Lake level marker

Research

We are involved in three general areas: drought monitoring and assessment, watershed modeling, and coastal climate extension.

Drought Monitoring and Assessment

We have developed a drought monitoring tool for the Carolinas. The tool reflects the diverse requirements of regional decision makers by including a wide choice of drought indices and allowing user selection of spatial and temporal aggregation scales. We continue work to evaluate the effectiveness of this tool to communicate drought information and plan to incorporate a component to measure drought impacts systematically. We also have begun measuring drought vulnerability with a variety of economic indicators. more

Watershed Modeling

Our hydrologic and water quality modeling efforts focus on two basins: the Catawba-Wateree and the Yadkin-Pee Dee. In the former, we have examined the influence of ENSO signals on stream discharge and water quality at a variety of spatial scales. Currently, we are calibrating EPA's HSPF model for the Yadkin-Pee Dee River Basin. Our goals are to evaluate stream flow and water quality response to climate variability and change in that basin and to provide the results to stakeholders that address water supply and quality planning. This effort is linked to a new collaboration with Dr. Larry Band at the University of North Carolina. Together, we plan to measure the potential impacts of climate change and variability on water resources. more

Coastal Climate Extension

We are collaborating with North and South Carolina Sea Grant Extension on coastal climate extension work. This relationship seeks to augment the existing SG Extension team in the Carolinas with a new extension specialist -- Dr. Greg Zielinski. Dr. Zielinski will extend science-based information to coastal communities, resource managers and interest groups in North and South Carolina. He will also contribute to related research motivated by community needs and provide hands-on operational and technical support for coastal climate issues. We expect to build on the long-established work of Sea Grant to communicate science to decision makers.

Drought Monitoring and Assessment

The Carolinas Drought Monitor allows selection of over two dozen variables and drought indices. Users can choose either raw values in appropriate units, or percentile of occurrence; the latter option allows blending of multiple indices. As appropriate, the variables or indices can be displayed at daily, weekly, or monthly time steps. By default, indices are displayed in choropleth map form with options to manipulate classification method (e.g., equal interval, quantile, natural breaks) and number of classes. The tool provides an original classification scheme for each drought index and allows classification using the same intervals and color scheme as the U.S. Drought Monitor. Another classification method conforms to intervals defined in state drought legislation. As with most geographical information systems, users may overlay points (weather stations, stream gages), lines (e.g., streams), and polygons (e.g., states, counties, drought management areas, climatic divisions, USGS HUCs). Some of these map layers are available simply for display, such as basin hydrology, stream gages, and shaded relief. For most other layers, users either can view the polygon borders as background, or as an analysis layer to which gridded data are aggregated to create choropleth maps. For example, it is possible to view drought conditions for the same month at several different spatial aggregation levels. The application allows users to select multiple polygons of the same spatial unit to create a new choropleth map, showing a drought index for a group of counties or for a particular watershed not defined by one specific 2-, 4-, 6-, or 8-digit HUC. Map navigation tools allow users to zoom in and out, to pan to a particular region, and to create a new map extent or map center.

Users also can produce graphs or tables for points or areas of particular interest. They initiate this feature by selecting a point or polygon on a map or directly from a list. The default setting produces a graph or table for the most recent selection. Users may adjust the beginning and end dates for tables and graphs, toggle between the two, and adjust aggregation regions or the individual components within them (e.g., weather stations, counties, river basins).

The application provides a variety of metadata features. When mapping drought indices, a roll-over feature displays the label and value for each point and polygon. Users can view color-coded polygons and points simultaneously showing percentiles of their custom drought blend. They may also see the weather stations used to produce the interpolated grid from which polygon values were derived. This feature also identifies any points with missing data. A list of relevant stations or areas accompanies each table and graph. Help features provide users with additional information. This includes comments on the functionality of each user option, methods used in application development, error messages, specific calculations for each drought index, references to relevant literature, and caveats associated with the methods used in the production of maps, tables, and graphs.

The development of this tool involved extensive research to evaluate the methods used in the final product. We have one paper in press detailing a new method for calculation of a weekly Palmer Drought Index. A second paper outlining the tool is forthcoming in the Bulletin of the American Meteorological Society. A third manuscript evaluating spatial interpolation and aggregation is under review.

The drought monitoring tool is receiving scrutiny by drought tasks forces in both states. In the Catawba-Wateree Basin, a Drought Management Advisory Committee (DMAC) has been formed to review, update, and improve the low-inflow protocol during the new license term. The Carolinas drought monitoring tool will be an integral part of that process and, therefore, must mature in ways that address decision-makers future needs. Kirstin Dow leads an effort to understand the design needs for a new front-end user interface. She has surveyed members of the Catawba-Wateree and South Carolina DMAC about their understanding of uncertainties represented in mapping drought indicators, preferences for graphical output, and interest in expanding the tool to include weekly timescales or drought forecasts. Preliminary findings of this survey were presented in two papers at the Annual Meeting of the Association of American Geographers. The effort generated insight into stakeholders' desire for regional as well as county and basin level information and their preference for some spatial scales of information over greater accuracy.

Watershed Modeling

The current and ongoing FERC relicensing processes in the Carolinas have focused attention drivers of both water quantity and quality, including climate variability and water management decisions at local, regional, and watershed scales. Further, there is now increased awareness of interstate linkages and dependencies. North Carolina needs sufficient quantity to accommodate population growth, economic development, hydropower generation, and related concerns. South Carolina has the same needs but is also especially sensitive to water quality as rivers enter the state from North Carolina, the effects of significant alteration of streamflow, flow variability, and water quality on coastal resources and economic development.

Although FERC can be a significant catalyst for large scale planning activities, particularly during the relicensing process, there are water management issues that it is ill-suited to address. This is because of either the scope-of-control for a given licensee, the relatively static and long-term nature of a license once it is completed, or the licensee's inability to deal with climate uncertainties in a business planning context. The process does not include a comprehensive assessment of whole-basin issues such as the confounding effects of multiple resource management authorities (local, regional, state), basin-wide resource use (industry, recreational and commercial use, private property concerns), or ecological impacts (river and coastal habitat quality). Interannual climate variability is assumed to be sufficiently characterized by the available historic streamflow record, and uncertainties associated with climate change are not considered.

The two watersheds of primary concern in the Carolinas are the Santee River and Yadkin-Pee Dee River watersheds. In our earlier modeling work we focused on the Catawba River, one of the two main tributaries of the Santee River. In that basin we used a watershed simulation model of hydrology and water quality to analyze ENSO effects on stream flow and quality at three spatial scales. Our current work extends those efforts into the Yadkin-Pee Dee River watershed. We are also expanding the scope of our objectives.

Like the Santee River and its watershed, the Yadkin-Pee Dee is highly dependent on climate and interannual climate variability. It will be especially sensitive to climate change, particularly if the change results in less water or has a negative impact on quality. It is an excellent watershed in which to expand our ENSO analysis and to address watershed resource management issues at multiple spatial and temporal scales in the Carolinas.

Once the model is developed, particular attention will be given to determining the sensitivities of various model components and outputs to forcings related to climate and watershed factors that interact most strongly with climate. We are especially interested in examining the combined effect of resource management decisions and climate uncertainties. Many of these decisions are highly constrained by regulatory mandates but are also driven in large part by such things as climate and the complex interactions of economic development, public policy, and population trends. Insight into how these drivers differentially impact resources of concern, such as water quantity and quality, is potentially useful information, particularly if it can be scaled to temporal intervals that are tractable and useful for policy-makers and managers.

The model we chose for this phase is HSPF, a model that is distributed by the USEPA as part of the BASINS modeling package. The model has been used for many years in situations where it is necessary to have a spatially discrete understanding of watershed hydrology and water quality. Model implementation is a two-step process, first hydrology followed by water quality. We are currently working on the hydrology sub-model.

For modeling purposes, we divided the entire watershed into three separate sub-basins: the Black, Yadkin-Pee Dee, and Waccamaw Rivers. This was done because the downriver extents are tidally influenced so there are no streamflow data with which to calibrate the model. Integration of the three sub-basin models may occur as a later step in model development. The model is being calibrated using 1989-1993 observations and verified with 1994-1995 data. We are using PEST (Parameter ESTimation), an automated calibration package, to assist with the model calibration. As with most models of watershed hydrology, precipitation and temperature are the most important drivers in the HSPF model. These climate inputs are compiled from COOP station data set prepared by EPA for HSPF modeling. Our initial work shows that the spatial density of COOP stations is not optimal for the spatial scale of the model.

This issue is particularly apparent in the Waccamaw Basin, where model calibration is compromised because none of the COOP stations are located within the watershed boundary. In order to complete the best model parameterization, we compared three possibilities. One calibrated the model using COOP data from the closest available station. We also tested two interpolation techniques to estimate daily precipitation and temperature inside the watershed based on stations outside the watershed. The two interpolation techniques, multiple linear regression and inverse distance weight, were compared by cross-validation. We are currently developing a two-step interpolation to estimate occurrence and amount separately.

We initially calibrated the Yadkin-Pee Dee sub-basin using a single USGS gaging station located furthest downstream on the river. We are currently working to achieve greater spatial resolution in the model by calibrating upstream reaches with data from other USGS gauging stations. This was motivated partly by our need to simulate the effects of a series of hydroelectric dams on streamflow and (eventually) water quality. We are also working with the two power companies that own the dams to acquire more detailed data of actual reservoir operations.