.wikipedia.org/wiki/Ecoinformatics

2 Essentials of Ecology, p17

3 http://envirosci.net/111/scope.htm

4 http://science.jrank.org/pages/403/Anthropocentrism.html

5 Environmental Applications of Data Mining, Saˇso Dˇzeroski

6 http://en.wikipedia.org/wiki/Scientific_method

7 http://www.participateschools.co.uk/resources/collecting-environmental-data

8 https://oira.syr.edu/oira/Assessment/AssessPP/Analyze.htm

9 http://www.gisdevelopment.net/tutorials/tuman001b.htm

10

http://en.wikipedia.org/wiki/Numerical_weather_prediction 11

http://en.wikipedia.org/wiki/Knowledge_representation_and_reasoning 12

http://adr.sagepub.com/cgi/content/full/17/1/69

Ch1 p.25

13 Ontology Development 101: A Guide to Creating Your

First Ontology 14

http://dssresources.com/history/dsshistory.html