Assessing Risk in Operational Decisions Using Great Lakes Probabilistic Water Level Forecasts

/ A method adapted from the National Weather Service's Extended Streamflow Prediction technique is applied retrospectively to three Great Lakes case studies to show how risk assessment using probabilistic monthly water level forecasts could have contributed to the decision-mak-ing process. The first case study examines the 1985 International Joint Commission (IJC) decision to store water in Lake Superior to reduce high levels on the downstream lakes. Probabilistic forecasts are generated for Lake Superior and Lakes Michigan-Huron and used with riparian inundation value functions to assess the relative impacts of the IJC's decision on riparian interests for both lakes. The second case study evaluates the risk of flooding at Milwaukee, Wisconsin, and the need to implement flood-control projects if Lake Michigan levels were to continue to rise above the October 1986 record. The third case study quantifies the risks of impaired municipal water works operation during the 1964-1965 period of extreme low water levels on Lakes Huron, St. Clair, Erie, and Ontario. Further refinements and other potential applications of the probabilistic forecast technique are discussed.KEY WORDS: Great Lakes; Water levels; Forecasting; Risk; Decision making     

LAKE ONTARIO REGULATION UNDER TRANSPOSED CLIMATES1

ABSTRACT: The implications of Lake Ontario regulation under transposed climates with changed means and variability are presented for seasonal and annual time scales. The current regulation plan is evaluated with climates other than the climate for which it was developed and tested. This provides insight into potential conflicts and management issues, development of regulation criteria for extreme conditions, and potential modification of the regulation plan. Transposed climates from the southeastern and south central continental United States are applied to thermodynamic models of the Great Lakes and hydrologic models of their watersheds; these climates provide four alternative scenarios of water supplies to Lake Ontario. The scenarios are analyzed with reference to the present Great Lakes climate. The responses of the Lake Ontario regulation plan to the transposed climate scenarios illustrate several key issues: (1) historical water supplies should no longer be the sole basis for testing and developing lake regulation plans; (2) during extreme supply conditions, none of the regulation criteria can be met simultaneously, priority of interests may change, and new interests may need to be considered, potentially requiring substantial revision to the Boundary Waters Treaty of 1909; (3) revised regulation criteria should be based on ecosystem health and socio-economic benefits for a wider spectrum of interests and not on frequencies and ranges of levels and flows of the historical climate; and (4) operational management of the lake should be improved under the present climate, and under any future climate with more variability, through the use of improved water supply forecasts and monitoring of current hydrologic conditions.     

POTENTIAL EFFECTS OF CLIMATE CHANGE ON GROUND WATER IN LANSING,MICHIGAN1

ABSTRACT: Computer simulations involving general circulation models, a hydrologic modeling system, and a ground water flow model indicate potential impacts of selected climate change projections on ground water levels in the Lansing, Michigan, area. General circulation models developed by the Canadian Climate Centre and the Hadley Centre generated meteorology estimates for 1961 through 1990 (as a reference condition) and for the 20 years centered on 2030 (as a changed climate condition). Using these meteorology estimates, the Great Lakes Environmental Research Laboratory's hydrologic modeling system produced corresponding period streamflow simulations. Ground water recharge was estimated from the streamflow simulations and from variables derived from the general circulation models. The U.S. Geological Survey developed a numerical ground water flow model of the Saginaw and glacial aquifers in the Tri‐County region surrounding Lansing, Michigan. Model simulations, using the ground water recharge estimates, indicate changes in ground water levels. Within the Lansing area, simulated ground water levels in the Saginaw aquifer declined under the Canadian predictions and increased under the Hadley.     

Comparison of Machine Learning Models Performance on Simulating Reservoir Outflow: A Case Study of Two Reservoirs in Illinois,U.S.A.

Reservoir outflow is an important variable for understanding hydrological processes and water resource management. Natural streamflow variation, in addition to the streamflow regulation provided by dams and reservoirs, can make streamflow difficult to understand and predict. This makes them a challenge to accurately simulate hydrologic processes at a daily scale. In this study, three Machine Learning (ML) algorithms, Random Forest (RF), Support Vector Machine (SVM), and Artificial Neural Network (ANN), were examined and compared to model reservoir outflow. Past, current, and future hydrologic and meteorological data were used as model inputs, and the outflow of next day was used as prediction. Simulation results demonstrated that all three models can reasonably simulate reservoir outflow. For Carlyle Lake, the coefficient of determination and Nash–Sutcliffe efficiency were each close to one for the three models. The coefficient of determination, relative mean bias, and root mean square error indicated that the SVM performed better than the RF and ANN, but the SVM output displayed a larger relative mean bias than that from RF and ANN. For Lake Shelbyville, the ANN model performed better than RF and SVM when considering the coefficient of determination, Nash–Sutcliffe efficiency, relative mean bias, and root mean square error. The study results demonstrate that the three ML algorithms (RF, SVM, and ANN) are all promising tools for simulating reservoir outflow. Both the accuracy and efficacy of the three ML algorithms are considered to support practitioners in planning reservoir management.     

MODELING THE IMPACTS OF WATER LEVEL CHANGES ON A GREAT LAKES COMMUNITY1

ABSTRACT: Recent research that couples climate change scenarios based on general circulation models (GCM) with Great Lakes hydrologic models has indicated that average water levels are projected to decline in the future. This paper outlines a methodology to assess the potential impact of declining water levels on Great Lakes waterfront communities, using the Lake Huron shoreline at Goderich, Ontario, as an example. The methodology utilizes a geographic information system (GIS) to combine topographic and bathymetric datasets. A digital elevation surface is used to model projected shoreline change for 2050 using water level scenarios. An arbitrary scenario, based on a 1 m decline from February 2001 lake levels, is also modeled. By creating a series of shoreline scenarios, a range of impact and cost scenarios are generated for the Goderich Harbor and adjacent marinas. Additional harbor and marina dredging could cost as much as CDN $7.6 million. Lake freighters may experience a 30 percent loss in vessel capacity. The methodology is used to provide initial estimates of the potential impacts of climate change that can be readily updated as more robust climate change scenarios become available and is adaptable for use in other Great Lakes coastal communities.     

CHANGING WATER BALANCE OVER TIME IN RUSH CREEK,EASTERN CALIFORNIA, 1860–19921

ABSTRACT: Rush Creek, the principal tributary to Mono Lake, has undergone profound hydrologic modifications as a result of flow regulation for hydroelectric generation and irrigation, diversions for irrigated agriculture, and diversions for water export to the City of Los Angeles. Lower Rush Creek (the lowermost 13 km downstream of Grant Lake Reservoir) was dry by 1970, but now receives flow as a result of court-ordered efforts to restore former ecological conditions. Using available historic data and recent field measurements, we constructed the water balance for Lower Rush Creek, identifying six distinct historical periods characterized by very different patterns of gain and loss. The hydrologic patterns must be understood as a basis for modeling ecosystem response to stream-flow alteration. A gradually gaining stream under natural conditions, the advent of irrigation diversions caused the middle reaches of Lower Rush Creek to be often completely dry, while irrigation-recharged springs still maintained a baseflow in the downstream “Meadows” ranch. Increased water exports from the basin subsequently reduced irrigation and dried up the springs.     

Lake Water Levels and Associated Hydrologic Characteristics in the Conterminous U.S.

Establishing baseline hydrologic characteristics for lakes in the United States (U.S.) is critical to evaluate changes to lake hydrology. We used the U.S. Environmental Protection Agency National Lakes Assessment 2007 and 2012 surveys to assess hydrologic characteristics of a population of ~45,000 lakes in the conterminous U.S. based on probability samples of ~1,000 lakes/yr distributed across nine ecoregions. Lake hydrologic study variables include water‐level drawdown (i.e., vertical decline and horizontal littoral exposure) and two water stable isotope‐derived parameters: evaporation‐to‐inflow (E:I) and water residence time. We present (1) national and regional distributions of the study variables for both natural and man‐made lakes and (2) differences in these characteristics between 2007 and 2012. In 2007, 59% of the population of U.S. lakes had Greater than normal or Excessive drawdown relative to water levels in ecoregional reference lakes with minimal human disturbances; whereas in 2012, only 20% of lakes were significantly drawn down beyond normal ranges. Water isotope‐derived variables did not differ significantly between survey years in contrast to drawdown. Median E:I was 20% indicating that flow‐through processes dominated lake water regimes. For 75% of U.S. lakes, water residence time was less than one year and was longer in natural vs. man‐made lakes. Our study provides baseline ranges to assess local and regional lake hydrologic status and inform management decisions in changing environmental conditions.     

REGION,LANDSCAPE, AND SCALE EFFECTS ON LAKE SUPERIOR TRIBUTARY WATER QUALITY1

ABSTRACT: In 1998 and 1999, third‐order watersheds in high mature forest (HMF) and low mature forest (LMF) classes were selected along gradients of watershed storage within each of two hydrogeomorphic regions in the Lake Superior Basin to evaluate threshold effects of storage on hydrologic regimes and watershed exports. Differences were detected between regions (North and South Shore) for particulates, nutrients, and pH, with all but silica values higher for South Shore streams (p < 0.05). Mature forest effects were detected for turbidity, nutrients, color, and alkalinity, with higher values in the LMF watersheds, that is, watersheds with less that 50 percent mature forest cover. Dissolved N, ammonium, N:P, organic carbon, and color increased, while suspended solids, turbidity, and dissolved P decreased as a function of storage. Few two‐way interactions were detected between region and mature forest or watershed storage, thus threshold based classification schemes could be used to extrapolate effects across regions. Both regional differences in water quality and those associated with watershed attributes were more common for third‐order streams in the western Lake Superior drainage basin as compared with second‐order streams examined in an earlier study. Use of ecoregions alone as a basis for setting regional water quality criteria would have led to misinterpretation of reference condition and assessment of impacts in the Northern Lakes and Forest Ecoregion.     

IMPACTS OF CLIMATE CHANGE ON MISSOURI RWER BASIN WATER YIELD1

ABSTRACT: Water from the Missouri River Basin is used for multiple purposes. The climatic change of doubling the atmospheric carbon dioxide may produce dramatic water yield changes across the basin. Estimated changes in basin water yield from doubled CO₂ climate were simulated using a Regional Climate Model (RegCM) and a physically based rainfall‐runoff model. RegCM output from a five‐year, equilibrium climate simulation at twice present CO₂ levels was compared to a similar present‐day climate run to extract monthly changes in meteorologic variables needed by the hydrologic model. These changes, simulated on a 50‐km grid, were matched at a commensurate scale to the 310 subbasin in the rainfall‐runoff model climate change impact analysis. The Soil and Water Assessment Tool (SWAT) rainfall‐runoff model was used in this study. The climate changes were applied to the 1965 to 1989 historic period. Overall water yield at the mouth of the Basin decreased by 10 to 20 percent during spring and summer months, but increased during fall and winter. Yields generally decreased in the southern portions of the basin but increased in the northern reaches. Northern subbasin yields increased up to 80 percent: equivalent to 1.3 cm of runoff on an annual basis.     

BSENEFICIAL USE POTENTIAL OF DRY WEATHER FLOW IN THE LAS VEGAS VALLEY,NEVADA1

ABSTRACT: Historically ephemeral washes in the Las Vegas Valley have become perennial streams in the urbanized area, and the primary source of these perennial flows appears to be the overirrigation of ornamental landscaping and turf. Overirrigation produces direct runoff to the washes via the streets and results in high ground water levels in some areas. Elevated ground water levels result in discharge to the washes because of changes in the natural balance of the hydrologic system and construction site and foundation dewatering. In recognition of the resource potential of these flows within the Las Vegas Valley, of the potential for dry weather flows to convey pollutants from the Valley to Lake Mead, and of the need to characterize dry weather flows under the stormwater discharge permit program, the quantity and quality of dry weather flow in Flamingo Wash was investigated during the period September 1990 through May 1993. This paper focuses on the resource potential of the flow (quantity and quality) as it relates to the interception and use of this water within the Valley. Economic and legal issues associated with the interception and use of this resource are not considered here.     

APPLICATION OF A HYDROLOGIC MODEL FOR LAND USE PLANNING IN FLORIDA1

ABSTRACT: An environmental simulation model of the Upper St. Johns River Basin, Florida, has been developed in order to predict hydrologic responses under proposed management plans. Land use projections for each of 19 hydrologic planning units are provided by a linear programming analysis of agricultural activities. Inputs to the model include rainfall, runoff, evapotranspiration (ET), aquifer properties, topography, soil types, and vegetative patterns. A water balance is developed in the uplands based on infiltration, ET, surface runoff, and groundwater flow. Valley continuity is based on stage-volume relationship for inflows and outflows and a variable roughness coefficient dependent on vegetative patterns. Land use changes form the basis for predicting hydroperiod variation under alternative management schemes. Plans are ranked according to two criteria, deviation from a natural hydroperiod and flood or drought control provided. Results indicate that (1) a single reservoir without irrigation and (2) floodplain preservation plans are superior to (3) multiple reservoir with irrigation and (4) uncontrolled floodplain plans with regard to both criteria.     

MULTIOBJECTWE ALLOCATION OF WATER SHORTAGE IN THE SVARTA RIVER,SWEDEN1

ABSTRACT: A model is proposed for allocation of water shortages among competing water uses in the Svarta River basin in Sweden. The three major competing uses in the basin are hydroelectricity generation, irrigation water supply, and urban water supply. Minor uses that impact upon the allocation are minimum river flow requirements for fishlife and for dilution of treated wastewater, and storage level restrictions for recreation purposes in the main storage facility, Lake Sommen. Analysis of the competing demands on the water are modeled through the method-of-weights multiobjective technique using a deterministic mixed-integer optimization formulation. The (0–1) variables in the formulation are required to synthesize the restricted validity of permits for withdrawal of irrigation water from the river and to simulate the complex operating rules of the major regulation facility on the river. Due to the deterministic nature of the formulation, the model is used on a hydrologic scenario basis. Use of the model is demonstrated by application to the Svarta River.     

USE OF A HYDROLOGIC MODEL IN A BASIN-WIDE WATER ALLOCATION PROCEEDING1

ABSTRACT: The Montana Department of Natural Resources and Conservation developed a hydrologic model to help analyze the effects of allocating water for consumptive and instream uses in the upper Missouri River basin of Montana. The model, a PC-based FORTRAN program, uses a mass-balance approach to compute monthly streamflows, reservoir operations, hydropower production, and irrigation and municipal water uses throughout the 54,000 square mile basin for a 59-year base period. Simulation results are presented as monthly mean and percentile-exceedence values. The model was run for baseline conditions and six hypothetical water-allocation alternatives. Results were used by staff resource area specialists to assess potential impacts to water quantity and distribution, water rights, water quality, stream channel form, fisheries, wildlife, recreation, hydropower production, and economics. These analyses were presented to the public and the decision-making board in an environmental impact statement (EIS). Though, in many instances, the model did not allow for detailed, site-specific analyses, the model was an important tool and its simulation results formed the hydrologic basis for the EIS.     

SIMULATION OF SPATIAL AND TEMPORAL CHANGES IN WATER QUALITY WITHIN A HYDROLOGIC UNIT1

Water quality must be considered in the development and planning aspects of water resource management. To accomplish this, the decision-maker needs to have at his disposal a systematized procedure for simulating water quality changes in both time and space. The simulation model should be capable of representing changes in several parameters of water quality as they are influenced by natural and human factors impinging on the hydrologic system. The objective of this work is two-fold. The first goal is to demonstrate the feasibility of developing and utilizing a water quality simulation model in conjunction with a hydrologic simulation model. The model represents water quality changes in both time and space in response to changing atmospheric and hydrologic conditions and time-varying waste discharges at various points in the system. This model has been developed from and verified with actual field data from a prototype system selected for this purpose. The second aim is to set forth procedural guidelines to assist in the development of water quality simulation models as tools for use in the quality-quantity management of a hydrologic unit.     

EFFECTS OF SIMULATED CLOUD SEEDING ON STREAMFLOW OF SELECTED WATERSHEDS IN PENNSYLVANIA1

A mathematical model was developed to simulate the hydrologic behavior of five small watersheds in central Pennsylvania. Continuous hydrographs for the 6-month period, April to September 1964, were simulated. Synthesized rainfall cycles consisting of increasing rainfall by 10, 20, and 30 percent to simulate the effects of cloud seeding were processed through the watershed model to determine the effects on low flow augmentation. Other rainfall cycles used consisted of increasing every third storm by 30 percent and of developing a rainfall cycle by processing daily radiosonde data through a mathematical cumulus cloud model to obtain a prediction of rainfall following seeding. A comparison of actual and predicted hydrographs indicated that simulated cloud seeding resulted in significant monthly and seasonal water yields. In general, the results of the study appear to indicate that on a theoretical basis cloud seeding would be a feasible method of augmenting low stream-flow during the summer months on watersheds in the northern Appalachian region.     

An Application of the T‐TEL Assessment Method to Evaluate Connectivity in a Lake‐Dominated Watershed after Drought

Lakes are landscape features that influence connectivity of mass and energy by being foci for the reception, mixing, and provision of water and material. Where lake fractions are high, they influence hydrological connectivity. This behavior was exemplified in the Baker Creek watershed in Canada's Northwest Territories during a two‐year drought in which many lake levels declined below outlet elevations. This study evaluated how lakes controlled surface runoff connectivity reestablishment following the drought using a new assessment method, T‐TEL (time scales — thresholds, excesses, losses). Analysis of daily data showed that during a summer period following the drought, connectivity occurred between 0% and 41% of the time. The size of run‐of‐the‐river lakes relative to their upstream watershed area, and the upstream lake fraction, are two factors for connectivity. These terms represent a lake's ability to control the size of storage deficits relative to rainfall, and evaporation and storage losses along pathways. The connectivity magnitude–duration curve only aligned with the watershed flow duration curve during high‐water conditions, implying lakes functioned as individuals rather than as part of a perennial watercourse during much of the study. The T‐TEL method can be used to quantify consistent metrics of hydrologic connectivity that can be used for regionalization exercises and understanding hydrologic controls on material transport.     

Water Quality and Plankton in the United States Nearshore Waters of Lake Huron

Our goal in the development of a nearshore monitoring method has been to evaluate and refine an in situ mapping approach to assess the nearshore waters across the Great Lakes. The report here for Lake Huron is part of a broader effort being conducted across all five Great Lakes. We conducted an intensive survey for the United States nearshore of Lake Huron along a continuous shoreline transect (523?km) from Port Huron, Michigan, to Detour Passage. A depth contour of 20?m was towed with a conductivity-temperature depth profiler, fluorometer, transmissometer, and laser optical plankton counter. Multiple cross-contour tows (10-30?m) on the cruise dates were used to characterize the variability across a broader range of the nearshore. The cross-contour tows were comparable with the alongshore contour indicating that the 20-m contour does a good job of representing the nearshore region (10-30?m). Strong correlations were observed between water quality and spatially associated watershed land use. A repeat tow separated by several weeks investigated temporal variability in spatial patterns within a summer season. Strong correlations were observed across each variable for the temporal repeat across broad- and fine-scale spatial dimensions. The survey results for Lake Huron nearshore are briefly compared with a similar nearshore survey in Lake Superior. The biomass concentrations of lower food web components of Lake Huron were notably approximately 54-59?% of those in Lake Superior. The towed instrumentation survey supported the recent view of a change in Lake Huron to an ultra-oligotrophic state, which has been uncharacteristic in recent history.     

HOW CLIMATE UNCERTAINTY SHOULD BE INCLUDED IN GREAT LAKES MANAGEMENT: MODELING WORKSHOP RESULTS1

ABSTRACT: In two workshops, we evaluated decision analysis methods for comparing Lake Erie levels management alternatives under climate change uncertainty. In particular, we wanted to see how acceptable and effective those methods could be in a public planning setting. The methods evaluated included simulation modeling, scenario analysis, decision trees and structured group discussions. We evaluated the methods by interviewing the workshop participants before and after the workshops. The participants, who were experienced Great Lakes water resources managers, concluded that simulation modeling is user-friendly enough to enable scenario analysis even in workshop settings for large public planning studies. They felt that simulation modeling can improve not only understanding of the system, but also of the options for managing it. Scenario analysis revealed that the decision for the case study, Lake Erie water level regulation, could be altered by the likelihood of climate change. The participants also recommended that structured group discussions be used in public planning settings to elicit ideas and opinions. On the other hand, the participants were less optimistic about decision trees because they felt that the public might view subjective probabilities as difficult to understand and subject to manipulation.     

GREAT LAKES WATER RESOURCES: CLIMATE CHANGE IMPACT ANALYSIS WITH TRANSIENT GCM SCENARIOS1

ABSTRACT: The U.S. Army Corps of Engineers conducted an assessment of Great Lakes water resources impacts under transient climate change scenarios. The integrated model linked empirical regional climate downscaling, hydrologic and hydraulic models, and water resource use sub-models. The water resource uses include hydropower, navigation, shoreline damages, and wetland area. The study is unique in that both steady-state 2°CO₂ and transient global circulation model (GCM) scenarios were used and compared to each other. The results are consistent with other impact studies in that high scatter in regional climate among the GCM scenarios lead to high uncertainty in impacts. Nevertheless, the transient scenarios show that in the near-term (approximately 20 years) significant changes could occur. This result only adds to the urgency of creating more flexible and robust management of water resources uses.     

MULTIPLICATIVE,SEASONAL ARIMA MODELS FOR LAKE ERIE AND LAKE ONTARIO WATER LEVELS1

ABSTRACT: Federal agencies in the U.S. and Canada continuously examine methods to improve understanding and forecasting of Great Lakes water level dynamics in an effort to reduce the negative impacts of fluctuating levels incurred by interests using the lakes. The short term, seasonal and long term water level dynamics of lakes Erie and Ontario are discussed. Multiplicative, seasonal ARIMA models are developed for lakes Erie and Ontario using standardized, monthly mean level data for the period 1900 to 1986. The most appropriate model identified for each lake had the general form: (1 0 1)(0 1 1)₁₂. The data for each lake were subdivided by time periods (1900 to 1942;1 943 to 1986) and the model coefficients estimated for the subdivided data were similar, indicating general model stability for the entire period of record. The models estimated for the full data sets were used to forecast levels 1,2,3, and 6 months ahead for a period of high levels (1984 to 1986). The average absolute forecast error for Lake Erie was 0.049m, 0.076m, 0.091 m and 0.128m for the 1, 2,3, and 6 month forecasts, respectively. The average absolute forecast error for Lake Ontario was 0.058m, 0.095m, 0.120m and 0.136m for the 1,2,3, and 6 month forecasts, respectively. The ARIMA models provide additional information on water level time series structure and dynamics. The models also could be coordinated with current forecasting methods, possibly improving forecasting accuracy.