Blok

Connie Blok

International Institute for Aerospace Survey and Earth Sciences (ITC)
Geoinformatics, Cartography and Visualization Division
P.O.Box 6,
7500 AA Enschede
The Netherlands
Blok@itc.nl

Cognitive models of dynamic phenomena and their representation: perspective on aspects of the research topic

1. General questions related to the research topic

Important questions for the investigation of the topic (from my perspective) seem:

Which factors influence the kinds of dynamic phenomena that are being observed in reality?
Factors that have to be considered include: purpose of observation, application (discipline), phenomenon being investigated, the ability/expertise of the observer, scale of observation, time frame and resolution considered, whether observations are direct or indirect (e.g. depending on registrations of measurements), and perhaps more factors.
Which kinds of dynamic phenomena are being observed in reality?

The above mentioned factors have to be taken into account to answer this question.

How do observers conceptualize these kinds of dynamic phenomena?

In other words, how do dynamic phenomena influence reasoning, understanding, decision making? Are there systematic differences among observers, and if so, can these be explained from the factors that influence observation mentioned above?

Can the dynamic phenomena observed in reality be categorized, based on the ways in which observers conceptualize them?

Which factors influence the kinds of dynamic phenomena that are being observed in a (graphic) representation of reality?

Most of the factors mentioned above are applicable here as well, except for direct/indirect observation. In addition, scale/resolution of the representation and the data processing and visualization methods applied can be mentioned here.

Which kinds of dynamic phenomena are being observed in representations?

The above mentioned factors have to be taken into account to answer this question.

How do observers conceptualize these kinds of dynamic phenomena?

In other words, how do observers (or in general people) react to different (graphic) representations of dynamic phenomena? Do different displays trigger different kinds of memory, reasoning or decision making?

Can the various (graphic) representations, depending on the way in which people react to them, be matched to ways in which the dynamic phenomena observed in reality are conceptualized (and perhaps classified in topologies)?

If that is possible, it may yield better representations of dynamic phenomena than the ones that are currently used.

2. Direct versus indirect observation of dynamic phenomena

I will further elaborate on a specific aspect mentioned in relation to the questions above. Many dynamic phenomena cannot be directly observed in reality by people, but only indirectly, through registrations of measurements. This is the case where the dynamics are too slow to be noticed (e.g. geological processes), where they are hidden (e.g. taking place in or under the earth's surface, like certain soil processes) or where they are only reflecting in parts of the spectrum for which human vision is not sensitive (e.g. reflections in the near infrared part of the spectrum).

Remotely sensed data are an important source for spatio-temporal applications, of which the influence will only increase in the future, when data can be acquired at higher spatial resolutions and with more sophisticated methods. In many applications of remotely sensed data, people are relying on indirect observations. Remotely sensed data, however, may be pre-processed before they reach the user. Interesting questions are: if people have to rely (partly or entirely) on indirect observation of dynamic phenomena in reality, does that lead to the adoption of different conceptualizations of the phenomena than the ones conceived by direct observation? How are these conceptualizations matched to reality? Is it possible to design(graphic) representations of the dynamic phenomena (or visualization tools) that facilitate the conceptualization and the matching to the dynamics in reality?

3. Example of indirect observation: NDVI data

One type of data that I intend to use in my Ph.D.-research is NDVI data. NDVI stands for Normalized Difference Vegetation Index. The data are mainly derived from NOAA satellites, which carry sensors that detect emitted radiation in several bands of electromagnetic spectrum, including the red (RED) and near-infrared (NIR) channels, where radiation from green leaves is strongly represented. Healthy vegetation reflects particularly well in the NIR-part of the spectrum.

The measure is also referred to as greenness index. It is an indication of the level of photosynthetic activity in vegetation: the more the photosynthesis is going on in green plant material, the higher the NDVI values. The values are used as crude measure of vegetation health, but they vary considerably. Because of the high temporal frequency of teledetection by the NOAA satellites, NDVI data are widely used for the monitoring of vegetation dynamics, e.g. for range land monitoring, pest and disease monitoring and food security programmes. The data are particularly useful in otherwise data-poor environments, like developing countries. Data costs are relatively low or even irrelevant, since dekadal images (showing ten days averages) are freely downloadable from the Internet.

The data are pre-processed by NASA before they are made available to other suppliers or users. Users can import image data (showing ten days averages, or images that are composed using the maximum NDVI value for every pixel per day, over a week or month) into application software for time series analysis (e.g. WinDisp3). They usually start with visual inspection to determine the fitness for use (e.g. completeness, cloud problems). For exploration and analysis purposes, these data are often further processed: classified and/or statistically/mathematically analysed. The results of the calculations are represented in graphs and images. Usually, the representation in images is a static one, although there may be a film mode available in the application software, but the design and user interaction possibilities in this mode are limited.

In this example, several of factors mentioned in relation to the kinds of dynamic phenomena observed (see questions above) are known or approximately known (e.g. purpose of observation, application, time frame and resolution considered). Among the factors that are not known are the effects of the way in which the data are processed, classified, modelled and graphically represented. This influences which dynamic phenomena are seen, and the way in which these phenomena are conceptualized.

Therefore, more specific questions related to the monitoring of NDVI data and to my research on cartographic animation for monitoring are:

Which dynamic phenomena are considered to be relevant for monitoring in reality?

Or: which dynamic phenomena is the expert looking for in reality?

How are the dynamic phenomena that are considered relevant related to later evidence in reality?

Consequences of indirectly observed dynamics may ultimately become directly visible in reality, but then it may already be too late to intervene in case of undesired developments (e.g. famine because of crop failure). Still, it may be relevant to be able to verify what kind of dynamics took place.

How do experts conceptualize the dynamics that are considered relevant and how do they match it to later evidence or to reality in other ways?

Which of the relevant dynamic phenomena can be captured from (graphic) representations, After which processing and in what kind of visualization?

How are the dynamic phenomena captured from (graphic) representations conceptualized?

Which of the relevant dynamic phenomena can not, or difficulty be captured from representations?

Are (for the application) new forms of visualization valuable for the monitoring of NDVI data?
E.g. can dynamic visualization variables in cartographic animations be applied to facilitate monitoring? A user might be able to work in other ways with animated representations of the data (e.g. instead of calculating absolute or relative differences between pixel values in consecutive images, and displaying the results of two or more calculations juxtapositioned, the original images can perhaps be shown in an animated sequence. In a shorter time span more images can be viewed, and a different, perhaps unusual perspective may reveal other phenomena. For exploration/hypothesis generation about dynamics, unusual perspectives can be stimulating. It is, however, necessary to know in what way people react to different representations, e.g. to (pseudo)motion in images, before this question can be answered.
Can the various (graphic) representations, depending on the way in which people react to them, be matched to ways in which the dynamic phenomena are conceptualized?

The research will require an interdisciplinary approach, in which at least cognitive scientists, vegetation scientists and cartographers/geographers should be involved. The research can be a part of broader, comparative investigations, in which all kinds of factors are involved that influence which dynamic phenomena in reality and in representations are observed (see section 1 of this appendix). If that is done, the following question can be answered:

Are the findings only valid for particular applications, data and problem contexts, or are they broader applicable?

The main purpose of the investigation is to be able to apply functional (graphic) representation methods/ tools. The representation and the interface should, as far as possible, match human reasoning and intuition. If this succeeds, workers with geographic data may be able to better capture dynamic phenomena from representations. They may also need less training to work with the data.

Brief curriculum vitae

In 1986, I obtained a degree (with honours) in human geography, specialization cartography, at the University of Utrecht, The Netherlands. Since then, I am working at the International Institute for Aerospace Survey and Earth Sciences (ITC) in Enschede, The Netherlands. My current position is Assistant Professor in Visualization in the Division Geoinformatics, Cartography and Visualization. I am involved in education, research and consulting activities. The main research interests are in cartographic visualization in general, and in cartographic animation and the use of dynamic visualization variables for exploration and analysis in particular. The promoters of my Ph.D.-research (see covering letter) are Prof.dr. F.J. Ormeling, University of Utrecht, and Prof.dr. M.J. Kraak of the ITC. Activities in professional societies include: editor of the Dutch Cartographic Journal (Kartografisch Tijdschrift; I have also been the chief editor for over two years), and corresponding member of the ICA Commission on Visualization.

Since the Ph.D.-proposal was officially accepted by the end of last year, the physical evidence that is relevant to the topic of the workshop is still limited. However, I can point at some work done that can be considered as (partially) relevant.

Dynamic visualization variables in animations; their use for the monitoring of spatio-temporal data.
Presented at the meeting of the ICA Commission on Visualization, 18-21 June 1997 in Gaevle, Sweden. See also:
http://www.geog.psu.edu/ica/icavis/gavleabs.html.
The presentation contained the initial ideas for my Ph.D.-research.
Dynamic visualization variables in a developing framework for the representation of geographic data.
Paper presented at the seminar of the French Association des Cartographes et Géographes, 12 and 13 December 1997 in Paris, France (forthcoming publication).
For abstract: see http://www.itc.nl/~carto.
The paper is an attempt to position the dynamic visualization variables in the broader context of a developing framework for the representation of geographic data.
The visualization of relationships between patterns in spatio-temporal data sets by cartographic animations (in preparation; with Barend Köbben, Tao Cheng and Agnes Alaba Kuterema).
Aspects related to one of the dynamic visualization variables (synchronization) are investigated. The sample animations developed for the article can be viewed at: http://www.itc.nl/~carto/webcartoforum.
This Website has been presented as a possible medium to obtain reactions to visualization tool development at the meeting of the ICA Commission on Visualization, 21-24 May 1998 in Warsaw, Poland.

NIR-part of the spectrum: The index is calculated as: NDVI = (NIR - RED) / (NIR + RED), and normalized to the theoretical range -1 <= NDVI<= 1 (to partially account for differences in surface slope and illumination), but realistic values (excluding values for water, soil, noise etc.) range from about - 0.1 (not very green) to 0.6 (very green).

This page is currently maintained by Varenius Workshop Webmaster
Last Updated: Sep. 18, 1998