Human Behavior and Land Use Models: The Patuxent Watershed, Maryland and Southern Yucatan, Mexico.

As an environmental economist, my research interests are in the study and modeling of human behavior with respect to their natural environment. With specific regard to land use, my research is in developing theory-based models, which are then estimated with revealed human choice preference data. My work in land use modeling is in two areas: the Patuxent Watershed in Maryland (with economists and ecologists at the University of Maryland, funded by EPA & NSF) and the southern Yucatan peninsular region of Mexico (with geographers at Clark University, funded by NASA).

Human-induced land-use/cover change is being modeled for the data-rich Patuxent Watershed of the Chesapeake Bay, thus providing the spatial configuration and dynamic evolution of a landscape by capturing ecological functions, human behavior, and their interaction. The effort links remote sensed data on land use and cover with a variety of spatially explicit socioeconomic and physical data as well as "communicating" ecological and economic models. Each model employs a landscape perspective that captures the spatial and temporal distributions of the services and functions of the natural system and human-related phenomena, such as surrounding land-use patterns and population distributions. Configuration and reconfiguration of the landscape follows from the intertwining of these phenomena, and the Patuxent work offers the potential for a richer model of land use and its change by accounting for spatial heterogeneity and linking land-use conversion to features of the landscape. The aim is to predict the probability that a given pixel, of a given description and in a given location, will remain in its current use or be converted to an alternative use. While the conversion process is affected by inertia and other disequilibrium considerations and constrained by zoning and other land use controls, the change in land-use probabilities are functions of the value of the parcel in alternative uses. Consequently the analysis must be able to explain what factors affect land values in alternative uses.
The land uses in the 7,000 km2 of seven counties of the Patuxent Watershed located in Maryland, range from Washington, D.C. suburbs to rural and agricultural southern Maryland. The conversion of agricultural and forested land (open use) to residential uses constitutes 78% of the total land-use change in the seven counties of the watershed during the past ten years. As a consequence, the economic modeling effort focuses on the prediction of "open" land use to residential use through a four part process: (i) analysis of residential value as function of a variety of spatially related economic and ecological variables that are hypothesized to affect residential land values, estimated on actual transactions of residential parcels; (ii) use of the estimated coefficients of land value estimated from actual transaction to predict values for "open" land were it to be converted to residential use; (iii) use of these predictions with other explanatory variables such as zoning, soil type, and costs of conversion to estimate the spatial distribution of the relative probability that any such land will be developed; and (iv) linking these relative probabilities with a macroeconomic model of the state of the local economy to predict annual housing starts to therefore predict how many of the pixels will change in a given year.
The model is a utility-theoretic econometric model of human behavior affecting land use decisions, not driven by GIS-determinism. Using spatial data leads to interesting complications such as spatial autocorrelation, temporal dynamics, and spatial structural change. Because of these, applying standard econometric techniques to either aggregate or disaggregate spatial data generates nonsperical disturbances, misspecification, and measurement error. Therefore new estimation techniques in spatial econometrics have been developed to take some of these issues into account in the Patuxent modeling effort (e.g. Bell and Bockstael, 1997). It is still an empirical issue as to whether the information gained by using spatial econometric techniques vastly improves the estimation. The field of spatial econometrics itself is still in its early stages. The initial spatial econometric modeling work in the Patuxent model demonstrates the potential improvements in explaining and predicting land values (Geoghegan and Bockstael, 1995). Further improvements and refinements in both the theoretical and applied econometric modeling techniques for use with the Patuxent model are presently under way.
Another theme of this research agenda is to use the remote sensed data more creatively in land use modeling. For example, in order to better capture the spatial externalities that often characterize land use, and therefore have a major influence on land value, indices based on the diversity and fragmentation of the surrounding landscape around each pixel have been included in the Patuxent land value model to further explain residential land values. The intuition on including these variables is that increasing diversity might adversely affect aesthetics, but may have convenience value signifying the proximity of important work, shopping, recreation and institutional destinations, so therefore it is an empirical question over which effect dominates. Fragmentation might be considered more obviously undesirable. Holding diversity constant, increasing fragmentation signals a hodge podge of land uses. A high fragmentation index is synonymous with a checkered landscape, and the potential for large negative locational externalities. Confusion over the sign of expected effects may be very much tied to the issue of scale, another issue that increasingly is important for the social sciences to consider as discussed above. Preliminary estimation demonstrates that these additional GIS-created variables, measured at different scales, can add explanatory power to the Patuxent model of housing values (Geoghegan, et al., 1997). The nature and pattern of the land uses surrounding a parcel have an influence on the price, implying that people care very much about the patterns of landscape around them, supporting the belief that severe externalities exist in land use and in land use patterns.
Another fruitful means for addressing the human-environment relationship, especially where spatially explicit data are sparse or of coarse resolution, is to link empirical models derived from the remotely sensed imagery with theory-based models of dynamics of land-use/cover change. An example is work underway in the southern Yucatán peninsular region (SYPR) that merges remote sensed data-based Markov modeling approaches with field-and statistical-based models of the various land managers that are producing the signals registered in the imagery.
SYPR covers the Mexican states of Quintana Roo and Campeche, northern Belize, and northern Petén, Guatemala, from the Caribbean to Gulf of México. The dominant semi-deciduous tropical forests of the region came under assault after a major highway was built through the center of SYPR, connecting the east and west sides of the peninsula. This road became the pathway for various new land-users: first slash-and-burn farmers on communally designated lands, followed private ranchers, NGO-sponsored rice projects, and more recently, biosphere reserves. Most of these changes coincide with Landsat imagery.
This imagery can be used to classify land-covers by various time periods, cross checked by field observations, and applied to standard Markov chain analysis. This approach, based on the assumption that the immediate past is the past predictor of the near future because of "stationarity" of the processes involved, uses transition probabilities of past states (e.g., land uses) to estimate future ones. Markov approaches have been used successfully by ecologists to assess land cover, and seem appropriate to explore regions with "low" levels of "chronic" change, such as the ephemerally used forests in much of the SYPR. Their applicability for multiple land-use changes is less certain, especially where stochastic processes operate as in SYPR.
Markov approaches can be made spatially explicit on a pixel by pixel bases and the transitions of each pixel weighted by its cover and use characteristics as well as the varying processes operating within it. Exploring such elaborations for SYPR revealed that by accounting for slope, elevation, and distance to nearest different land cover predicted more than 95% of the spatial variance in forest cover (Ogneva-Himmelberger & Turner 1996). The results for other transitions, such as those of various kinds on agricultural land, were not so successful, in part because the stationarity principle applied less well to them. Since the initial cultivators entering SYPR were smallholder subsistence farmers and since the dynamics of agricultural change among these kinds of farmers tend to track well with population density, the transitions of the pixels were weighted by this factor. The overall results for the SYPR were disappointingas anticipatedbecause so much of the land cleared for agriculture in the region involved large-scale, NGO projects that were not responding the production logic of smallholders. Eliminating these projects from consideration and focusing on those areas in the SYPR where smallholder subsistence remains, the simple addition of population density improved the predictive capacity of the basic Markov approach (Ogneva-Himmelberger & Turner 1996).
Indepth understanding of the various land-use/cover changes in the SYPR, however, reveals that the processes leading to so much of the changes in the 1970s and 1980s are no longer operatingthe petro-boom period with significant investment in large-scale agricultural experimentsand that new processes related to changes in tenurial institutions are currently affecting smallholder production and livestock production. The variation in the different types of land managers can be incorporated into this kind of analysis, operating from approaches outlined in the Patuxent study above, essentially "socializing" each pixel in the analysis according to the kind of land managers associated it. The stochastic factors per se cannot be model, but they can be introduced into the basic approach through scenarios. The larger point, however, is that working from models of land managers and from the RSD, hybrid approaches emerge that prove fruitful for near-term projections and monitoring of land-use/cover change.


References

Bell, K. and N. Bockstael. 1997. "An Example of Spatial Economic Modeling: Land Use Conversion in Howard County, Maryland". Paper presented to the Association of Environmental and Resource Economists at the Allied Social Sciences Association meeting, New Orleans, LA, January 1997.

Geoghegan, J., and N. Bockstael. 1995 "Economic analysis of spatially disaggregated data: explaining land values in a regional landscape." Paper presented to the Association of Environmental and Resource Economists at the Allied Social Sciences Association meeting, Washington, DC, January 1995.

Geoghegan, J., L. Wainger, and N. Bockstael. 1997. "Spatial Landscape Indices in a Hedonic Framework: An Ecological Economics Analysis Using GIS", Ecological Economics 22(3).

Ogneva-Himmelberger, Y. and B. L. Turner, II. 1996. "Markov Modelling Experiments of Deforestation in Yucatan". Paper presented to the American Association of Geographers NESTVAL Conference, Clark University, Worcester, Massachusetts, November, 1996.