Floods, droughts and other weather-related extremes have inflicted, and are expected to inflict, growing costs on society [1
]. According to the National Climatic Data Center of the National Oceanographic and Atmospheric Administration, over the past 30 years weather and climate-related disasters caused to communities in the U.S. total standardized losses in excess of $750 billion [6
]. In 2011 alone, the estimated total damage cost due to wildfires, droughts and unprecedented flood events in the U.S. exceeded $52 billion [6
]. Records of climatic variability and forecasts generated by climatic models provide decision-makers with the capability to assess risks of future conditions, develop scenarios, and increase resilience through practices and management options [3
]. Coupled ocean-atmosphere General Circulation Models (GCMs) can be particularly useful in understanding future climates as they are capable of simulating climatic trends over decades [8
]. Despite the advances in global climate modeling, considerable uncertainty exists with regard to regional variation and extent of future climatic impacts [9
]. Arnell and Reynard [10
] used a daily rainfall-runoff model and equilibrium vs.
transient climate change scenarios to investigate variability in river flows in twenty one catchments in Great Britain. The study estimated an average river flow increase by 20% in wet periods and roughly 20% decrease in dry periods by 2050. Monthly flows were found to exhibit greater variability than annual flows with sharp increase in streamflow over the winter months. Inter-annual change was found to be less pronounced than inter-decadal streamflow variability [10
]. Arnell [11
] incorporated the UKCIP98 climate change scenarios into a calibrated hydrological model to investigate seasonal effects on mean monthly flows and low flows. The study compared natural multi-decadal variability to anthropogenic climate change effects and predicted increases in average monthly flows accompanied by substantial decreases in low flows in headwaters [11
Rosenberg et al.
] developed a comprehensive assessment of climate change impacts by regions and sectors for the conterminous U.S. Climate data for the analysis was provided by the Hadley/United Kingdom Meteorological Office (UKMO) general circulation model (GCM; HadCM2). Water yields for various time frames between 2030 and 2095 were modelled using the Hydrologic Unit Model for the United States (HUMUS). Overall, HadCM2 projections indicated wetter than normal conditions in the Pacific Northwest and the Ohio Valley and lower than normal water yields in the Lower Mississippi and Texas Gulf basins. Seasonal changes were also predicted including increased streamflow discharges in late winter and early spring [12
]. Ficklin et al.
] used GCM-projected variations in atmospheric CO2
, temperature and precipitation to model impacts associated with climate change on evapotranspiration, water yield, streamflow, and water usage in San Joaquin Valley, California. The study predicted decrease in evapotranspiration by 37.5% and increases of water yield and stream flow by 36.5% and 23.5%, respectively. The study suggests high level of sensitivity in hydrologic endpoints with regard to potential changes in climatic conditions [13
Denault et al.
] explored the potential effect of future climate scenarios including increased rainfall intensity on urban stormwater peak discharges in a small catchment in British Columbia, Canada. The study examined the vulnerability of urban stormwater infrastructure to the effects of both urbanization and climate variability using rainfall-runoff simulations. The investigators found that upgrading existing infrastructure to projected streamflow alterations could be cost-efficient if incorporated in long-term water management planning [9
]. Dessai and Hulme [14
] argue that the implementation of successful water resource management strategies is often obstructed by the uncertainties associated with climate models predictions. The investigators suggest a framework to assess adaptation measures that are insensitive to ambiguities in climate model projections and can justify future investments in regional climate change adaptation strategies [14
In addition to climate variability, conversion of land to urban uses is recognized as a major factor contributing to alterations in watershed hydrology [13
]. Replacement of vegetation with impervious surfaces as a result of urban development affects microclimate and hydrology [16
]. Urban development tends to remove vegetation and soil, increase imperviousness, and reduce natural infiltration capacity and ability to store floodwaters [15
]. Alterations of a watershed’s hydrological characteristics due to urban development can significantly impact peak discharges, volume, and frequency of floods [13
]. Over the past two decades, cellular automata (CA) models of urban simulation found numerous applications in practically every research area in the field of urban planning [18
]. Researchers focus on the CA models in their explorations of the urban space because, for the most part, CA models are capable of conducting a number of previously intractable research tasks, such as modeling of spatial dynamics, simulation of micro-levels interactions, and capacity to predict emergent patterns [18
]. Torrent et al
] developed a meta-simulation framework capable of running simulations incorporating both coarse scale (e.g., population growth) and fine scale (e.g., space and time) sprawl driving factors. Batty [20
] explores a host of urban simulations models conceptually rooted in the complexity theory demonstrating their applicability to bottom-up stochastic temporal dynamics and phenomena associated with the processes of urban spatial evolution. Torrens et al.
] represented a set of spatial determinants of sprawl on a geographic lattice to simulate the drivers of this well-known urban phenomenon.. Overall, research has demonstrated CA models applicability to spatially explicit representations of urban processes on a spatially referenced cellular lattice at incremental time steps governed by specific transition rules [21
]. The CA model configurations have also increased in complexity. Onsted and Clarke [24
] explored the applicability of SLEUTH to model the impact of regulatory policies such as voluntary differential assessment programs where lands are excluded from particular type of urban development in exchange of tax breaks. The approach successfully represented the shifting easement dynamics and contributed to improved goodness-of-fit statistics. Tang [26
] used remote sensing and sub-cell fuzzy cellular automata to improve the accuracy of urban landscape change projections. Vancheri et al.
] coupled the cell-based dynamics of cellular automata with multi-agent systems to analyze future real estate value and population distributions. Recent studies in cellular automata linked cell state transition in land use to associated activities such as residential development and employment [28
]; and movement of capital and population [29
]. Hansen [30
] developed a scenario-based model framework which includes a land-use model, a runoff model, and a flooding screening model. Santé et al.
] provide a detailed overview of recent developments in CA modeling.
This research links advances in urban growth modeling to downscaled climate projections to derive insights into the sensitivity of hydrologic response to changes in climatic conditions and level of urbanization. Land cover change is projected using a cellular automata (CA-Markov) model incorporating Markov transition probabilities and multi-criteria evaluation (MCE) [32
]. Simulations from seven GCMs downscaled to individual Historical Climate Network (HCN) stations for the period 2010–2039 were obtained from the Consortium for Atlantic Regional Assessment (CARA) [33
]. CARA’s website specifies simulated changes in temperature and precipitation under two of the IPCC commonly used future greenhouse gas emissions scenarios–A2 associated with rapid population growth and high energy use, and B2 associated with slower population growth and moderate energy use [33
]. The sensitivity of hydrologic endpoints including mean seasonal streamflow (m3
), minimum (m3
), maximum (m3
), 100-year flood and 7Q10 low-flow (m3
) to climate drivers is examined using a calibrated hydrological model—the Hydrological Simulation Program–Fortran (HSPF) [35
]. The seasonal effects of downscaled climate projections are incorporated into the model using the Climate Assessment Tool (CAT) built within the BASINS-HSPF modeling framework [7
]. BASINS (Better Assessment Science Integrating Point and Nonpoint Sources) is a software package which combines the functionality of Geographic Information Systems (GIS) with state-of-the-art hydrological models and assessment tools to support comprehensive watershed and water-quality studies and guide the design and implementation of effective management strategies [7
This study examines the combined effect of land cover change and projected climate variability on the probability of exceeding the baseline (1971–2000) streamflow at the East Fork Little Miami River, Ohio. A cellular automata model of land cover is developed to simulate alterations in seven land cover categories in the Greater Cincinnati metropolitan area. The results from the simulation are entered into a calibrated BASINS-HSPF model for the East Fork Little Miami River watershed. Smoothed projections for the 2010–2039 mean annual and seasonal temperature and precipitation, derived from seven IPCC-endorsed GCMs, were entered into the meteorological time series for the calibrated HSPF model using BASINS-integrated Climate Assessment Tool. The potential impact of anticipated climate variability on the streamflow was examined under two IPCC scenarios: A2 (mid-high) and B2 (mid-low). Due to the wide range of projected precipitation changes, the discrete CAT output was used as random number seeds for a Monte Carlo simulation. The output from the simulation was used to estimate the probability of exceedance of the baseline (1971–2000) values. In addition, sensitivity analysis was performed using various climatic conditions, levels of imperviousness and LID practices.
The results suggest that by the year 2030 nearly 25 percent of the watershed area will be converted to urban uses if the current trends of development continue. The alterations of the landscape, including increases in impervious surfaces, are well-known to affect hydrological processes, increase runoff volume and peak discharge, and decrease base flows especially over the summer months. This study suggests that the changes in land cover and precipitation will generate various runoff scenarios. The results strongly suggest a probability of decreased low flows, especially during the summer months.
The outcomes of this research indicate that projected changes in rainfall and runoff generation will have implications for both urban stormwater management and natural systems protection and preservation. The study indicates that the short-term impacts of projected changes in temperature and precipitation combined with the effects of urbanization will result in higher probabilities of exceeding the baseline values for the 100-year flood discharges. Low impact development practices are found to affect the infiltration rates and therefore the overall amount of generated runoff. The sensitivity analysis is a useful tool that would allow stormwater managers to address insufficient conveyance capacity through routine replacement and scheduled upgrades in the future [9
]. Due to the stochastic nature of the climate processes and the uncertainty associated with modeling sub-scale variation and heterogeneity [53
] many researchers consider the outcomes of localized studies such as this as indication of potential changes in precipitation, temperature and runoff generation rather than guidelines for upgrading stormwater management infrastructure and practices [9
The anticipated changes in 7Q10 low flows and more specifically, the increased probability of having these flows decline below the seasonal minimum can have deleterious effect on the aquatic ecosystems especially during the summer months. Denault et al.
] reached a similar conclusion emphasizing the effects of urbanization and the associated increase in impervious surfaces on the reduction of the summer base flow. In addition, increased imperviousness and runoff volumes will certainly impact water quality in the affected streams of the watershed [54
]. Therefore, the results from this study bring once again to the forefront the need for future urban development planning based on understanding that innovative approaches to reduce the negative impacts of increased imperviousness will certainly contribute to mitigating potential short- and mid-term climate change effects. Furthermore, developing priorities with regard to additional data collection and environmental goals most sensitive to climate-related variables will provide the basis for future management actions. Another possible solution is to revisit and update storm water management plans to include climate- related adaptation measures. A framework based on coupling climate and urban growth model can provide the basis for a decision-support tool to investigate scenarios, evaluate management options, and track the implementation of best management practices under changing climate conditions.