Satellite SAR Remote Sensing of Great Lakes Ice Cover,Part 2. Ice Classification and Mapping

During the 1997 winter season, shipborne polarimetric backscatter measurements of Great Lakes (freshwater) ice types using the Jet Propulsion Laboratory C-band scatterometer, together with surface-based ice physical characterization measurements and environmental parameters, were acquired concurrently with Earth Resource Satellite 2 (ERS-2) and RADARSAT Synthetic Aperture Radar (SAR) data. This polarimetric data set, composed of over 20 variations of different ice types measured at incident angles from 0° to 60° for all polarizations, was processed to radar cross-section to establish a library of signatures (look-up table) for different ice types. The library is used in the computer classification of calibrated satellite SAR data. Computer analysis of ERS-2 and RADARSAT ScanSAR images of Great Lakes ice cover using a supervised classification technique indicates that different ice types in the ice cover can be identified and mapped, and that wind speed and direction can have an influence on the classification of water as ice based on single frequency, single polarization data. Once satellite SAR data are classified into ice types, the ice map provides important and necessary input for environmental protection and management, ice control and ice breaking operations, and ice forecasting and modeling efforts.     

Great Lakes ice classification using satellite C-band SAR multi-polarization data

The objective of this study is to advance development of algorithms to classify and map ice cover on the Laurentian Great Lakes using satellite C-band synthetic aperture radar (SAR) multi-polarization data. During the 1997 winter season, shipborne polarimetric backscatter measurements of Great Lakes ice types, using the Jet Propulsion Laboratory C-band scatterometer, were acquired together with surface-based ice physical characterization measurements and environmental parameters, concurrently with European Remote Sensing Satellite 2 (ERS-2) and RADARSAT-1 SAR data. This fully polarimetric dataset, composed of over 20 variations of different ice types measured at incidence angles from 0° to 60° for all polarizations, was processed and fully calibrated to obtain radar backscatter, establishing a library of signatures for different ice types. Computer analyses of calibrated ERS-2 and RADARSAT ScanSAR images of Great Lakes ice cover using the library in a supervised classification technique indicate that different ice types in the ice cover can be identified and mapped, but that wind speed and direction can cause misclassification of open water as ice based on single frequency, single polarization data. Using RADARSAT-2 quad-pol and ENVISAT ASAR dual-pol data obtained for Lake Superior during the 2009 and 2011 winter seasons, algorithms were developed for small incidence angle (< 35°) and large incidence angle (> 35°) SAR images and applied to map ice and open water. Ice types were subsequently classified using the library of backscatter signatures. Ice-type maps provide important input for environmental management, ice-breaking operations, ice forecasting and modeling, and climate change studies.     

Retrieval of ice/water observations from synthetic aperture radar imagery for use in lake ice data assimilation

High-resolution lake ice/water observations retrieved from satellite imagery through efficient, automated methods can provide critical information to lake ice forecasting systems. Synthetic aperture radar (SAR) data is well-suited to this purpose due to its high spatial resolution (approximately 50 m). With recent increases in the volume of SAR data available, the development of automated retrieval methods for these data is a priority for operational centres. However, automated retrieval of ice/water data from SAR imagery is difficult, due to ambiguity in ice and open water signatures, both in terms of image tone and in terms of parameterized texture features extracted from these images. Convolutional neural networks (CNNs) can learn features from imagery in an automated manner, and have been found effective in previous studies on sea ice concentration estimation from SAR. In this study the use of CNNs to retrieve ice/water observations from dual-polarized SAR imagery of two of the Laurentian Great Lakes, Lake Erie and Lake Ontario, is investigated. For data assimilation, it is crucial that the retrieved observations are of high quality. To this end, quality control measures based on the uncertainty of the CNN output to eliminate incorrect retrievals are discussed and demonstrated. The quality control measures are found to be effective in both dual-polarized and single-polarized retrievals. The ability of the CNN to downscale the coarse resolution training labels is demonstrated qualitatively.     

Tree-Structured Modeling of the Relationship Between Great Lakes Ice Cover and Atmospheric Circulation Patterns

Seasonal maximum ice concentration (percentage of lake surface covered by ice) for the entire Laurentian Great Lakes and for each Great Lake separately is modeled using atmospheric teleconnection indices. Two methods, Linear Regression (LR) and Classification and Regression Trees (CART), are used to develop empirical models of the interannual variations of maximum ice cover. Thirty-four winter seasons between 1963 and 1998 and nine teleconnection indices were used in the analysis. The ice cover characteristics were different for each Great Lake. The ice cover data lent itself better to CART analysis, because CART does not require a priori assumptions about data distributions characteristics to perform well. The stepwise LR models needed more variables, and in general, did not explain as much of the variance as the CART models. Two variables, the Multivariate ENSO index and Tropical/Northern Hemisphere index, explained much of the interannual variations in ice cover in the CART models. Composite atmospheric circulation patterns for threshold values of these two indices were found to be associated with above-and below-normal ice cover in the Great Lakes. Thus, CART also provided insight into physical mechanisms (atmospheric circulation characteristics) underlying the statistical relationships identified in the models.     

Great Lakes Ice Thickness Prediction

Weekly ice thickness data, collected from 24 bay, harbor, and river sites on the Great Lakes, were correlated with freezing degree-day accumulations to develop regression equations between ice thickness and freezing degree-days. The data base at ice measurement sites was 3 to 8 winters in length. Ths standard error of estimate varied for individual regression equations and averaged between 7 and 8 cm for five forms of regression equations. Because the regression equations are empirical, the range of input data used to predict ice thickness should be limited to the range of values used in the derivation.     

Wind Fields over the Great Lakes Measured by the SeaWinds Scatterometer on the QuikSCAT Satellite

This paper demonstrates the utility of satellite scatterometer measurements for wind retrieval over the Great Lakes on a daily basis. We use data acquired by the SeaWinds Scatterometer on the QuikSCAT (QSCAT) satellite launched in June 1999 to derive wind speeds and directions over the lakes at a resolution of 12.5 km, which is two times finer than the QSCAT standard ocean wind product at a resolution of 25 km. To evaluate QSCAT performance for high-resolution measurements of lake wind vectors, we compare QSCAT results with Great Lakes Coastal Forecasting System (GLCFS) nowcast wind fields and with standard QSCAT measurements of ocean wind vectors. Although the satellite results over the Great Lakes are obtained with an ocean model function, QSCAT and GLCFS wind fields compare well together for low to moderate wind conditions (4–32 knots). For wind speed, the analysis shows a correlation coefficient of 0.71, a bias of 2.6 knots in mean wind speed difference (nowcast wind is lower) with a root-mean-square (rms) deviation of 3.8 knots. For wind direction, the correlation coefficient is 0.94 with a very small value of 1.3° in mean wind direction bias and an rms deviation of 38° for all wind conditions. When excluding the low wind range of 4–12 knots, the rms deviation in wind direction reduces to 22°. Considering QSCAT requirements designed for ocean wind measurements and actual evaluations of QSCAT performance over ocean, results for high-resolution lake wind vectors indicate that QSCAT performs well over the Great Lakes. Moreover, we show that wind fields derived from satellite scatterometer data before, during, and after a large storm in October 1999, with winds stronger than 50 knots, can monitor the storm development over large scales. The satellite results for storm monitoring are consistent with GLCFS nowcast winds and lake buoy measurements. A geophysical model function can be developed specifically for the Great Lakes using long-term data from satellite scatterometers, to derive more accurate wind fields for operational applications as well as scientific studies.     

Origin and Evolution of the Great Lakes

This paper presents a synthesis of traditional and recently published work regarding the origin and evolution of the Great Lakes. It differs from previously published reviews by focusing on three topics critical to the development of the Great Lakes: the glaciation of the Great Lakes watershed during the late Cenozoic, the evolution of the Great Lakes since the last glacial maximum, and the record of lake levels and coastal erosion in modern times.The Great Lakes are a product of glacial scour and were partially or totally covered by glacier ice at least six times since 0.78 Ma. During retreat of the last ice sheet large proglacial lakes developed in the Great Lakes watershed. Their levels and areas varied considerably as the oscillating ice margin opened and closed outlets at differing elevations and locations; they were also significantly affected by channel downcutting, crustal rebound, and catastrophic inflows from other large glacial lakes.Today, lake level changes of about a 1/3 m annually, and up to 2 m over 10 to 20 year time periods, are mainly climatically-driven. Various engineering works provide small control on lake levels for some but not all the Great Lakes. Although not as pronounced as former changes, these subtle variations in lake level have had a significant effect on shoreline erosion, which is often a major concern of coastal residents.     

An algorithm to retrieve chlorophyll,dissolved organic carbon,and suspended minerals from Great Lakes satellite data

An algorithm that utilizes individual lake hydro-optical (HO) models has been developed for the Great Lakes that uses SeaWiFS, MODIS, or MERIS satellite data to estimate concentrations of chlorophyll, dissolved organic carbon, and suspended minerals. The Color Producing Agent Algorithm (CPA-A) uses a specific HO model for each lake. The HO models provide absorption functions for the Color Producing Agents (CPAs) (chlorophyll (chl), colored dissolved organic matter (as dissolved organic carbon, doc), and suspended minerals (sm)) as well as backscatter for the chlorophyll, and suspended mineral parameters. These models were generated using simultaneous optical data collected with in situ measurements of CPAs collected during research cruises in the Great Lakes using regression analysis as well as using specific absorption and backscatter coefficients at specific chl, doc, and sm concentrations. A single average HO model for the Great Lakes was found to generate insufficiently accurate concentrations for Lakes Michigan, Erie, Superior and Huron. These new individual lake retrievals were evaluated with respect to EPA in situ field observations, as well as compared to the widely used OC3 MODIS retrieval. The new algorithm retrievals provided slightly more accurate chl values for Lakes Michigan, Superior, Huron, and Ontario than those obtained using the OC3 approach as well as providing additional concentration information on doc and sm. The CPA-A chl retrieval for Lake Erie is quite robust, producing reliable chl values in the reported EPA concentration ranges. Atmospheric correction approaches were also evaluated in this study.     

Effects of Great Lakes Water Loading upon Glacial Isostatic Adjustment and Lake History

Over the last century geological studies of the ancestral Great Lakes have confirmed that the large surface load of the Laurentide ice sheet deformed the region causing tilting of ancient lake shorelines. Models of this glacial isostatic adjustment mechanism have promoted understanding of this process but have only included ice sheet loads as the source of earth deformation in the region. We describe a method, utilizing a model of glacial isostatic adjustment combined with GIS, that recreates the paleohydrology of the Great Lakes. Predictions include the extent of late glacial, postglacial, and Holocene lakes and their associated outlets and bathymetries. This predicted history of the Great Lakes is similar to that obtained from a century of detailed field studies but our method uses only the present digital elevation model, a prescribed ice sheet chronology, and an assumed earth viscoelastic rheology. Ancient lake bathymetry predictions provide an estimate of water loads associated with each lake. The effect of these lake loads upon vertical deformation of the Great Lakes region is shown to be small, less than 15 m, but not insignificant when compared to approximately 150 m of deformation forced by ice and ocean loads. Maximum lake-induced deformation is centered upon Lake Superior where water depths were greatest. Where topography is low relief, prediction of shoreline locations should include the lake loading effect as well as the ice and ocean loads.     

Ecological data as a resource for invasive species management in U.S. Great Lakes coastal wetlands

The coordinated use of ecological data is critical to the proper management of invasive species in the coastal wetlands of the Laurentian Great Lakes. Researchers and government programs have been increasingly calling for the use of data in management activities to increase the likelihood of success and add transparency in decision making. Web-enabled databases have the potential to provide managers working in Great Lakes coastal wetlands with relevant data to support management decisions. To assess the potential value of these databases to managers in Laurentian Great Lakes states, we surveyed wetland managers to determine their current data usage as well as their future data interests and catalogued the online databases currently available. Surveys were disseminated via email to managers in 56 different organizations overseeing invasive species management efforts in Great Lakes coastal wetlands; 46 responses were included in this analysis. Of the survey respondents, all reported using raw biotic data for decision making, (i.e. presence of target species) but many indicated that they would prefer to incorporate a greater variety of data, as well as more complex information. Our survey found that managers used web-enabled databases, but most databases that we catalogued only provided presence data for wetland biota. We concluded that databases can provide the types of data sought by invasive species managers but have unmet potential to be integrated into responsive management processes.     

Impacts on biodiversity and species changes in the lower Great Lakes

The Great Lakes basin ecosystem evolved after the retreat of the last ice sheet, about 10 000 years ago. Today, the complex of species present in the Great Lakes and much of the visible landscape bears little resemblance to that found some 400 years ago. Rather, the effects of various aspects of human development have caused major changes in the natural biodiversity. Lessons learned in the lower Great Lakes are applicable to many lakes around the world that have undergone agricultural, industrial and urban development in their drainage basins and have become managed, artificial ecosystems.     

Environmental exposure of freshwater mussels to contaminants of emerging concern: Implications for species conservation

Contaminants of emerging concern (CECs) are prevalent in aquatic landscapes and may be a factor in population declines of aquatic and terrestrial fauna. Yet, there are limited data to assess the impacts of CECs to species. Understanding CEC impacts is particularly important for imperiled freshwater mussels which provide valuable ecosystem services. CEC exposure of freshwater mussels was characterized by evaluating sites with and without the federally endangered mussel (Villosa fabalis) in three subwatersheds of the Maumee River, Ohio, USA, a tributary of the Great Lakes Basin. Analyses of water, sediment, and tissue concentrations of two common mussels (Eurynia dilatata and Lampsilis cardium) indicated different CEC exposures across all 6 sites. Distinct CEC signatures were found across the three media types suggesting as mussels interact with water and sediment they may be experiencing different exposure concentrations and mixtures of CECs at different life stages. Of the 83 CECs which were detected, agricultural CECs dominated sediments, pharmaceuticals were common in tissues and water, and 16 of the 83 CECs were found co-occurring in mussel tissue, water, and sediment. There were no species differences in the CEC signatures indicating all mussels, including species of concern, may be experiencing similar exposure. Comparisons to known CEC standards indicate some exceedances in the Maumee watershed including locations of federally listed mussel species. This study provides evidence of the complexity of CEC mixes in a Great Lakes watershed and the need to understand how CECs impact declining aquatic fauna.     

Development of the Great Lakes Ice-circulation Model (GLIM): Application to Lake Erie in 2003–2004

To simulate ice and water circulation in Lake Erie over a yearly cycle, a Great Lakes Ice-circulation Model (GLIM) was developed by applying a Coupled Ice-Ocean Model (CIOM) with a 2-km resolution grid. The hourly surface wind stress and thermodynamic forcings for input into the GLIM are derived from meteorological measurements interpolated onto the 2-km model grids. The seasonal cycles for ice concentration, thickness, velocity, and other variables are well reproduced in the 2003/04 ice season. Satellite measurements of ice cover were used to validate GLIM with a mean bias deviation (MBD) of 7.4%. The seasonal cycle for lake surface temperature is well reproduced in comparison to the satellite measurements with a MBD of 1.5%. Additional sensitivity experiments further confirm the important impacts of ice cover on lake water temperature and water level variations. Furthermore, a period including an extreme cooling (due to a cold air outbreak) and an extreme warming event in February 2004 was examined to test GLIM's response to rapidly-changing synoptic forcing.     

Dividing the waters: The case for hydrologic separation of the North American Great Lakes and Mississippi River Basins

Legislation has been introduced this year in the U.S. Congress, but not yet enacted, that would direct the U.S. Army Corps of Engineers to complete a study of the options that would prevent the spread of aquatic nuisance species between the Great Lakes and Mississippi River Basins. Hydrologic separation is the only option which closes the aquatic connection between the two basins and does not require continuous operation and maintenance of various technologies that have some risk of failure. The one-time, capital cost to separate the two basins is widely acknowledged to be high, and the outstanding question is whether the costs are justified given the significant risk of future ecological damages and long-term economic losses. Interests opposing separation have mounted a public campaign that the news media have picked up to deny that hydrologic separation should be considered or that a problem even exists. The campaign rests on four assertions: (1) existing electric barriers in the Chicago canals are effective; (2) it is too late--the carps are already in the Great Lakes or soon will be; (3) Asian carps will not thrive in the Great Lakes due to inadequate food and spawning habitat; and (4) Asian carps are unlikely to cause serious harm. Our review of these assertions and the ecological and socio-economic threats to both basins supports our recommendation that the pending legislation be passed and that it include analysis of hydrologic separation of the two basins.     

A Statistical Representation of Wind Events in the Great Lakes

We have analyzed hourly observations of overtake wind speed obtained from the eight National Data Buoy Center buoys deployed in the Great Lakes to determine the long-term (10 to 12 year) statistical characteristics of wind events over the lakes. We find that a generalized double-exponential model accurately describes both the return period and the duration of the wind events observed at each buoy, with only slight spatial variations in the model parameters. Wind observations made at several shore stations on Lake Michigan and filtered through a simple overland-overlake wind model confirm the model parameter values determined from the buoy measurements. The generalized function will be useful for engineering and climatological studies of the effects of winds on the Great Lakes.     

Researcher disciplines and the assessment techniques used to evaluate Laurentian Great Lakes coastal ecosystems

The Laurentian Great Lakes of North America have been a focus of environmental and ecosystem research since the Great Lakes Water Quality Agreement in 1972. This study provides a review of scientific literature directed at the assessment of Laurentian Great Lakes coastal ecosystems. Our aim was to understand the methods employed to quantify disturbance and ecosystem quality within Laurentian Great Lakes coastal ecosystems within the last 20 years. We focused specifically on evidence of multidisciplinary articles, in authorship or types of assessment parameters used. We sought to uncover: 1) where Laurentian Great Lakes coastal ecosystems are investigated, 2) how patterns in the disciplines of researchers have shifted over time, 3) how measured parameters differed among disciplines, and 4) which parameters were used most often. Results indicate research was conducted almost evenly across the five Laurentian Great Lakes and that publication of coastal ecosystems studies increased dramatically ten years after the first State of the Great Lakes Ecosystem Conference in 1994. Research authored by environmental scientists and by multiple disciplines (multidisciplinary) have become more prevalent since 2003. This study supports the likelihood that communication and knowledge-sharing is happening between disciplines on some level. Multidisciplinary or environmental science articles were the most inclusive of parameters from different disciplines, but every discipline seemed to include chemical parameters less often than biota, physical, and spatial parameters. There is a need for an increased understanding of minor nutrient, toxin, and heavy metal impacts and use of spatial metrics in Laurentian Great Lakes coastal ecosystems.     

Genetic diversity of lake whitefish in lakes Michigan and Huron; sampling,standardization, and research priorities

We combined data from two laboratories to increase the spatial extent of a genetic data set for lake whitefish Coregonus clupeaformis from lakes Huron and Michigan and saw that genetic diversity was greatest between lakes, but that there was also structuring within lakes. Low diversity among stocks may be a reflection of relatively recent colonization of the Great Lakes, but other factors such as recent population fluctuation and localized stresses such as lamprey predation or heavy exploitation may also have a homogenizing effect. Our data suggested that there is asymmetrical movement of lake whitefish between Lake Huron and Lake Michigan; more genotypes associated with Lake Michigan were observed in Lake Huron. Adding additional collections to the calibrated set will allow further examination of diversity in other Great Lakes, answer questions regarding movement among lakes, and estimate contributions of stocks to commercial yields. As the picture of genetic diversity and population structure of lake whitefish in the Great Lakes region emerges, we need to develop methods to combine data types to help identify important areas for biodiversity and thus conservation. Adding genetic data to existing models will increase the precision of predictions of the impacts of new stresses and changes in existing pressures on an ecologically and commercially important species.     

Deformities,PCBs, and TCDD-Equivalents in Double-Crested Cormorants (Phalacrocorax auritus) and Caspian Terns (Hydroprogne caspia) of the Upper Great Lakes 1986–1991: Testing a Cause-Effect Hypothesis

Deformities have been reported in many species of colonial waterbirds from several localities on the Laurentian Great Lakes. The hypothesis that deformities were caused by either polychlorinated biphenyls (PCBs) or contaminants measured as 2,3,7,8-tetrachlorodibenzo-p-dioxin equivalents (TCDD-EQs) is tested in this review of available data on concentrations of contaminants in eggs and observed deformities in embryos and chicks of double-crested cormorants (Phalacrocorax auritus) and Caspian terns (Hydroprogne caspia) between 1986 and 1991. Hatched chicks, live and dead eggs retrieved from 37 colonies in the upper Great Lakes were assessed for gross anatomical deformities. Rates of embryo death from seven regions of the upper Great Lakes were measured annually between 1986–1991. Half the embryos found dead in eggs were deformed. Nineteen types of abnormalities or deformities were observed. Subcutaneous edema in cormorants and gastroschisis in terns were the most common abnormalities in live or dead eggs. One of ten crossed-billed cormorant embryos survived to hatch. No bill-deformed terns hatched, although tern embryos had a greater rate of crossed-bills than cormorants. The suite of deformities and abnormalities found was similar to that produced in chickens by exposure to planar polychlorinated biphenyl (pPCB) and dioxin congeners. Hatching and deformity rates were correlated with concentrations ofpPCBs and TCDD-EQs. Planar PCB congeners that contributed most of the TCDD-EQs were present at concentrations sufficient to cause the observed effects. TCDD-EQs measured by H4IIE rat hepatoma cell 7-ethoxyresorufin O-deethylase (EROD) bioassay were highly correlated with deformity rates observed in cormorant chicks, live and dead eggs, and egg death rates. Similar correlations of TCDD-EQs with deformity rates were found in hatched tern chicks, dead eggs, and egg death rates, but not in live eggs. TCDD-EQs were more highly correlated to deformity and embryo death rates than total PCBs. The weight of evidence and these data are sufficient to reject the null hypothesis that there is no causal relationship between the incidence of deformities in cormorants and terns and exposure to planar halogenated compounds measured as TCDD-EQs or total PCBs in the Great Lakes.     

Predicting physical and geomorphic habitat associated with historical lake whitefish and cisco spawning locations in Lakes Erie and Ontario

The Great Lakes basin was historically populated by multiple, coevolved coregonine species, but much of that diversity has been lost. In Lakes Erie and Ontario, both lake whitefish (Coregonus clupeaformis) and cisco (Coregonus artedi) occurred in high numbers before habitat degradation, overfishing, invasive species, and other factors caused significant declines. There is growing interest in restoring these populations, and suggested actions include restoration of critical habitats such as spawning habitat. Unfortunately, our current understanding of lake whitefish and cisco spawning habitat characteristics and locations in these lakes is limited. To highlight areas of potential importance for conservation and restoration, we used random forest models and data on historical spawning locations to predict lake whitefish and cisco spawning habitats based on hypothesized key factors including wind fetch, ice cover duration, distance from 1st and 6th order tributaries, and lake bottom substrate. Our model accurately predicted spawning habitat locations for 71% and 54% of cases for lake whitefish and cisco, respectively. Fetch was the most important variable in the lake whitefish model, with spawning habitats being most likely to occur in regions of low to moderate fetch. Cisco spawning habitats were most likely to occur in areas of relatively low fetch near a 1st order stream. We used these models to predict spawning habitat locations for both species across Lakes Erie, Ontario, and St. Clair. Our results improve our understanding of lake whitefish and cisco spawning habitat characteristics and will aid in the spatial prioritization of actions to restore these native fishes.     

Milestones of Research on the Physical Limnology of the Great Lakes

Two important themes of research on the physical limnology of the Great Lakes during the past 20 years have been: (1) the controlling influence of bottom topography on circulation and (2) the boundary layer character of circulation. The first is highlighted by the generation of topographic gyres by wind and by the propagation of vortex wave modes; the second by the distinct current-climatology of the coastal boundary layer and the spectacular upwelling and downwelling events occurring there. Both themes developed as advances in theoretical insight could be quantitatively coupled to observational evidence, the latter arising predominantly from the International Field Year on the Great Lakes.