Preprints of the Metadiversity Conference Proceedings

  Session 3: The Challenge in Earth Observation, Ecosystem Monitoring, and Environmental Information

The Challenge in Earth Observation, Ecosystem Monitoring, and Environmental Information

ROBERTA BALSTAD MILLER, Director, the Center for International Earth Science Information Network (CIESIN) at Columbia University

It has been a very interesting day thus far. We have had people discussing what should be done, people discussing what will be done, and then people discussing what is currently being done. It is that last category that I think we will take further today with what really is a very outstanding panel reporting on some systems efforts both at the national level and international level.

But first what I would like to do is look at the PCAST report of the Panel on Biodiversity and Ecosystems, in historical perspective. Then I will address three major challenges that are raised by that report and that we need to deal with in responding to the report. Those challenges, I would argue, are the challenge of data integration, the challenge of data access and dissemination, and, finally, the challenge of public information.

Background

The PCAST Panel on Biodiversity and Ecosystems emphasizes the need to use data and information resources for monitoring ecosystems and for integrating data. What the panel members had in mind when they wrote their report was economic data and biodiversity data for providing policymakers and the public with information for resource management. All of this is important. But the desire to build links among scientific research, data management, and public policy is not new.

National Income Accounts. In the 1930s, the economist Simon Kuznets, working at MIT, developed the National Income Accounts, which were time series data of economic activity (in terms of productivity) in all sectors of the economy. His goal in doing this was to improve our understanding of the economy and to provide a means of measuring change in productivity in various sectors. National Income Accounts are the economic indicators that still affect economic policy and, some would argue, are responsible for our economic prosperity in the period since then. I think this can be seen as a very successful attempt to link the scientific research in economics with information series and public policy.

Social Indicators

There was a second attempt also. This attempt took place in the late 1960s and the late 1970s. Taking off from the idea of economic indicators, there was in this country–and in a number of other countries–something called the Social Indicators Movement. The Social Indicators Movement was based on the recognition that what is most important in national policy is not economic in nature and that, while economic indicators are themselves useful, they are not enough. National policy needs to be informed by statistical data on both tangible and intangible changes in the society.

Tangible kinds of indicators that were discussed included education test scores, infant mortality rates, literacy levels, and unemployment statistics. Intangible indicators included topics such as health and well-being and confidence in government. (If you think that confidence in government is not very important, look at the United States and Russia today and see how important confidence in government is for the smooth working of a country and, in particular, an economy.)

Economics vs. Social Indicators

There are significant differences between economic and the social indicators. The economic indicators have a common metric: They use dollars and cents. Everything is expressed in money. As a result, the economic indicators are additive. You could add these indicators together and get a gross national product.

Social indicators are different. They use diverse metrics. You don't measure literacy, unemployment, and confidence in government along the same metric. They are very, very different. As a result, social indicators are not additive. They are all separate. There have been attempts to combine various kinds of indicators into a composite indicator. (One of the more well-known attempts resulted in the Indicator of Development, which I believe was composed of literacy, infant mortality, and education of women.) But in almost every case, the additive social indicators left out something. When you combine social indicators, you lose information and detail. Consequently, social policy based on these additive indicators did not work terribly well.

The Current Situation

Today, we are facing similar problems in some respects and different problems in other respects. We need data series that will help us in research, in policy, and in resource management. But the circumstances today are very different than they were in the ‘30s or in the ‘60s or in the ‘70s. Biodiversity and ecosystems are complex systems. They are not closed systems. They are affected by economic activities. They are affected by social, demographic, and cultural activities and phenomena. They are affected by politics and by public policy, including treaties, regulations, war, foreign policy, and transportation policy. They are affected by physical and environmental change at regional and global levels. Then, of course, there are the biological functions that take place within this broad, shifting framework of many other types of change.

A second difference in the situation today is related to advances in information technologies. We have the capability to obtain and save vast quantities of information–probably much more information than any one person could use. So, part of the problem is that we have an embarrassment of riches. The problem addressed by this meeting is to bring order to all this information. But policymakers are not going to be able to deal with the vast quantities of information that scientists can deal with or that data managers are going to be able to deal with. The translation from science and data management requirements to the policy framework has to be a matter of imposing order on that chaos.

A third difference from earlier attempts to create data series for public policy is the growing emphasis upon public as well as policy information. In the 1930s, Simon Kuznets did not worry about informing the public about environmental indicators. This was–even then–a public policy issue but not something about which the public was concerned.

Things are different today, and such issues must be addressed by this group and by others who wish to respond to the PCAST Panel on Biodiversity and Ecosystems.

The Challenge of Integrating Data

As I noted at the start, there are three challenges for us to accept. First of all is the challenge of data integration. We have already heard a great deal about the data problems of biodiversity. But it is a lot more complicated than data. In a very real sense, biodiversity is about people. Biodiversity is about economic markets, and biodiversity is about global environmental change. In order to understand biodiversity, you have to have data integration. You have to pull data on all of these together into a single data series or a single database or a single type of data. This is true whether you approach the topic from a scientific, a policy, or a public information perspective. Because the science of biodiversity involves so many fields, the data series themselves have to involve multidisciplinary data.

Integrating multidisciplinary data is not an easy task. For example, you may have to compare or combine remote sensing data with population data, with transportation data, with in situ data in order to have the background to deal with certain kinds of land-management issues. We don't have a common metric. We are not like the economists in the 1930s and thereafter. We don't have dollars and cents that we can use for all of these data series. Space–spatial representation–frequently becomes the framework for integrating the data.

The Problem of Space

But space also creates a problem in data integration, because the unit of analysis or the basic spatial unit differs for the three major types of data that I am talking about. Remote sensing data is provided to us in an image, and that image then is superimposed with an imaginary grid. Scientists use that grid to analyze images.

But socioeconomic data are collected in political jurisdictions, and those political jurisdictions never approach a grid (except for a few places in the Midwest). Jurisdictions are determined by historical forces, historical practice, or historical agreements. Jurisdictions also can be determined by rivers or by, in some cases, the shifting boundaries created by war, politics, and treaties. All socioeconomic data that are collected by the government are collected for political jurisdictions. So you have to break out of the tyranny of those political jurisdictions in order to put the data in a framework where you can use them with the gridded data that are available through remote sensing.

To complicate things further, you have ecosystem data. Now, ecosystems don't translate easily to a grid or to a political jurisdiction. So, you have still a third geographical area that you have to put together when you are integrating data. Therefore the data themselves often need to be transformed before they can be integrated and used together. This is a really difficult task.

Examples of Integrated Data

What I want to do now is give you a couple of examples of integrated data. One example is the newly created gridded population of the world map. It is roughly a five-minute-by-five-minute grid, and, obviously, the areas must differ up at the poles. The map is imperfect, but it is being corrected right now. In addition, parts of the map are better than others. However, this map marks the first time we have ever been able to produce population data that wasn't expressed by national boundaries but instead by the grid. Of course, you don't want to lose the national boundaries, because that is where the laws and the regulations are enforced. So, you need to move between those two means of representation. In fact, at CIESIN, we have also gridded the Mexican population on a one-kilometer-square grid. These data are all available online.

The next example is a program that we maintain online called the Demographic Data Viewer, or DDViewer. This provides you with mapping capability for the U.S. Census. You can map the entire country, a state, a county or group of counties, and you can even map counties across state boundaries. And since census databases are created from massive state databases, to be able to cross these state boundaries is quite a feat. You are able to go in and select the unit you want to map. You select the variables that you would like to have on the map and it goes down to the census-block group. You can specify the parameters of what you are doing through a program. Then you basically press the Map-It button, and you get your data instantly. Again, it is a way of visualizing demographic data, population data, and census data in a different way so that you can integrate them with various kinds of information. It enables you to get away from the tyranny of the political jurisdictions in the display of socioeconomic data.

Still another product that we maintain online is something we call DDCarto that translates census data to other kinds of units. From the counties and the states, you can translate data to zip codes, you can translate data to congressional districts, and you can translate data to eco regions. Therefore, you can move from one kind of geography to another kind of geography.

It is this kind of work that needs to be done for data integration. It is as much data preparation and data development as it is data analyses. And yet, if you are going to integrate disparate types of data and if you are going to provide data for public policy and public information, you have got to go through these exercises and you have to transform your data.

The Challenge of Data Access and Dissemination

The second challenge that needs to be addressed is the challenge of data access and dissemination. This is something that we have talked about a number of times already today. People have talked about the need for interoperable metadata. People have talked about the need for common metadata standards. People have talked about the need for distributed information management systems. These topics are going to be addressed in very experienced and able detail by the panel today, so I am not going to go into too much detail now. I would emphasize though that distributed information systems for biodiversity must link with multiple kinds of data. It is not enough simply to have biodiversity or biological data. You have got to link into the socioeconomic data and the global-change data.

Let’s look at the data and information system from the Socioeconomic Data and Applications Center (SEDAC), which is part of NASA’s Earth Observing System. Because many of the data on socioeconomic factors have to be pulled from many different sources, the SEDAC search system provides a means of searching multiple data catalogues, either singly or all at once. You can search it through a structured system, through a key word, and through a geographical interface. This is a metadata search tool. You are not searching the data themselves–you are searching the metadata to identify metadata that might be of interest. We do have a version of this that allows people, particularly in developing countries, to do an e-mail search of this catalogue, because although many of the people who use it do not have the bandwidth for full connectivity, they do have e-mail available.

The Challenge of Public Information

The third challenge that is raised by the Report of the PCAST Biodiversity and Ecosystems Panel is the challenge of public information. There are a number of reasons to focus on public information: The panel recommends it. The convention on biodiversity recommends it. Agenda 21 recommends it. A host of other public reports recommend that a data management strategy must include a public information component as well.

But another reason for doing so is because the technology has a bias for public dissemination. The days when information was placed in libraries or provided only to those who had a "need to know" are fast disappearing.

A third reason for emphasizing public access, as well as policy access and scientific access, is related to democratic traditions. This is an argument that comes out of the earlier Social Indicators Movement. One of the leaders of that movement, Sten Johanson, a Swedish sociologist who was in charge of the Level of Living study in Sweden, argued that in a democratic society, a government had a duty to inform (to give data to) its citizenry on public policy, and, furthermore, the citizens had a right to the kind of data that would enable them to evaluate how well they were being governed. For the first time in history, we have the technological capability to make this happen.

Biodiversity and ecosystem research can have indicators, can have a steady stream of public data because, again, the technology has changed. We have moved from the book to the electronic medium. But there are some requirements that this technology lays on us.

First, it requires having a user interface that is very friendly. Not everyone is going to be technologically sophisticated. Computer data programs must be easy to use.

Second, I would argue that there should also be multiple means of dissemination of this information. There should be computers and electronic information systems. There should also be computers with printing capabilities available in public information centers or public information places. I am thinking of libraries, civic buildings, and the contemporary market place: shopping centers, convenience stores. If there were computers there that had data available and a printing capability, even someone who wasn't able to run the computer could get the information he or she needed and walk away with a printout.

Still another aspect of providing indicators for public policy is providing training. It is going to be easier, obviously, for younger people than older people, but some kind of training program is a valuable and logical part of the public information program.

In Summary

In summary then, one of the central recommendations of the PCAST Panel on Biodiversity and Ecosystems is to translate scientific research into data that can be used in monitoring ecosystems, in managing biodiversity and ecological resources, and in forming public policy. This will only happen, I would argue, if we are able to improve our capacity to integrate data, if we are able to document and disseminate the data, and if we are able to make the resulting data and information available to both policymakers and the general public.

Previous | Next

Copyright © 2003 NFAIS. All rights reserved. No part of this product or service may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written consent.

Privacy Policy