Final Report: Integrating Economic and Physical Data to Forecast Land Use Change and Environmental Consequences for California's Coastal Watersheds.EPA Grant Number: R829803
Title: Integrating Economic and Physical Data to Forecast Land Use Change and Environmental Consequences for California's Coastal Watersheds.
Investigators: Merenlender, Adina , Biging, Greg , Landis, John
Institution: University of California - Berkeley
EPA Project Officer: Michaud, Jayne
Project Period: July 1, 2002 through June 30, 2004
Project Amount: $259,454
RFA: Futures: Research in Socio-Economics (2001) RFA Text
Research Category: Nanotechnology , Economics and Decision Sciences , Water and Watersheds
Land-use change has received less attention than other threats to natural systems such as global climate change and air and water pollution. This is true despite the fact that land-use change is the primary driver of habitat loss and ecosystem degradation and greatly exacerbates most of the other threats to the environment (Harte, 2001). Land-use change has multiple socio-economic drivers and complex biophysical outcomes that generally elude our ability to provide a technical fix for the problems generated. Our approach has been to address the process and pattern of past land-use change, forecast future land-use change, and determine the risk of these changes to the environment to help plan for a more sustainable future. The first step in this process is to establish a causal relationship between land use and environmental impacts. For this research project, we did this by examining the relationship between land cover and the condition of downstream habitat.
The likelihood of future land development or conversion also needs to be considered. The expected probability of land-use conversion often increases as a function of the value of developable land. One approach we take for finer scale decisionmaking is to actually estimate land-use change as a function of the underlying biophysical and socioeconomic characteristics based on sites that have previously been converted (Chomitz and Gray, 1996; Wilson, et al., in press, 2005). Mapped biophysical and socioeconomic characteristics derived from a geographic information system (GIS) can serve as explanatory variables to estimate the relative probability of each land-use alternative. For example, forest conversion to agricultural use will be more likely in areas with suitable soil quality, slope, access to water or precipitation, and access to markets. The results from these analyses may be used to predict the relative probability of land-use conversion for remaining developable sites. For this study, we used this approach to forecast future land-use change across multiple watersheds.
Land-use changes and resulting impacts to natural habitats happen incrementally with seemingly subtle effects. It often is difficult to assess the cumulative impacts of these incremental changes, and most environmental impact assessments do not take them into account because the analysis is done at the site scale and does not evaluate the potential for larger scale impacts. This research explicitly addresses cumulative impacts at a watershed scale.
The objective of this research project was to develop spatially explicit economic and reduced land-use change models and use the best of these models to examine the environmental consequences of land-use change for California’s coastal watersheds that are experiencing rapid urban and agricultural expansion. These foremost land-use stresses can result in cumulative impacts to coastal watersheds that impact anadromous fish. The specific objectives were to: (1) develop spatially explicit economic land-use change models; (2) compare an economic modeling approach to a more traditional noneconomic (or reduced form) land-use change model; (3) estimate changes to future land cover based on the best performing land-use change model; and (4) quantify potential impacts of continued land-use change for coastal Mediterranean watersheds and instream habitat for endangered salmon.
These objectives were met first by developing two types of economic models. The primary model used hedonic analyses on recent sales of extensive use parcels unrestricted from future development to estimate residential and vineyard values. The second model used a similar model framework but utilized expected net present value (NPV) as an alternative economic measure of vineyard land value. Expected NPV was determined directly from farm-level information on grape price, yield, and establishment costs, as well as physical attributes. The data for this approach came from a survey that we conducted in 2001.
Results from the modeling exercises showed that slope had a significant negative effect on land value, whereas distance to the largest nearby city and growing degree days had a positive effect on land value. Lot size had a significant negative effect on land residential value. For vineyard land value, floodplain, slope, and lot size had a negative effect on land value whereas grape price and lag in land price positively influenced vineyard land value. Significant lot size and zoning coefficients indicated that current or potential residential can influence vineyard parcel values. In total, these models could explain 67 percent and 61 percent of the total variation in land price for residential and vineyard land values, respectively.
We compared these two economic modeling approaches to a reduced or physically based land-use model. For this reduced model, we used recent land use transitions to predict three possible terminal land-use states similar to the structural models described above: residential, vineyard, or extensive use. Site characteristics served as the explanatory variables that determined the underlying suitability under each alternative land use. The reduced-form model results in the highest accuracy for predicting land-use change, 73.7 percent overall for the three types.
One of the major advancements of our work was the calibration of land-use change models at the land parcel-level to distinguish impacts among different levels of residential density. In Sonoma County, the area used for this research, the extent of low-density development (1 unit per 1-5 acres) and very low-density (1 unit per 5-40 acres) is several times larger than the urban footprint (> 1 unit per acre). The increased rate of exurban development, along with the larger land area required to support it, means that 10 times the amount of land in the United States was converted to low-density development as compared to urban densities (1 unit/4-16 ha [10-40 acres]) in 2000 (Theobald, 2001). Estimates based on nighttime satellite imagery suggest that 37 percent of the U.S. population now lives in exurban areas that account for 14 percent of the land area; purely urban areas, including traditional dense suburban development, account for only 1.7 percent of the land area and house 55 percent of the population; in contrast, rural areas (84% of the land area) contain only 8 percent of the population (Sutton, et al., in press, 2005).
Projecting changes in land cover involved developing the land-use change model and then using it to convert current land cover types to those that are most likely to exist in the near future. As the results discussed above demonstrate, the reduced models more accurately predict land-use change transitions and so were used for the remaining research tasks. The reduced land-use change model used for the final application of this research was constructed using parcel-level data (Bell and Irwin, 2002). The model is conditioned on the initial land-use state, taken as “developable” parcels in 1990. This excludes those lands protected in parks and reserves and parcels already converted to residential, vineyard, or other high-intensity land uses prior to 1990 based on existing land-use maps. Land-use conversion is defined as transitions from developable parcels in 1990 to either a residential type or vineyard use during the period 1990-2000. The conversion decision is considered irreversible because of the substantial up-front fixed costs. The classes of residential densities used were adapted from Theobald (2003). The land-use change models were calibrated separately for each of the major land uses as the likelihood of development for each land use type depended on different factors.
These models were used to convert the current land cover types for each parcel to their estimated future type of land cover. Land-cover conversions for 2002-2010 were estimated using a Markov decision process, where transitions were considered stochastic with decisions partly informed by site-specific characteristics. Monte Carlo simulation methods were repeated 1,000 times to obtain the average area converted to exurban, urban, and vineyard development. Then we were able to calculate the percent of the different land uses in each watershed (% of total watershed area) for the past (1990), recent (1997), current (2002), and future (2010) time periods. The relationships between these land-cover type percentages in a watershed and stream habitat were addressed as part of our final objective.
We developed statistical models to explore the relationship between upland land use and stream habitat at two different resolutions. The first was done using satellite data to identify land-cover classes and the second used land-cover types designated by Sonoma County at the parcel-level. The later were then used to forecast the impacts of likely future land-use change determined by the parcel-level land-use change model. In both cases, we draw on existing habitat data at the reach scale from field surveys by the California Department of Fish and Game. In particular, we used embeddedness scores as the dependent variable. Between 1997 and 2000, field crews recorded the concentration or level of fine sediment within gravel and cobble substrate, termed embeddedness, at each potential spawning site on a four-level ordinal scale, from one (very low levels of fine sediment) to four (very high levels of fine sediment).
Results from our coarse-scale analyses showed a strong relationship between embeddedness and proportion of watersheds in urban and agricultural land use. The power of the empirical regression model depended on the size of the watershed. Generally, the watershed scale was the best predictor of embeddedness compared to other local or drainage network scales of influence. Detailed methods and results from this study are in Opperman, et al. (in press, 2005).
We developed multiple ordinal logistic regression models to examine the relative contribution of land use at the parcel level on embeddedness. These models again demonstrated that agricultural land in the watershed is a significant predictor of embeddedness, but they also reveal the importance of even low-density housing on stream condition. We used the resulting models to predict substrate quality in streams in year 2002 based on the observed land-use change in parcel-level housing and digitized vineyard between 1997 and 2002. Maximum likelihood estimates (MLE) and confidence intervals were estimated based on their probability distributions. We evaluated model predictions with a subset of watersheds that were surveyed from 1998-2002. Accuracy of MLE was used as the main criterion of model performance. Finally, we forecasted substrate quality and its uncertainty based on the average land-use change for 2010. Our preliminary results show that low-medium residential housing and agriculture have a strong impact on the concentration of fine sediment in streams.
Application of the Results
We implemented our research findings through a planning effort for the Sonoma County Agricultural Preservation and Open Space District designed to prioritize land acquisition in Sonoma County. The District was created by voters in November 1990 and funded with a local sales tax. We worked with District planners and decision-boards to identify and prioritize desirable conservation benefits. Initially, we compiled available digital data on open space and agricultural and natural resources of Sonoma County from available databases. The District Authority defined four categories of benefits: Agriculture, Greenbelts, Natural Resources, and Recreation.
The parcel-level models of land-use change and economic models estimating easement value for each developable parcel produced as part of this research are now being integrated into the Districts planning process to further refine acquisition priorities. These models allow staff planners to make tradeoffs for the relative development threat and acquisition cost for each available parcel. This grant allows us to develop a methodology for what we call benefit-loss-cost (BLC) targeting that aims to minimize the expected loss of biological benefits because of future land-use conversion and takes acquisition costs into account (Newburn, et al., in press, 2005). BLC-targeting selects parcels with the highest ratio of expected benefit-loss to land costs, and in this case, provides the District with an important tool to increase their efficiency in protecting open space values with a limited annual budget. Finally, we helped County computer programmers to design a user-friendly GIS interface that allows District staff to query any number of parcels and generate a report of the benefits, their relative likelihood for agricultural conversion or urban development, and predicted easement values.
The District staff uses the models we developed and digital resource maps in a variety of different ways corresponding to the goals of a specific acquisition effort and the funding partners involved. In addition, these science-based decision support tools have eased public concerns over subjective decisionmaking, allowed transparent exploration of alternative conservation strategies, and reduced the time it takes for project approval in what was formerly a heavily politicized process.
The final integration of land-cover information with watershed variables was done at two different scales, resulting in new insights gained by examining the system at different resolutions. The results from our coarse-scale examination of the relationship between land cover and in- stream habitat demonstrate that the proportion of agriculture and urban development at a watershed scale can be useful for predicting embeddedness downstream (Opperman et al., in press, 2005). The higher resolution models allowed us, for the first time, to address environmental impacts associated with different residential development densities. We estimated the expected changes to land cover caused by land-use change and how these changes may impact future stream conditions. We are in the process of refining our estimates of future stream condition to better account for the variation observed in the system prior to publishing these results.
The models developed through this research project demonstrate important improvements to the way that land conservation targeting are done. This application of our work was published in the economic and conservation biology literature. In addition, local government has adopted the use of these models and our recommended approach to better allocate funds for open space and agricultural land protection.
Our research took advantage of existing stream habitat data on embeddedness (the extent to which fine sediment is mixed with fish spawning gravels) as a measure of stream condition. There are, however, other important limiting factors to salmon recovery, such as stream flow. Therefore, we have launched new research, funded by the U.S. Environmental Protection Agency Science To Achieve Results Program, to better quantify the causal relationships between water use and stream flow. This will allow us to address an important limiting factor for salmonid recovery that is associated with land-use change but for which we have very little information on. The ultimate goal will be to apply the same approach to forecasting land-use change developed during this project to help improve future water management-a necessary step for coastal salmonid recovery.
In summary, we made significant progress on all four objectives outlined for this research project, published three papers, extended the models to local decision makers, and have a final manuscript in preparation that makes several interdisciplinary advances to the literature on environmental management.
Chomitz K, Gray D. Roads, land use and deforestation: a spatial model applied to Belize. The World Bank Economic Review 1996;10:487-512.
Harte J. Land use, biodiversity, and ecosystem integrity: the challenge of preserving earths’ life support system. Ecology Law Quarterly 2001;27:929-965.
Sutton P, Cova TJ, Elvidge CD. Mapping “Exurbia” in the conterminous United States using nighttime satellite imagery. Geocarto (in press, 2005).
Theobald DM. Land-use dynamics beyond the American urban fringes. Geographical Review 2001;91:544-564.
Theobald DM. Targeting conservation action through assessment of protection and exurban threats. Conservation Biology 2003;17:1624-1637.
Wilson K, Pressey R, Newton A, Burgman M, et al. Measuring and incorporating vulnerability into conservation planning. Environmental Management (in press, 2005).
Journal Articles on this Report : 2 Displayed | Download in RIS Format
|Other project views:||All 26 publications||3 publications in selected types||All 3 journal articles|
||Newburn D, Reed S, Berck P, Merenlender A. Economics and land-use change in prioritizing private land conservation. Conservation Biology 2005;19(5):1411-1420.||
||Opperman JJ, Lohse KA, Brooks C, Kelly NM, Merenlender AM. Influence of land use on fine sediment in salmonid spawning gravels within the Russian River Basin, California. Canadian Journal of Fisheries and Aquatic Sciences 2005;62(12):2740-2751.||
Supplemental Keywords:land-use change, habitat loss, ecosystem degradation, coastal watersheds, urban expansion, agricultural expansion, in-stream habitat, downstream habitat, watershed,, RFA, Economic, Social, & Behavioral Science Research Program, Scientific Discipline, Air, Geographic Area, Water, ECOSYSTEMS, Ecosystem Protection/Environmental Exposure & Risk, Hydrology, Water & Watershed, climate change, State, Air Pollution Effects, Monitoring/Modeling, Habitat, Species, decision-making, Urban and Regional Planning, Atmosphere, Social Science, Watersheds, Economics & Decision Making, coastal wetlands, ecosystem valuation, urbanization, habitat dynamics, environmental monitoring, biodiversity, assessing ecosystem vulnerability, economic research, policy making, watershed, urban planning, coastal watershed, fish habitat, endangered species, decision making, environmental decision making, land use effects, community based environmental planning, socioeconomics, management alternatives, endangered salmon, habitat disturbance, environmental policy, predictive model, changing environmental conditions, hedonic models, coastal ecosystems, urbanizing watersheds, watershed sustainablility, Anadromous fish, conservation biology, water quality, California (CA), ecology assessment models, econometric analysis, land use, aquatic habitat protection , econometrics