A. Liu

Geography Department, University of Cincinnati, Cincinnati, OH 45221-0131, U.S.A.

email: liu_amy@msn.com

S. Tong

Geography Department, University of Cincinnati, Cincinnati, OH 45221-0131, U.S.A.

e-mail: susanna.tong@uc.edu

J. Goodrich

U.S. Environmental Protection Agency, Cincinnati, OH 45268-0690, U.S.A.

e-mail: goodrich.james@epa.gov

Abstract This study uses an integrative approach to study the water quality impacts of future global climate and land use changes. In this study, changing land use types was used as a mitigation strategy to reduce the adverse impacts of global climate change on water resources. The climate scenarios were based on projections made by the Intergovernmental Panel on Climate Change (IPCC) and the United Kingdom Hadley Centre’s climate model (HadCM2). The Thornthwaite water balance model was coupled with a land use model (L-THIA) to investigate the hydrologic effects of future climate and land use changes in the Ohio River Basin. The land use model is based on the Soil Conservation Service’s curve number method. It uses the curve number, an index of land use and soil type, to calculate runoff volume and depth. The ArcView programming language, Avenue, was used to integrate the two models into a geographic information system (GIS). An output of the water balance model, daily precipitation values adjusted for potential evapotranspiration, served as one of the inputs into the land use model. Two watersheds were used in the present study: one containing the city of Cincinnati on the mainstem of the Ohio River, and one containing the city of Columbus on a tributary of the Ohio River. These cities represent two major metropolitan areas in the Ohio River Basin with different land uses experiencing different rates of population growth. The projected hypothetical land use changes were based on linear extrapolations of current population data.

Results of the analyses indicate that conversion from agricultural land use to low-density residential land use may decrease the amount of surface runoff. The land use practices which generate the least amount of runoff are forest, low-density residential, and agriculture; whereas high-density residential and commercial land use types produce the highest runoff. The hydrologic soil type present was also an important factor in determining the amount of runoff and non-point source pollution. A runoff depth matrix and total nitrogen matrix were created for Cincinnati and Columbus to describe possible land use mitigation measures in response to global climate change. The differences in Cincinnati and Columbus were due to differences in geographic location, air temperature, and total runoff. The results of this study may be useful to planners and policy makers for defining the possible impacts of future global climate and land use changes on water resources.

The earth’s climate is predicted to change because human activities are altering the chemical composition of the atmosphere through the buildup of greenhouse gases (National Research Council 1999). The Intergovernmental Panel on Climate Change (IPCC) concluded that a rise of mean surface temperature will be felt in North America in the next century (Shriner et al. 1998). Based on projections made by the IPCC and the United Kingdom Hadley Centre’s climate model (HadCM2), temperatures in Ohio could increase by 2 to 4⁰ C and precipitation could change by -20% to +20% by the year 2010 (Karl et al. 1996; USEPA 1998). Research is needed to translate changes in climatic variability into environmental impacts - such as water quality in specific river basins (National Research Council 1999).

The Ohio River Basin was selected as the study area in this project. It is a source of drinking water for more than three million people. Drinking water quality, urban and industrial discharges, and storm water overflows are important water quality issues in this region. Lower stream flows in the summer could reduce water availability for municipalities and industries. Higher summer temperatures and lower flows could also degrade water quality by concentrating pollutants. This could also lead to an increase of potentially toxic algae (Ohio EPA 1996). These water quality and quantity issues are therefore intimately related to surface water runoff (Mather 1981; Jacoby 1990; Schaake 1990).

The proliferation of computer technologies has made it possible to generate complicated climatic models, such as the General Circulation Models (GCMs), for assessing future climate change and its impacts on water resources. Nevertheless, the spatial resolutions used in most GCMs are too crude to provide reliable climatic and hydrological estimates for any type of detailed regional analysis. In 1979, the U.S. National Academy of Sciences concluded that "at present, we cannot simulate accurately the details of regional climate and thus cannot predict the locations and intensities of regional climate changes with confidence." Attempts to model regional hydrologic changes with GCMs are also hampered by the lack of accurate and suitable regional hydrological parameters (Rind 1988). Without knowing these parameters, it is difficult to describe regional impacts and develop policy. Because the current generation of global climate models is not well suited to the evaluation of detailed water resource problems, other impact assessment tools must be developed and tested (Thompson 1992). One tool that has proven to be effective is the water balance model. The original water balance model was developed by Thornthwaite and Mather (1955) to evaluate the importance of different hydrologic parameters under a variety of hydrologic conditions. This is a bookkeeping procedure whereby water supply (precipitation) is balanced against the climatic demand for water (evapotranspiration). Only air temperature and precipitation data are needed as inputs. Advantages of this approach include moderate data requirements and flexibility of use. However, there are some limitations. For instance, land use is not considered in calculating potential evapotranspiration. The model also does not account for changes in the energy budget and wind effects (Idso and Brazel 1984). Despite these limitations, water balance models have been employed over the last 40 years to evaluate the hydrologic impacts of climatic change (Cohen 1986; Gleick 1986, 1987, 1990).

During the past few years, we have experienced many extreme climatic events. We also have witnessed numerous examples of the devastating consequences caused by floods, droughts, and hurricanes. In many cases, these impacts are compounded by the inadequacies of our current water treatment systems to cope with climatic anomalies. Once the hydrologic impacts of climatic change are recognized, it is possible to undertake measures to reduce the impacts. Mitigation strategies can be grouped into engineering and management practices. Engineering practices can be costly and their benefits difficult to predict (Sale et al. 1997; Francfort and Rinehard 1994). A variety of factors contribute to the high cost of clean water infrastructure. First, the development of wastewater treatment and transport facilities requires heavy construction; it requires site preparation, poured concrete and steel structures, and major piping, pump, and other hydraulic machinery. Second, sewage treatment facilities require an array of technologies, including biological, chemical, and hydraulic controls. Third, these highly sophisticated facilities require electronic and other related control systems to ensure cost-effective operation and reliability and to protect public health and the environment.

In 1996, EPA reported a need of $139.5 billion in wastewater infrastructure construction based on eligible costs under the Clean Water Act. On March 18, 1999, EPA revised its estimates to $199.6 billion in wastewater needs over the next 20 years. This recent revision substituted $81.9 billion in Sewer System Overflows for $10.3 billion in previously reported sewer infiltration and inflow correction estimated needs. The other needs are: $44.0 billion for remaining secondary treatment and advanced treatment for attainment of water quality standards; $21.6 billion for new collector and interceptor sewers; $44.7 billion for combined sewer overflows; and $7.4 billion for stormwater (AMSA 1999).

Unlike costly infrastructure changes, management of land use patterns offers promising approach to mitigate the hydrologic effects of climate change (Henratty and Stefan 1998; Krysanova et al. 1998). For instance, it is well known that increasing urbanization will lead to an increase of impervious surface area, which in turn will increase surface water runoff (Chow et al. 1988). When areas are urbanized, wetlands are sometimes used to maintain the quality and supply of surface runoff (Mitsch and Gosselink 1993).

A tool recently has been developed to assist environmental scientists and local planners to make rapid, general estimates of the potential hydrologic effects of proposed land use changes. L-THIA, or Long-Term Hydrologic Impact Assessment, is a GIS tool based on computations of annual runoff obtained from the long-term local climate records (Bhaduri et al. 1997). L-THIA has been developed as a straightforward analysis tool that provides estimates of changes in runoff, recharge and nonpoint source pollution resulting from past or proposed land use changes. It is based on the Soil Conservation Service’s curve number method and uses the curve number, an index of land use and soil type, to calculate runoff volume and depth (USDA 1986). It gives long-term average annual runoff for a land use configuration, based on actual long-term climate data for that area. By using many years of climate data in the analysis, L-THIA focuses on the average impact, rather than an extreme year or storm. Consequently, L-THIA results do not predict what will happen in a specific year (Harbor 1994). As a quick and easy approach, L-THIA results are intended to provide insight into the relative hydrologic impacts of different land use scenarios. Although it is not physically based, L-THIA has been shown to produce comparable results to that of SWMM (Bhaduri, personal communication). It is an ideal tool to assist in the evaluation of potential effects of land use change and to identify the best location of a particular land use so as to have minimum impact on the natural environment of the area (Bhaduri et al. 1997).

To date, no studies have integrated the water balance model and L-THIA into a single evaluation tool to examine the relative and combined effects of global warming and land use changes on the hydrology of a watershed. By combining the two models, environmental scientists can construct a more powerful and accurate hydrologic model than if the models are used separately. The original water balance model does not use land use type in its calculation of evapotranspiration and runoff. This deficit is compensated by the L-THIA model, which uses the precipitation data, land use, and soil type to calculate runoff volume and depth. In addition, traditional water balance models use either average monthly or yearly climatological data. By coupling the water balance model with L-THIA (which requires daily precipitation data), it is possible to calculate the adjusted precipitation data under different land use and soil types and provide more accurate estimates of runoff based on daily climatological data.

The objective of this paper is to derive a methodology to assess the combined impacts of global warming and land use changes in the Ohio River Basin. By coupling a water balance model with a long-term hydrologic impact assessment model into a single GIS package, the efficacy of land use changes in mitigating the hydrologic effects of global warming can be evaluated. This is a pilot study of a larger research project aimed at detecting water quality impacts of global warming and land use changes.

The entire Ohio River Basin drains 548,000 km² and covers parts of 14 states, as shown in Figure 1. The principal states on the right bank are Indiana (78,300 km²), Ohio (79,470 km²), and Pennsylvania (41,980 km²). The principal states on the left bank are Kentucky (105,820 km²) and West Virginia (55,390 km²). The basin’s population is almost 25 million (or 10 percent of the U.S. population) in major cities such as Pittsburgh, Columbus, Cincinnati, Louisville, Indianapolis, and Nashville. The Ohio River begins in Pittsburgh, PA, at the confluence of the Allegheny and Monongahela Rivers. It flows 1,579 km to Cairo, IL, where it empties into the Mississippi River.

The physiography is defined by a gradual transition from the Appalachian region along the east and southeast to the glaciated plains in the north and northwest (Spencer 1965; Johnson 1975). The Appalachian region consists of the Valley and Ridge and Blue Ridge mountainous segments that form the eastern divide of the Ohio River Basin, and the Appalachian Plateaus that border the mountains on their west. The mountains consist of unglaciated, folded and faulted sedimentary formations, and some crystalline and metamorphic rocks, while the plateaus possess elevated sedimentary deposits dissected with deep canyons. Here forested regions predominate.

The region between the Ohio River and south to the Appalachian plateau contains unglaciated, horizontal or gently dipping consolidated sedimentary rocks. These are overlain by some alluvial deposits in the major tributaries. Most of the Basin that lies north of the main stem of the Ohio River consists of the relatively flat, glaciated Central Lowland region. Here agriculture predominates.

The Ohio River Basin possesses a humid climate with a northwest-southeast gradient. Summers are warm and humid. Winters range from moderately cold with few annual winter frosts in the south to severe with significant winter snowfall in the northeast. The average annual precipitation ranges from up to 2024 mm in the eastern Appalachians to about 506 mm in the northwest.

Water-related resource problems of the Ohio River Basin identified by Ohio River Basin Consortium for Research and Education (ORBCRE) are mostly pollution-oriented and include effluent from municipal waste water treatment plants, combined sewage and stormwater overflows, coal mine drainage and resulting sedimentation, urban stormwater, agricultural and forest runoff, toxic pollutants, problems from oil and gas recovery brines, reservoir eutrophication, ground water pollution, and drinking water contamination. The region has a number of hazardous waste disposal sites (Karl et al. 1996).

Because of the high precipitation and its relatively uniform distribution throughout the year, the Ohio River contributes substantial stream flow to the Mississippi. A comparison of the mean monthly stream flow of the Ohio River near its mouth at Metropolis, IL (drainage area 545,600 km²) with the Mississippi River at Vicksburg, MS (drainage area 3,065,000 km²) reveals that, although representing only 18 percent in drainage area of the entire Mississippi River Basin, the Ohio River contributes 49 percent of the total Mississippi stream flow observed at Vicksburg.

Although the Ohio River Basin is one of the most important watersheds in the continental United States in terms of its economy and water resources, it is one of the under-studied areas. Cincinnati and Columbus were chosen as the sites for detailed hydrologic analysis in the Ohio River Basin for a number of reasons. First, they represent two metropolitan areas in the Ohio River Basin where the population has been changing. Columbus is in a predominantly agricultural watershed. The metropolitan area is experiencing a positive population growth. Conversely, the Greater Cincinnati metropolitan area is experiencing a negative population growth. Suburban sprawl is more pronounced in Cincinnati and there is a positive growth trend in the suburban areas. Thus, developing sound land use policies will be important to adjust for the population changes. Second, Cincinnati and Columbus represent two of the largest cities in Ohio. Servicing such a large population will require an effective utilization of water resources. Third, these sites were chosen to discern the effects of geography on water resources impacts. Cincinnati is a city on the main stem of the Ohio River, while Columbus is on a tributary of the Ohio River, north-east of Cincinnati.

In addition, specific sites were chosen in each watershed to illustrate the effects of global climate and land use changes on runoff and water quality. These sites were chosen to represent the different land use and soil types in the watershed. In this study, six sites were chosen in the watershed containing Cincinnati
(Fig. 2)
, and four sites were chosen in the watershed containing Columbus
(Fig. 3).

Daily climatological (precipitation and air temperature) data from 1964 to 1996 for Cincinnati and Columbus were obtained from the National Climatic Data Center. The average daily air temperature was calculated from the average of the maximum daily temperature and the minimum daily temperature. The mean temperature and precipitation values were treated as the average or "base case" climatic conditions for Cincinnati and Columbus. They were further used to derive the other climatic data for the different climatic change scenarios. Current land use coverages from 1994 were obtained from Thematic Mapper data acquired by the Multi-Resolution Land Characterization (MRLC) Consortium. The land use data were classified into the following categories: water, commercial, agricultural, high-density residential, low-density residential, grass/pasture, forest, and industrial. Categories were defined using the National Land Cover Data (NLCD) Land Cover Classification system. Water was defined as areas of open water. Commercial areas were defined as infrastructure (e.g. roads, railroads, etc.) and all developed areas not classified as high-density residential. Agricultural areas were defined as pasture, hay, row crops, small grains, fallow, or urban/recreational grasses. High-density residential areas were defined as highly developed areas where people reside in high numbers. Examples include apartment complexes and row houses. Vegetation in this area accounts for less than 20% of the cover. Constructed materials account for 80% to 100% of the cover. Low-density residential areas were defined as areas with a mixture of constructed materials and vegetation. Constructed materials in these areas account for 30% to 80% of the cover. Vegetation may account for 20% to 70% of the cover. These areas most commonly include single-family housing units. Grass/pasture was defined as herbaceous grasslands and wetlands. Forest was defined as areas characterized by tree cover, which accounts for 25% to 100% of the cover. Forest includes deciduous forest, evergreen forest, and mixed forest. Industrial areas were defined as areas of bare rock, sand or clay; quarries, strip mines, or gravel pits, and transition areas (areas of sparse vegetative cover that are dynamically changing from one land cover to another, often because of land use activities).

Soil data were obtained from the National State Soil Geographic (STATSGO) database produced by the U.S. Department of Agriculture - National Resources Conservation Association. The soil data were classified into hydrologic soil groups, which describe the hydraulic conductivity of the soil (Table 1). As depicted by Darcy’s law, the water flow rate is dependent on the driving force (potential quotient) and on the ease of flow through the medium (hydraulic conductivity). Hence, hydraulic conductivity is used here as a surrogate for infiltration loss. Population data of 1990 (see Table 2) were obtained from the Bureau of the Census (1993).

Spatial coverages were merged and clipped using Arc/Info so that the entire Ohio River Basin was covered. Individual watersheds which depicted the spatial extent of the two cities were selected and clipped.

Until current GCMs improve their spatial resolution and hydrologic parametrizations, information on the hydrologic effects of global climate changes can best be obtained using regional hydrological models. Such studies are limited to sensitivity analyses that describe the vulnerability of hydrologic basins to a range of reliable climate scenarios (Nash and Gleick 1991).

Three scenarios were utilized in this study depicting changes in daily values of temperature and precipitation. They illustrated a warm and wet case scenario (+4 degrees Celsius, +20% precipitation of the current average climatic condition), a dry case scenario (+4 degrees Celsius, -20% precipitation), and the base case scenario (no change in temperature and precipitation). These scenarios were based on projections made by the IPCC and the United Kingdom Hadley Centre’s climate model (HadCM2), which indicate that temperatures in Ohio could increase by 2 to 4⁰ C and precipitation could change by -20% to +20% by the year 2010 (Karl et al. 1996; USEPA 1998).

Future land use scenarios for each city were based on demographic projections of each city. Population levels were used as a surrogate for urban growth. Regression models on the population data from 1850 to 1990 were used to predict future population levels in the year 2050. The area of land use change was predicted by multiplying the predicted year 2050 population level by a density factor that was characteristic of each city. The density factor was computed as the ratio between the 1990 population and the 1990 area provided by the 1990 U.S. Census Bureau (Table 3).

As a scenario for future land use change in Cincinnati, the existing high-density residential development was converted to low-density residential development to depict the negative population growth in the metropolitan area. This scenario was based on the recent revitalization programs in Cincinnati. For the last two decades, the city has been making a deliberate effort to pull down some old apartment and single/multi family buildings in some old communities and replace them by new townhouses and single family housing with more green space. Incentives, such as tax abatement, are used by the city to encourage land developers to redevelop and rejuvenate the city core. Hence, it is likely that in the future, many of the existing high-density residential development will be converted to low-density residential areas. The conversion occurred in the central metropolitan area, which spans approximately 200 km².

In Columbus, existing forest and agricultural areas were converted to low-density residential development to depict the positive population growth in the area. Likewise, the existing low-density residential developments were converted into high-density residential developments. This is a likely scenario, as the city of Columbus is growing at a very fast pace and some agricultural areas have been converted to low-density residential areas; and some low-density residential areas are being converted to high density residential developments. As in Cincinnati, the conversion occurred in the central metropolitan area, which spans approximately 650 km².

The Thornthwaite climatic water balance computer program was originally developed by C. Willmott (1977). It was later translated by Thompson (1992). Part of the software algorithm was rewritten using Avenue programming language, so that it could be incorporated with the L-THIA (Harbor 1994) program based on ArcView 3.1 and Spatial Analyst. The water balance model uses the air temperature, heat index, number of days in the year, elevation, soil moisture capacity, and the latitude of the site to calculate the difference between precipitation and potential evapotranspiration. This difference is then used as the new daily precipitation values for the L-THIA analysis to generate the annual average runoff depth map. In addition to the long-term daily precipitation data, L-THIA also utilizes hydrologic soil groups, curve number values, and land use data of the area. The analysis is based on the Soil Conservation Service’s curve number, or CN, method (USDA 1986). The CN is used in an empirically based formula to determine how much of a given rainfall event becomes surface runoff. The relationship between rainfall, runoff and CN value is non-linear, meaning that small changes in land use or rainfall can produce large changes in runoff. Due to its simplicity, its use of readily available watershed information, and the fact that it gives reasonable results, the CN method is often used in everyday simple stormwater management. Indeed, this method is widely accepted and has been applied to situations ranging from simple calculations of runoff from small watersheds to use in comprehensive hydrologic/water quality modeling for a regional catchment (see, e.g., Mostaghimi and Mitchell 1982). It is also built into many complex hydrological models, such as SWAT (Soil and Water Assessment Tool; Arnold et al. 1995, Srinivasan and Arnold 1994) and has been used in many sophisticated analyses. Hence, the use of the CN equation in L-THIA is a simple alternative to far more complicated hydrological models that require extensive data inputs which are often not available for most areas.

The time of the growing season (between April and August) is used to adjust the curve number values for antecedent moisture condition of soils. By using many years of climatic data in the analysis, the model focuses on average impact, rather than an extreme year or storm. From the runoff depth map, a runoff volume and an annual non-point source pollution map are created. Non-point source pollution was estimated from the runoff volume and event mean concentration non-point source data in L-THIA. The event mean concentration (EMC) is the average concentration of an urban pollutant measured during a storm runoff event. The EMC is calculated by flow-weighting each pollutant sample measured during a storm event (Chow et al. 1988). The EMC non-point source data were used to characterize runoff from the various land-use categories. In this study, the total nitrogen level per year was used as an example to illustrate the plausible changes in the patterns of non-point source pollution under different simulation conditions.

Previous investigation (McClintock et al. 1995) has shown that this method can produce reliable estimates of storm water runoff volume since the majority of the runoff produced during a year is not the result of large storm events. Instead, it is the sum of runoff generated during minor precipitation events. As the watershed becomes increasingly developed, these events will produce more frequent runoff.

Cincinnati

Six sites in the watershed were chosen to illustrate the effects of global climate and land-use changes on runoff and water quality. As expected, the wet climate scenario increased runoff the most (from the base case scenario), and the dry climate scenario decreased runoff the most (from the base case scenario). The ranking of land use types which produced the most runoff is: commercial (most runoff), high-density-residential, agricultural, low-density residential, and forest (least runoff).

In terms of non-point source pollution, the highest amounts of total nitrogen per year occurred in the commercial land use type. In the wet scenario, the total nitrogen was 14.86 mg; in the base case scenario, the total nitrogen was 10.89 mg; in the dry scenario, the total nitrogen was 6.60 mg. The lowest amounts of total nitrogen occurred in the forested areas. In the wet scenario, the total nitrogen was 0.14 mg; in the base case scenario, the total nitrogen was 0.06 mg; in the dry scenario, the total nitrogen was 0.02 mg.

Conversion of land from high-density residential use to low-density residential use decreased the amount of runoff in all climate scenarios. However, the amount of runoff varied according to hydrologic soil group and climate scenario. When the conversion from high-density to low-density residential area occurred on soil group B (Site 2), the total volume of runoff was reduced by 81.60% in the wet scenario, 84.74% in the base case scenario, and 88% in the dry scenario. When the conversion occurred on soil group C (Site 4), runoff was reduced by 73.24% in the wet scenario, 77.26% in the base case scenario, and 82.28% in the dry scenario. This shows that the dry climatic regimes will have more impacts on runoff when the land use type is changed from high to low density residential. Under this dry condition, places with soil group B will experience the highest impact with the most diminished runoff volume.

The effect of different soil types is very obvious when Site 5 (soil group B) was compared with Site 6 (soil group C). Soil group B has higher hydraulic conductivity, and thus allows more water to be infiltrated into the soil, resulting in reduced runoff. On the contrary, soil group C has lower hydraulic conductivity, resulting in less reduction in runoff. Moving from an agricultural land type on soil group C to one on soil group B reduces runoff by 51.82% in the wet scenario, 54.79% in the base case scenario, and 60.68% in the dry scenario (Table 4).

Four sites in the watershed were chosen to illustrate the effects of global climate and land-use changes on runoff and water quality. As in Cincinnati, the wet climate scenario increased runoff the most (from the base case scenario), and the dry climate scenario decreased runoff the most (from the base case scenario). The ranking of land use types which produced the most runoff is: commercial (most runoff), high-density-residential, agricultural, low-density residential, and forest (least runoff).

In terms of non-point source pollution, the highest amounts of total nitrogen occurred in the commercial land use type. In the wet scenario, the total nitrogen was 174.05 mg; in the base case scenario, the total nitrogen was 126.57 mg; in the dry scenario, the total nitrogen was 75.84 mg. The lowest amounts of total nitrogen occurred in the forested areas. In the wet scenario, the total nitrogen was 3.72 mg; in the base case scenario, the total nitrogen was 1.89 mg; in the dry scenario, the total nitrogen was 0.78 mg.

Conversion of land from low-density residential to high-density residential increased the total volume of runoff in all climate scenarios. Similar to Cincinnati, the amount of runoff change varied according to climate scenarios. There was no variation in hydrologic soil groups, because all soils in the central metropolitan area of Columbus were in hydrologic soil group C. Conversion of land from low-density residential to high-density residential increased runoff by 303.61% in the wet scenario, 376.87% in the base case scenario, and 454.10% in the dry scenario. Conversion of land from agriculture to low-density residential decreased runoff by 30.05% in the wet scenario, 32.60% in the base case scenario, and 32.97% in the dry scenario. Conversion of land from forest to low-density residential increased runoff by 91.03% in the wet scenario, 98.65% in the base case scenario, and 103.33% in the dry scenario. This finding concurs with the result from Cincinnati in that the effects of land use change are more pronounced under the dry conditions.

Changes in runoff from the following land use changes (although they did not occur in the same site) can provide insight to possible alternate management schemes in the future. For instance, comparing the runoff from an agricultural land use and a commercial land use indicates that conversion of agriculture to commercial use increases the total runoff volume by 793.43% in the wet scenario, 1063.80% in the base case scenario, and 1594.50% in the dry scenario. Furthermore, comparing the runoff from a forest land use and a commercial land use indicates that conversion from forest to commercial use increases runoff by 2340% in the wet scenario, 3377.03% in the base case scenario, and 5040% in the dry scenario (Table 5). This result is interesting as it implies that changes in land use can modify, to a certain extent, the effects of climate change on runoff.

In order to investigate the combined effects of global climate change, land use, and soil type on surface runoff and non-point source pollution, matrices ranking the level of surface runoff and non-point source pollution were created. For Cincinnati, a matrix (Table 6) was created which ranked the surface runoff depth of each land use/soil type combination and climate scenario. Across the top of the matrix is a ranking of the land use/soil type combinations based on runoff depth: forest/B soil, low-density residential/B soil, agriculture/B soil, low-density residential/C soil, agriculture/C soil, high-density residential/B soil, high-density residential/C soil, and commercial/C soil. Down the first column are the climate scenarios: dry, base, and wet. Structuring the matrix in this fashion allows one to rank the runoff depth from the lowest depth (rank 1) in the top left corner to the highest depth (rank 23) in the bottom right corner. Thus, traversing the matrix from top left to bottom right allows one to follow the increase in runoff depth as the land use, soil type, and climate scenario changes.

A similar matrix (Table 7) was constructed to depict the amount of total nitrogen (non-point source pollution) occurring in each land use/soil type combination and climate scenario. As in the runoff depth matrix, one can find the lowest total nitrogen amounts in the top left corner and the highest total nitrogen amounts in the bottom right corner. However, the ranking of the land use/soil type combination across the top of the total nitrogen matrix differs from that of the runoff depth matrix. The agriculture/B soil and low-density residential/C soil categories are juxtaposed, as are the agriculture/C soil and high-density residential/B soil categories. This might be related to the fact that agricultural land use often contributes more nitrogen than residential land use.

For Columbus, a similar pair of matrices was created. However, no soil type was considered because all of the land use changes that were projected occurred in an area defined by hydrologic soil type C. For Columbus, the runoff depth (Table 8) and total nitrogen (Table 9) matrices were very similar, in that the rankings of the land use types were identical in both cases: forest, low-density residential, agriculture, high-density residential, and commercial. As in the Cincinnati matrices, traversing the Columbus matrices from the top left corner to the bottom right corner indicates an increase in runoff depth and amount of total nitrogen as non-point source pollution. However, the ranges of the actual values differ between Cincinnati and Columbus. The highest total nitrogen in Columbus is 174.05 mg, whereas the highest total nitrogen in Cincinnati is 14.86 mg. This finding may reflect the fact that Columbus has more agricultural land use than Cincinnati. Since agriculture is one of the major sources of total nitrogen, it is reasonable to have a higher nitrogen level in Columbus.

The present study is one of the first studies to integrate models describing global climate change (Thornthwaite) and land use (L-THIA) change to examine the relative effects of climate change and land use change on surface runoff and water quality. The projected climate change was based on the IPCC and the United Kingdom Hadley Centre’s climate model (HadCM2). The projected hypothetical land use change was based on a linear extrapolation of current population data. Only a few previous studies have attempted to integrate climate models with land use models. Olejnik and Kedziora (1991) used a heat and water balance model to estimate surface runoff under changed climatological and land-use conditions. Although their heat and water balance model was well documented, their algorithm for predicting land-use change was poorly described. They only state that the plant development stage for every land use was estimated from phenological data from the National Atlas of Poland, published in 1964. They do not describe how the estimates of future land use changes were derived. In fact, they do not even claim that their land use changes are indicative of future conditions, only that they are different from current conditions.

Previous studies have reported that urban runoff is a potential source of water pollution (Pitt and Bozeman 1980) and that watershed urbanization causes poorer water quality (Tong 1990). Similar findings were noted in the current study showing that the degree of urbanization is an important factor in determining water quality. In fact, the results of the present study seem to indicate that, in these two cities of the Ohio River Basin under the simulation conditions and parameters estimated, conversion from agricultural land use to low-density residential land use may even decrease surface runoff and non-point source nitrogen pollution. This information is of great use to city planners in a growing metropolitan area, such as Columbus, where the watershed is mostly agricultural, and the positive growth rate is driving forth new residential and commercial developments in the metropolitan area. However, the results of the present study also indicate that commercial land use results in the greatest amount of runoff and non-point source pollution. Communities, such as Columbus, need to decide how to best apportion their existing land to accommodate future growth, in the threat of pending global climate change (Minner et al. 1998). Adopting Cincinnati and Columbus as case studies in this project has allowed us to examine the various possible future land use scenarios. Tools which integrate both climate models and land use models, like in this study, can provide a way for planners and resource managers to quickly predict the environmental effects of future developments.

There are many policy implications of such modeling studies. Federal agencies can use this information to promote zoning regulations which recommend certain land uses which they feel may improve water quality. This information will also be useful for city managers and planners working to develop communities around a metropolitan area. Changing land use is a long-term adaptation strategy to global warming. Employing land use changes would save money that would normally be used for structural adaptations to global warming. Improved water quality would not only affect water resources for human consumption, but it would also affect the environmental health of aquatic organisms. Thus, both human and environmental health would be affected.

The results from the present study also suggest that the hydrologic soil type can influence the amount of runoff and non-point source pollution. A comparison of the runoff depth matrix (Table 6) and the total nitrogen matrix (Table 7) reveals that there is some variability in the combined effects of land use and hydrologic soil type. The agriculture/C soil type, which was adjacent to the low-density residential/C soil type in the Cincinnati runoff depth matrix, was moved away from the low-density residential/C soil type in the total nitrogen matrix. This indicates that soil and land use may play different roles in runoff generation and non-point source pollution. Hence, the interplay between land use and soil type needs to be elucidated.

The differences in the Cincinnati-based watershed and Columbus-based watershed can be attributed to their geographical locations and net precipitation amounts. Columbus, which is on a tributary of the Ohio River northeast of Cincinnati, usually experiences lower temperatures and higher rainfall. This accounts for the greater amount of runoff and non-point source pollution in this area. The outfall of toxic and oxygen-demanding substances can cause drastic water quality changes, destroy recreation and aesthetics, and threaten biotic communities.

The runoff depth and total nitrogen matrices are useful tools for planners and policy makers in devising mitigation measures to the impending global climate change. For example, if a planner or policy maker is planning to mitigate the water quality effects of climate change, he or she would merely have to find the predicted future condition on the matrix, and then move either up or left to find mitigating land use practices (since the matrix reads from low values at the top left corner to high values at the bottom right corner). Constructing a matrix is useful in this type of study because it allows the investigator to rank the impacts from low to high when the impacting forces are multidimensional.

The impacts of global climate change are evident. However, it appears that land use changes offer a promising strategy to offset the negative water quality impacts of global climate change. For instance, the Cincinnati runoff depth matrix shows that, under the wet scenario, the annual average runoff depth in an area dominated by low-density residential development is 1.50 cm, whereas the runoff depth is 0.74 cm under the base case scenario. In order for the runoff depth to decrease to 0.74 cm, the low-density residential development should be converted to forest cover (Table 6). This matrix presents a powerful tool for policy and decision makers to use to quickly determine the water quality impacts of a number of scenarios. Soil type, land cover, and climate scenario are only three of the many possible factors that may be used in this fashion.

This paper outlines an easy methodology for simulating global changes. By coupling L-THIA with the water balance model, the likely effects of land use and climate changes can be demonstrated. As such, this tool provides a rapid assessment of water quality impacts under various land use scenarios. However, since this project is a pilot study, the results from this study are preliminary and are to be used as general guidelines for planning alternatives. More detailed hydrologic assessments need to be made to obtain more quantitatively accurate and predictive results. Future plans for developing this research include verifying and validating the model results, as well as using a more intensive hydrologic-land use model which explicitly accounts for slope using digital elevation models (DEMs), models infiltration loss, and offers a higher resolution. A land use prediction algorithm which considers adjacency measures as well as population growth will also be incorporated. Moreover, it will be important to conduct cost-benefit analyses on adaptation strategies to global climate change.

We would like to thank J. Harbor of Purdue University for the use of the L-THIA model and S. A. Thompson of Millersville University for the Thornthwaite water balance model. We also thank R. Clark and B. Lykins of the National Risk Management Research Laboratory of the U.S. Environmental Protection Agency for their helpful assistance.

Table 1

The four hydrologic soil groups, as defined by the Soil Conservation Service

Table 2

Population of Cincinnati and Columbus from 1850 to 1990 

Table 3

Regression models and density factors used in generating the future land use scenarios in Cincinnati and Columbus metropolitan areas