Hydrologic Impacts, Spatial Simulation
Spatial simulation of future hydrologic impacts involves deterministic or probabilistic modeling approaches that attempt to simulate likely changes in hydrology, and subsequent hydrologic response (impacts) of these changes, for a particular study area. The modeling approach might be focused on understanding the spatial impacts of predicted hydrologic changes (e.g., rainfall intensity), and/or changes in parameters impacting rainfall-runoff response and flow routing (e.g., changing land use). The goal is to produce spatial (map) and other data outputs that can assist planners and managers better understand the spatial ramifications of an uncertain future. Where appropriate and possible, estimates of uncertainty should be embedded in the map output. This information might be used to develop more informed and hence effective land use plans, flood mitigation strategies, or management strategies for habitat. The approach utilizes recent technological advances in the geospatial and hydrologic sciences to develop predictive modeling applications that can help environmental planners and managers better understand the spatial implications of hydrological processes changing in response to issues such as climate change and rapid urban development.
The approach has been stimulated by the development and utilization of spatially distributed parameters for hydrologic modeling brought on by advances in computing and geographic information systems (GIS). GIS have increasingly become a valuable management tool, providing an effective infrastructure for managing, analyzing, and visualizing disparate datasets related to soils, topography, land use, land cover, and climate (Liao and Tim 1997; Miller et al. 2004). The integration of GIS with hydrologic and hydraulic models as a data pre/postprocessor have simplified data management activities by enabling relatively easy and efficient extraction of multiple modeling parameters at the watershed scale (Ogden et al. 2001).
Some methods for integrating hydrologic models with GIS have been categorized as “loose,” “close,” or “tight” coupling (Liao and Tim 1997). Loose coupling methods usually involve data exchange using ASCII or binary data formats. An interface program is normally used to convert and organize GIS data into a format required by the hydrologic or land use model. Advantages of loose integration include ease of development and use with a wide range of commercial GIS software. Close coupling incorporates slight modifications to the control programs in the GIS software, providing improved data transfer between the model and the GIS database. There tends to be overlap between loose and close coupling methods. However, the close coupling method passes information between the GIS and the model via memory-resident data models rather than external files. This enhancement leads to improved model interactions and performance (Di Luzio et al. 2004). Tightly coupled model integration focuses on incorporating the functional components of one system within the other (i.e., the model within the GIS program). The GIS and model are no longer maintained separately. They instead share processes and data in order to reduce the redundancy in development and operation. This approach eliminates the use of interface modules and transfer files, and thereby promotes better system performance (Liao and Tim 1997).
Initial attempts to link existing hydrologic models with GIS utilized loose coupling, and frequently involved manual as well as automated parameter development. For example, Warwick and Haness (1994) used Arc/Info to determine hydrologic parameters directly for the Hydrologic Engineering Center-1 model (HEC-1), while separate line coverages defining the runoff routing were created manually. Suwanwerakamtorn (1994) derived semi-distributed hydrologic modeling using GIS for the management of watersheds and assessed the effect of land-use change using an integrated approach with HEC-1 and ILWIS (Integrated Land and Water Information System). The ability of the model to simulate future and past flood hydrographs based on hypothetical future as well as historical land-use conditions were demonstrated. The results of simulation runs demonstrated that for the study area, when forest area was reduced, more runoff would occur in every sub-catchment and also at the outlet.
Water resource planners and managers can be required to provide information to facilitate preparation for future flood hazard situations and to develop responsible regulations for sound floodplain development. If already in existence, historical hydrologic flow records, stage, and precipitation records can sometimes satisfy these planning needs. However, in many cases watershed runoff must be predicted to provide the needed information to support these preparatory and regulatory decisions. For example, a flood-damage reduction study may require an estimate of the increased volume of runoff that will result due to proposed changes to land use within a watershed. Unfortunately, no data record is available because the land use change has not yet occurred. Waiting to observe the future hydrologic impacts of proposed land use changes could result in losses of property, and even human life.
An alternative to “waiting and observing” is to use a hydrologic or hydraulic mathematical model to provide the information to aid decision-making processes that may impact future hydrologic responses of a given area (United States Army Corps of Engineers 2001). Caution is required when applying mathematical models to describe complex hydrologic systems. Models are only approximate representations of these complex natural processes, and usually incorporate the assumptions of model developers who have attempted to define the critical processes occurring within the watershed, and to develop relationships between these processes. Models may involve oversimplifications, or conversely over-specification, of hydrologic processes, which may or may not be valid for other site-specific applications due to the uniqueness of the study area for which they were developed (Kalin and Hantush 2003). Careful hydrologic model selection is therefore critical.
However, to gain a precise and complete understanding of the hydrologic impacts that land use change may impose upon every location within a watershed would require impossibly high degrees of data collection. This level of data collection would be financially unfeasible at the watershed scale (Santhi et al. 2001). Hydrologic and hydraulic models represent the most practical and realistic means to examine the flow of water on a watershed scale, in a manner allowing planners and managers to determine the effects of different land use scenarios over long time periods (Di Luzio et al. 2004; Santhi et al. 2001). Moreover, the application of hydrologic and hydraulic models allows users to identify high priority areas, valuable information when allocating scarce resources to, for example, flood mitigation efforts (Srinivasan et al. 1998).
A general approach to simulating the spatial variability of hydrologic impacts typically involves a multi-step procedure that might involve several or all of these components: (1) creation of an accurate hydrologic component representation of the watershed area; (2) determination of values for hydrologic modeling parameters and associated calibration for a specific storm, or range of storm events (which may incorporate predicted changes in magnitude and frequency); (3) generation of land use distribution patterns for initial and target years based upon a selected comprehensive land use policy, or forestry cover for initial and target years based upon a selected harvesting strategy; (4) use of forecasted land use/forestry cover datasets as inputs into the calibrated hydrologic model to generate storm hydrographs for target time periods; (5) terrain modeling of the receiving channel(s) (using high resolution terrain data, such as LiDAR, where available); (6) mapping of hydrologic/hydraulic data outputs, representing uncertainty when possible.
Worldwide, climate change and climate variability is causing significant and often unpredictable impacts on the hydrological system. In many locations, the risk of and vulnerability to floods is increasing due to changes in rainfall patterns and increased frequency of large events. In others, agricultural industries are threatened by prolonged droughts. Potential impacts may be exacerbated by rapid changes in land cover, development into flood-prone areas as a result of socio-economic pressure, and increasing municipal demands for water traditionally utilized by agriculture.
An increasing amount of research has been focused upon the modeling of future land use change via urbanization and the associated changes in hydrologic/hydraulic responses to precipitation events under increased amounts of urban land uses. In the US, data collected through the National Resources Inventory have shown that from the periods 1982–1992 to 1992–2001, the annual rate of land conversion has nearly doubled (United States Department of Agriculture-Natural Resources Conservation Service, 2003). Studies have revealed that conversions of this type can alter a watershed’s response to precipitation events, leading to increased volumes of surface water runoff, greater incidence of flooding, altered downstream river channel geometry through erosion, and the degradation of aquatic habitat of fish and other biota. Recent applications that integrate GIS technologies, land use forecasting models, and hydrologic models (e.g., Beighley et al. 2003 and McColl and Aggett 2006) have presented an opportunity to examine watershed response to precipitation under different future land use patterns reflective of various land use planning policies under consideration. These applications initially develop a hydrologic model that is calibrated to a historical precipitation event. Hydrologic parameter data required to run the hydrologic model is obtained for the study area, coinciding with the precipitation event date. These parameters include slope, soil type, land cover type, and river-channel slope and cross-sectional geometry, each of which can be readily extracted from available geospatial datasets (e.g., digital elevation models, digital soil and land cover datasets). River discharge volumes are computed by a hydrologic model at specified points within the stream network, which can then be compared to observed storm event discharge volumes. The ability to compare observed storm event discharge volumes to simulated volumes enables modelers to determine whether the model is reasonably characterizing the storm event and its associated runoff volumes.
The main goal of spatial simulation of future hydrologic impacts is to produce maps that can assist planners and managers better understand the potential spatial ramifications of an uncertain hydrologic future. Moglen et al. (2003) have stressed that ‘truly intelligent smart growth’ ought to be quantifiably superior to other proposed land development plans. However, because there are many data uncertainties embedded in this general approach (including uncertainties in the input hydrometeorologic, hydrologic and terrain data), there is a requirement to develop methods that can convey this uncertainty to limit misuse of map products. This concern is chiefly focused on three demands: (i) a fundamental scientific obligation to describe how close modeled information may be to a ‘truth’ it is intended to represent; (ii) a requirement that individual and agency reputations be protected, particularly when geographic information is used to support administrative decisions subject to appeal; and (iii) protection against litigation by those who might, in the future, allege to have experienced damage through planning based on map products of deficient quality. Where appropriate and possible, estimates of uncertainty should thus be embedded in the map output. For example, in mapping outputs from various hydraulic models both Legleiter and Goodchild (2005) and Aggett and Wilson (2007) point out the utility of fuzzy logic for incorporating uncertainty. Fuzzy logic addresses one of the fundamental limitations of spatial databases, namely its reliance on either/or logic (Burrough 1996; Hunter and Goodchild 1996). Fuzzy logic allows for the existence of transition zones between features using a function that assigns boundary components a grade of membership which corresponds to the degree of similarity between representative components of each feature. de Bruin (2000) used probabilities to determine the membership functions for fuzzy membership classes. Swan and Aggett (2007) utilize this strategy to represent the inundation extent of the modeled flows with some indication of variability. In their research, each individual cell in the depth layer required a membership value based on its depth. In deriving the membership function for the layer, the highest membership values were assigned to the cells containing the deepest inundation depths, centering the function around the maximum depth. In a similar fashion, fuzzy logic can also represent transition zones through some indication of uncertainty or error in the dataset. For example, the membership function could only be applied to areas of questionable credibility. In modeling future flood hazards, these areas might typically be transition zones between features such as the area dividing the wetted and dry extents of the channel.
- Aggett GR, Wilson JP (2007, in review) Creating and coupling a high-resolution DEM with a 1-D hydraulic model in a GIS for scenario-based assessment of avulsion in a gravel-bed river. GeomorphologyGoogle Scholar
- Burrough PA (1996) Natural objects with indeterminate boundaries. In: Burrough PA, Frank AU (eds) Geographic objects with indeterminate boundaries. Taylor & Francis, Ltd., London, p 345Google Scholar
- Kalin L, Hantush MM (2003) Evaluation of sediment transport models and comparative application of two watershed models (U.S. Environmental Protection Agency, National Risk Management Research Laboratory Paper EPA/600/R-03/139). U.S. Environmental Protection Agency, CincinnatiGoogle Scholar
- Suwanwerakamtorn R (1994) GIS and hydrologic modeling for the management of small watersheds. Int J Appl Earth Obs Geoinf (ITC) 4(1994):343–348Google Scholar
- Swan D, Aggett G (2007, in review) Two-dimensional modeling of stream channel dynamics in the Naches river, Washington. Trans GISGoogle Scholar
- United States Army Corps of Engineers (2001) Hydrologic modeling system HEC-HMS: user’s manual. USACE Hydrologic Engineering Center, Washington, DCGoogle Scholar