ANIMAPS
Sandra L. Arlinghaus, William D. Drake, and John D. Nystuen
all of both University of Michigan and Community Systems Foundation
with data and other input from Audra Laug, Kris S. Oswalt, Diana Sammataro
University of Michigan; Community Systems Foundation; Pennsylvania State University (respectively).

Introduction

Spatial analysis is often conducted in a single time slice. One sees thematic maps showing per capita Gross National Product by county for a single year, yet others showing pH level from soil samples interpolated across an agricultural surface, and still others showing, at the global scale, the association between total fertility rates and level of women's education in developing nations. To be sure, each of these is useful and each offers directions for further analysis that might otherwise go undetected by a student of the data set alone; each can also serve as a baseline map from which to build over time.

How much more useful these, and other maps, can become when change over time and diffusion processes can be tracked in the underlying variables being mapped. A sequence of static maps can be grouped as a single animated file, flipping from one map to the next, with exact overlays. When the animated file is uploaded to the Internet, the reader is offered a unique opportunity, not possible in conventional hardcopy publishing: to have color animated maps at virtually no cost.

Animated maps ("animaps") can be made in specific software packages, such as Microsoft PowerPoint; however, conversion of the file format to one that is readily uploadable to the web will be required (current versions of PowerPoint offer such an option). Another route is simply to use animation software that makes animated files from files in .gif or .jpg format. These packages have the advantage of universality. Any map in a GIS can be exported, for example, to Adobe PhotoShop and saved there in either .gif or .jpg format. The manner of exporting the map from the GIS in which it was created may not always be straightforward (although in most modern GISs it is). When the export process is not obvious, it is always possible to use the command that is universal in Windows (3.0, 3.1, 95, etc.), alt+PrintScreen, to capture an image of the map on the Windows clipboard and paste it into a blank canvas in PhotoShop (cropping the pasted image as required). Universality has many advantages: it offers flexibility to users with many state-of-the-art options and it offers perhaps unforeseen opportunity to users who are still several generations removed from state of the art.
 
 

Long-range Planning

From 1992 to 1996, Community Systems Foundation (501(c)3 corporation) in Ann Arbor developed a Management Information System (directed by Drake and Oswalt) for the country of Syria for UNFPA (United Nations Family Planning Agency). During the course of that project, there were five missions  (MCH/FP:  Maternal and Child Health/Family Planning) to develop a system for the monitoring and evaluation of health care and family planning at Syrian Ministry of Health facilities (both urban and rural). Among other things, sequences of GIS maps were developed (Oswalt and Arlinghaus) to show distribution patterns of acceptance of various health care practices at various geographic scales: from the highly local to country-wide. Considerable effort was expended in getting best-possible spatial information from field sources. Further effort involved training and education on MIS/GIS use of indigenous personnel, both in-country and in Ann Arbor, with the eventual goal that with mastery of mapping and other software would come independence and the capability for the Syrians, themselves, to continue the process begun with our assistance.

Even in a situation of sparse spatial data, thousands of maps were generated over a period of five years. For the maps to be useful, they needed to become available to a widely scattered audience. What was possible given the state of technology at the time in Syria, was to make a series of wall-map sized posters showing various GIS maps and offering a brief comment next to the map on the poster. When laminated and mounted on the walls in selected Syrian health centers, these maps told a story that the mounds of data collected at those centers never revealed to the workers in the centers. They served as a continuing source of motivation (as they, for the most part, transcended language barriers) and as a glimpse of what the future might bring. It was for related purposes that data accumulated over a fourteen month period was made into an animated map (Arlinghaus, February, 1997). This map compressed volumes of data, of Syrian Ministry of Health data mapped by health center, into a single 20 second file.

Animated map: the small red dots represent Syrian Ministry of Health Centers. The red triangles represent the total number of women visiting the health center in a one-month period. Data sets were from November, 1994 to December, 1995. Because there is no animated legend on this animated map, the single maps are enumerated below by month.

Single maps, by month:

November, 1994
December, 1994
January, 1995
February, 1995
March, 1995
April, 1995
May, 1995
June, 1995
July, 1995
August, 1995
September, 1995
October, 1995
November, 1995
December, 1995

In order to analyze an animated map it can be helpful also to have the individual maps, from which the animation was made, available as well. In the case of the Syrian animap, it appears that, when viewed at the country scale, much of the variation occurs in the province of Aleppo (see reference map for place-name information).  Overall, there appeared to be general expansion of acceptance of MCH/FP over time; however, more fluctuation in pattern, from month to month appeared than one might expect.  If the expansion is real, certainly the trend as measured by the data is NOT one of steady increase.  This observation might lead one to consider the level of steadiness and local cultural preferences in reporting data on a month by month basis.  Accumulated data could easily account for the high level of variability.  The animation points to one direction for further investigation:  that of tracking timeliness of reporting of information.

Global Animap

In the case of the Syrian animap, more information might suggest level of significance of observed patterns. In some cases, diffusion, be it of the "infill" or "sprawl" sort, is clearer than it is in others. A few months ago, Nystuen discussed the concern of his colleague in Entomology, Sammataro, in tracking the global diffusion of the Varroa Mite, a pest which threatens the honeybee population. This seemed a good opportunity to press into service the technique developed for Syria of using animaps.

Data was provided for most of the world's countries for much of the 20th century. The same basic strategy as was used for the Syrian data was to be used for the mite data, as well.  One additional suggestion (Nystuen) was to alter the interval between images to reflect the uneveness in time points for collecting the data (longer spacing between frames show a wider gap between data observations). Yet another was to have previous data in one color and data relating to the current frame in a different color (Nystuen). We used a simple color selection to track the advancing wave of the varroa mite across the nations of the world. Future work with these maps might involve deeper numerical analysis of the data, as percentages or other, and subsequent remapping (Drake) as the data quality permits.

Varroa Mite Animap

In this case, the pattern of advance is clear and, unlike the Syrian case, is one of steady increase.  Still, one might wonder about variability in the timing of reporting.   It is interesting that from early beginnings in Southeast Asia, the initial spread was quite slow. With air travel becoming more frequent, post World War II, the spread of the mite accelerated; indeed, with the more generally interconnected world, the pace of mite diffusion has also accelerated. Whether or not there is a causal connection would need verification. There is, however, an obvious spatial association that is enough to suggest such additional study.

Surrogates for lack of data

Sometimes it is difficult to acquire data over a period of time. With a bit of imagination, one may instead be able to use a surrogate variable to capture easily what might otherwise have been difficult to capture. To illustrate this sort of technique, in a time-dependent framework, consider the following animap. When African-Americans first came to North America, they entered often along the south and southeastern shores of modern-day U.S.A. Over time, population migrated and moved throughout the country. If one considers as a surrogate to having year by year data for that movement, the fact that not many people move over time all that far from their point of entry to the country, then one might capture the temporal movement pattern over centuries by the spatial density pattern at a single time slice. To test this idea from the standpoint of simple mapping, the U.S. was mapped by county according to density of African-American population (1990 Census data) (S. Arlinghaus and A. Laug). The mapping of this initial test-run was kept simple:  the default lat/long framework, rather than a conventional projection, were used in the GIS (Atlas GIS, version 3.03) for eventual ease in switching to other projections.  As had Nystuen, Laug wished to color percentages from previous frames all in one color, with percentages in the current frame colored in a different set of colors. She also wanted to track the advancing edge, as had Nystuen, but in addition wanted to see gradations in that edge. There is a tradeoff in clarity; how many categories should one use on the edge?

U.S. African-American Population Density by County, 1990.

The pattern that emerges at this one time slice does appear to mimick the general history of African-American migration in the U.S., over centuries. To have a firmer idea as to the extent to which the internal U.S.migration/density assumptions actually migration to the U.S. over time, a number of additional steps would be required.  The animap guides the research direction.  Indeed, some of the issues one might reasonably examine involve (but are not limited to)

All of these might be captured within a broader fractal/chaos framework.  Self-similarity is at the heart of this transformation from time to space: migration frequency is similar to density patterns within counties.

Animaps display spatial and temporal pattern together in a single .gif file.