In this paper we explore two questions that we feel are important to investigate further if geographic information observatories are to be fruitful research endeavor in GIScience. The rst question is simply `what is a geographic information observatory (and what is it not)?' and the second question is `what use is a geographic information observatory?'. The construction of large-scale geographic information observatories has the potential to be an exciting development, but it remains unclear what forms they might take. Furthermore, the reasons articulated so far for building these observatories remain largely in the domain of information science and have not been motivated in context to broader scienti c problems, such as global climate change. We investigate how geographic information observatories can support science in other elds, focusing on the example of socio-ecological system research. We argue that it is in the application of geographic information observatories toward solving big problems that they can garner community buy-in and demonstrate real impact.
Astronomical and terrestrial observatories have existed since antiquity with
perhaps the rst state-sponsored observatory as research institute having been built
in Baghdad in the 9th century CE during the reign of al-Ma'mun [
Recently, the analogy of building information and web science observatories has
been proposed, though this idea is nascent and remains largely underspeci ed
[
Type of observatory Examples
Geographic observatory Long-term Ecological Research Network (LTER)[
Both of these interpretations of geographic information observatory can be viewed as distinct from the idea of a geographic observatory (sans information), where the objects of interest{i.e., what are being observed{are terrestrial processes, though the observatory's products are information objects. These information sources and products (from geographic observatories as well as other less-organized and emergent sources) might be the information events that an information observatory observes, however. Even so, since all kinds of observatories perform information processing, any line between geographic observatories and geographic information observatories is quite fuzzy and more exploration of the di erence{e.g., in terms of information processing steps subsequent to physical measurement, or in terms of contextual meta-analysis{might help us clarify the form that we would like these observatories to take. For example, depending on the de nitions we choose, the Urban Observatory1 can be considered either a geographic observatory or a geographic information observatory (or perhaps both). As well, although geographic information observatories are in some sense one step removed from the direct process of making observations of the Earth and human behavior, geographic information observatories can still be considered important tools for geographically-related scienti c research. Understanding the primary sources of data production, data use, re-use, and integration all fall within the purview of geographic information observatories as we envision them.
However, even if we better de ne what we mean by geographic information observatories that does not necessarily give us a meaningful road map of priorities for how we should proceed toward building the observatories. Instead, what gives us that roadmap is an investigation into why geographic information observatories are an important development. One lesson we can take from real-world observatories is that for there to be community buy-in to the investment of time and money necessary to build large-scale observational infrastructure requires clear purpose. In the remainder of this paper we argue that a good strategy is to look deeper into how geographic information observatories can support scientists trying to solve big social and environmental problems. 2
2.1
Not to ignore our curiosity as information scientists about the nature of the
geographic information universe for its own sake, but as we consider the question of
why we need geographic information observatories we feel it is necessary to look
to problems that are not restricted solely to the domain of information science.
That is, the question of why is driven by the need to solve important problems
irrespective of discipline. There is a tendency among geographic information
scientists to assert their desire to do GIScience (with a big-S on science) and
not be simply seen as tool-builders. This perspective focuses on Goodchild's [
Geographic data is obviously extremely useful in many kinds of research, but in order to get at what we want geographic information observatories to look like, we might rst start by asking: what are the elds of research where we are severely hampered by a lack of understanding of the kinds, forms, and diversity of geographic data that exist in the information universe? There are likely many answers to this question, but interdisciplinary and transdisciplinary research elds that rely on the synthesis of geographical data from disparate sources are obvious candidates. We would like to highlight one subject of study in particular, socio-ecological systems, which we believe would greatly bene t from a better understanding of the geographic information universe, and therefore help justify why we want to build geographic information observatories. 2.2
Socio-ecological systems (also known as social-ecological systems or coupled
human-environment systems) concern the holistic study of the complex
relations between human social action and biogeochemical and physical systems [
The socio-ecological system perspective advocates an integrative approach.
However, although there is conceivably a vast amount of geographic data that
can shed light on understanding socio-ecological systems, most of the data is
not integrated, nor even accessible to a scientist in any useful form. In fact, in
many cases even when the data is available, it is unclear what data is useful for
answering a speci c question. In part, this is due to the fact that social data and
environmental data are collected and stored using very di erent procedures and
formats, and for di erent purposes. Furthermore, the study of socio-ecological
systems is fundamentally the study of equilibrium, resilience, robustness,
adaptability, and vulnerability [
Socio-ecological systems are adaptive, complex systems operating at
multiple spatial and temporal scales, and our sensors of the geographic world are
also operating at several spatial, temporal, and thematic scales. Ideally, there
is great potential to exploit the increasing availability of multi-resolution and
multi-thematic geographic data to model socio-ecological systems. However,
because we do not have a good picture of the geographic information universe,
it is extremely di cult to match and integrate the relevant data that might
be scattered across the information universe to speci c socio-ecological research
problems. For many socio-ecological analyses, it is often the case that in order
to get at a composite social-environmental picture requires using heterogeneous
information artifacts that people \leave behind," e.g., through social media,
GPS tracking, etc., in combination with environmental sensor data of varying
resolutions and themes, such as metagenomic analysis and spatio-temporal
microparticle measurements. The investigation of exposure over the life-course of
an individual (exposome) is a prime example of this kind of analysis [
In the context of answering what form we want geographic information observatories to take we might also identify some of the various aspects geographic information that we want to be able to observe in order to support the scienti c process.
Finding what we need. We need observatories that specialize in nding and ingesting geographical data from a multitude of sources. Along many fronts these are already being developed in the context of spatial data infrastructures, linked geo-data repositories, and the like. For the example of socio-ecological research this entails mapping the vast variety of environmental and social geographicallyreferenced data that are available.
Understanding what we nd. We must also understand what kinds of data
are useful for what purposes, which is in many ways a much more di cult
research problem [
Watching the detectives. We want to measure the change of geographical information, both in terms of content and related conceptual framing and interpretation. For example, in order to recognize that data from a source is used for one purpose by the originating community but has been re-purposed by an unrelated community. We want to be able to observe these di erences and their patterns through space and time. Also, at the community-level scienti c knowledge is in ux and interpretation of data recorded in the past through a present-day lens will often be distorted.
tance to geographic information observatories, we want to be able to observe the myriad relationships between attributes of geographic information and its use. For example, exploring the relationships between scale and process, and the implications of those relationships for speci c scienti c research questions, which could also entail observing methodological di erences in how scale is conceptualized by di erent users of geographic information. We could also study the geographic footprint of data in terms of where (when) it is used as opposed to where (when) it depicts, or what it was used for as opposed to what it was made for. 3
More often than not, geographic information scientists interested in organizing information about socio-environmental and other complex geographic systems are limited to using their \table-top telescopes" that are only pointed in one direction and which collect data that are easy to get, but biased and of limited interest (e.g., Twitter API). Or worse yet, they are information astrologers designing ontologies with no basis in observation! Information observatories (both geographic and non-geographic) do already exist{companies like Google and Facebook and governmental organizations like the United States National Security Agency have developed information observatories that are designed to answer the needs of commercial industry and the intelligence community. Where do we as academic information scientists position our information observatories in this environment? The observatories these organizations have built are powerful, but they are not designed to support scientists working to solve the many di cult problems of the next century. We have an opportunity (and imperative) to build similarly powerful observatories that will support critical research questions in climate change, health care, environmental science, social sciences, and the like.
We would like to thank Cameron McLean and Brandon Whitehead for contributing to the discussions that led to this paper. 16. Wood, A.W., Leung, L.R., Sridhar, V., Lettenmaier, D.P.: Hydrologic implications of dynamical and statistical approaches to downscaling climate model outputs.
Climatic change 62(1-3), 189{216 (2004) 17. Young, O.R., Berkhout, F., Gallopin, G.C., Janssen, M.A., Ostrom, E., van der Leeuw, S.: The globalization of socio-ecological systems: an agenda for scienti c research. Global Environmental Change 16(3), 304{316 (2006)