Assessment of Management Strategies for a Straightened Lowland Agricultural Stream in Southwestern Québec
Channel straightening and dredging were extensively used in the 20th century to improve agricultural fields' drainage efficiency and facilitate crop maintenance and harvest. Although the adverse geomorphologic and ecological effects of this practice on hydrologic networks are widely acknowledged, alternative management strategies remain marginal in Southwestern Québec. Furthermore, bank stabilization projects are often carried out to mitigate erosion at a local scale with little concern about the watershed scale, i.e. treating the symptom rather than the cause of erosion, and with little guidance on suitable designs for specific areas. We present here the results of a case study conducted in a small straightened agricultural stream in the St. Lawrence Lowlands (Richer stream, 45 km east of Montreal, Qc) to assess the efficiency of diverse instream structures to enhance bank stability. The loss in sinuosity and resultant increase in stream power in this stream resulted in acute erosion problems. A conceptual model developed in Java was developed to examine erosion processes at the watershed scale. This model is also used to parameterize a three-dimensional computational fluid dynamics model (Phoenics) to investigate the effects on flow field of employing various types of instream structures in a section of the Richer Stream which is particularly problematic for river management because of residential development and very limited space available for riparian zones. Model validation was achieved using flow velocity measurements obtained with an Acoustic Doppler Velocimeter. Simulation outcomes are analysed with regards to strategies using geomorphological, economical and ecological criteria in an attempt to identify their efficiency. The methods presented here can help decreasing the high degree of uncertainty generally involved in restoration activities by better assessing the efficiency of specific stabilization techniques prior to their implementation while considering unique stream/watershed characteristics and ecological factors.
Three-Dimensional Numerical Modelling of Flow and Sediment Transport for Field Scale Application of Stream Barbs at Sawmill Creek, Ottawa
Stream barbs (a type of submerged groyne or spur dike) are low-profile linear rock structures that prevent the erosion of stream banks by redirecting high velocity flow away from the bank. Stream barbs are becoming a popular method for stream bank protection as they can be built at a relatively low cost and provide added ecological benefit. The design and construction of stream barbs in Sawmill Creek, a small urban stream in the city of Ottawa, Canada, will serve as a demonstration project for the use of barbs as a bank stabilization technique that will contribute to the rehabilitation of urban creeks while reducing erosion threats to property and infrastructure. As well as providing bank protection, these structures promote vegetated stream banks, create resting pools and scour holes for fish habitat, and increase bio-diversity for aquatic species. Despite these benefits, stream barbs are not a common means of stream bank protection in Canada, due largely to a lack of suitable design guidelines. The overall goal of stream habitat restoration in incising channel systems should be to accelerate natural processes of channel equilibrium recovery, riparian re-vegetation, and stream-floodplain interaction. Incorporating stream barbs, instead of traditional bank protection measures, attempts to achieve these goals. A three-dimensional numerical model: 'Simulation in Intakes with Multiblock option' (SSIIM), was used to model the effects of placing a series of stream barbs along an unstable section of Sawmill Creek. The average bankfull depth, width, and discharge of the creek are 1.2 m, 7.5 m, and 9 m3/s respectively. The model was used to assess various design alternatives for a series of seven stream barbs at two consecutive channel bends requiring stabilization measures along their outer banks. Design criteria were principally based on the reduction of velocity, shear stress and subsequent erosion at the outside bank of each bend, and on the relocation of a new thalweg towards the centre of the channel, away from the outside bank. Sawmill Creek has the added complexity of having predominately clay bed and banks. The erosional behaviour of cohesive sediments such as clay is difficult to model correctly, due to the complex site-specific physio- chemical properties of clay particles. Following the construction of the proposed barbs at our field test site this summer (2009), and data collection the following spring and summer, we hope to advance the current knowledge of cohesive sediment transport processes in a complicated three-dimensional turbulent flow field. For the present modelling effort, erodibility of the consolidated clay bed and bank material was estimated based on establishing an entrainment threshold at near-bankfull conditions. The focus of this research is on (i) the unique site conditions and environmental protection requirements, (ii) design methodology, and (iii) results of the numerical simulation. The three-dimensional numerical model was capable of reproducing the expected distribution of secondary flow in a channel bend, the unique three- dimensional flow field resulting from a series of submerged structures and the associated patterns of soil erosion and deposition. The numerical modelling also demonstrated to be a useful tool for optimizing barb design for stream bank protection at the proposed field test site. Modelling results confirmed that in the vicinity of the barbs, the addition of the proposed barb layout achieved substantial reduction in erosion (up to 98 %), bed shear stress (up to 59 %) and streamwise velocity (up to 51 %).
Determining the Effectiveness of Riparian Revegetation Projects in the Maroochy Catchment, Queensland, Australia
Restoration projects are often performed without a post-project assessment, which leads to uncertainty if they are producing the desired results. Local government and community organizations in the Maroochy Catchment, in Queensland, Australia, organized a series of stream revegetation projects along streams flowing through areas with little or no riparian vegetation. The purpose of the project was to prevent high sediment fluxes and nutrient fluxes in the catchment, which were adversely affecting the Maroochy Estuary. Between October and November 2008, field data on selected physical, chemical and biological indicators were collected from three types of stream riparian zones in the catchment: mature forest, revegetated (3-8 years), and pastoral. Indicators included nutrient concentrations of riparian soils and stream water, water quality parameters (e.g. dissolved oxygen, conductivity, etc.), canopy percentages and sediment chlorophyll-a. Data were analyzed statistically to determine post-project success of the revegetated sites by comparing their key indicators to that of mature forest and pastoral sites. The purpose of this poster is to present the effectiveness of riparian zone revegetation projects in the Maroochy Catchment.
Asymmetrical Changes in Hydraulic Gradient Along Valley and River Transects During Meander Cutoff Evolution
This work is motivated by the uncertainty about the presence and rate of steepening in river and valley hydraulic gradients bounding meander bends during meander cutoff, and how these gradients may affect river restoration designs. Our science question is whether the river and valley hydraulic gradients are equal and unchanging during meander cutoff. Laboratory investigations are underway prior to field experiments and model simulations. The laboratory physical model is a 2.1m x 0.9m EmRiver Process Simulator, running approximately 20 minutes per simulation, using particulate matter with a specific gravity of 1.6, a flow rate of 45ml/s, an initial river cross-section averaging 24cm2, and a range of radius of curvature from 15cm to 18cm. We calculated gradient using river water surface and valley watertable measurements, and their separation along a fixed orientation parallel to the valley slope, not along the thalweg. Measurements were taken with auto level, rod, and tape, and orthoimagery was captured to refine estimates of channel geometry. The ERDAS Leica Photogrammetry Suite processes digital images to generate digital elevation models (DEM) of the system. Initial results have confirmed a steepening of the river hydraulic gradient, from 4% to 5.5% for initial radius of curvature of 15cm, and from 7.7% to 10.9% for curvature of 18cm. The valley watertable gradient has a slight reduction or flattening of about 0.4%. Changes in channel geometry during cutoff include adjustments to cross-sectional area and increasing meander wavelength and sinuosity.
Model Simulation of In-Channel Restoration Structures and the Disruption of Cross-Stream Circulation and Streamwise Velocity Patterns in Meander Bends
Channel restoration in developed areas often attempts to slow or stop lateral migration at meander bends by using in-channel restoration structures, such as vanes and j-hooks. Our modeling experiment examines how in-channel structures affect cross-stream circulation and streamwise velocity patterns, with the results guiding field monitoring campaigns. Simulations were conducted with the FLOW 3D® computational fluid dynamics (CFD) package. Before modeling in-channel structures, the CFD model framework was verified using laboratory flume data generously provided by Ecole Polytechnique Federal in Lausanne, Switzerland. The flume was 1.3 m wide, had a 0.15 m boundary flow height, a straight inflow section sloped at 0.0022, a 193° meander bend, and a 1.7 m radius of curvature, and was used a rectangular and a sloping bed. The subsequent simulation included one or more in-channel structures in the meander bend, and changes in the cross-stream circulation strength and streamwise velocity patterns at meanders were recorded. The in- channel structures were able to reverse and stop cross-stream circulation along the in-channel structure, and disrupt the streamwise velocity core. The magnitude of these disruptions varied with changes in radius of curvature and inflow depth. Simulation results suggest radius of curvature and in-channel restoration structures could be jointly modified to maintain cross-stream circulation when needed for key point bar or cut- bank processes.
A Conceptual Model for Stream Restoration in Western New York: A Developing Case Study in the Elton Creek Watershed
The practice of ecosystem restoration is still often viewed as an "art" versus a science. This view is further fueled by the manner in which many restoration projects are identified and implemented- on a case by case basis with little documentation on the drivers of the ecosystems in question or the establishment of success criteria once the project is complete. While monitoring data may exist to quantify the existing conditions of a stream (biological, chemical, and physical), these data are not always easily translated into design criteria for restoration. Rather, most restoration designs rely upon best professional judgment by highly experienced practitioners. Various agencies in Western New York have completed numerous stream restoration projects in the region to improve in-stream habitat and address impaired streams. In order to capture these successful restoration practices in a manner that can be translated to other streams in the region and inform future restoration designs, a conceptual framework is proposed for the Western New York region. The conceptual framework will identify the ecosystem drivers in the stream riparian corridors and discuss trajectories of these ecosystems. A project site on Elton Creek near Delevan, New York has been selected as a test case for applying this conceptual framework. The focus of this paper will be on determining the drivers of the stream ecosystem through the use of rapid geomorphic assessments. The results of two rapid geomorphic assessment methods be presented and contrasted in the context of the conceptual framework.
Crustal Structure of the Hudson Bay Region: Insights from Receiver Function Analysis
Cratons are vast areas of continental lithosphere that have remained relatively intact and stable since the Precambrian. They tend to be underlain by deep roots, characterized by anomalously fast seismic velocities (∼3%), which tomographic images show can extend to depths of ∼350km. The process by which these roots are formed is presently not well understood, but is believed to be associated with the accretion and imbrication of subducted oceanic material during the Archean. In order to test hypotheses on Precambrian crustal and lithospheric evolution in the North American craton and its influence on the intracratonic Hudson Bay basin, we undertake a study of P-to-S receiver functions to obtain estimates of crustal thickness and Vp/Vs ratio. Our high quality data come from the POLARIS, CHASME and CNSN networks, and the recently deployed HuBLE-UK broadband network of ten stations presently recording in northern Hudson Bay. Receiver functions from stations on Archean basement exhibit a sharp Moho Ps phase and clear subsequent reverberations, in contrast to those in the Trans-Hudson, which show much more complexity. Our results indicate Moho depths in Archean domains are consistent with other cratons worldwide (35-40km), with significant crustal thickening occurring within the Proterozoic Trans-Hudson Orogen (up to 52km). Increased crustal thickness (>40km) is also seen on Baffin Island, another region dominated by Proterozoic geology. Variations in Vp/Vs ratio appear to characterise certain geological regions, with elevated values (>1.74) in the Hearne craton, northern Hudson Bay and Southern Baffin Island regions, possibly due to reworking during Hudsonian collision. The lowest Vp/Vs ratios (∼1.70) are found within the Rae Craton, indicative of a felsic to intermediate lower crust. Combining results from the H-κ analysis with the nature of the crust-mantle transition zone will provide insights into the composition of the lower crust, and also shed light onto crustal evolution during the Precambrian.
Structure of the Crust and Uppermost Mantle Beneath Hudson Bay Based on Ambient- Noise Tomography
The Hudson Bay basin is the least studied of the four major Phanerozoic intracratonic basins in North America, which include the hydrocarbon-rich Williston, Illinois and Michigan basins. The Hudson Bay basin is more than 1000 km wide and contains up to 2 km of Palaeozoic sedimentary rocks, yet the origin of the basin is still unknown. This study focuses on determining how the Hudson Bay basin formed and on regional crustal structure based on ambient-noise tomography. Twenty-one months of continuous ambient-noise recordings have been acquired from 43 broadband seismograph stations that encircle Hudson Bay. These stations are part of the Hudson Bay Lithospheric Experiment (HuBLE), an international project that is currently operating more than 40 broadband seismograph stations around the periphery of Hudson Bay. Of the 43 stations, 10 stations, located in northern Hudson Bay, belong to the NERC array and 2 stations, located in northern Manitoba, belong to the University of Manitoba. Following established processing procedures that include trace normalization and spectral whitening, cross-correlations are computed for all possible station pairs. The resulting waveforms are treated as Green functions, from which group velocity dispersion measurements can be made. Since Hudson Bay freezes during winter months, there is a pronounced asymmetry to the Green functions indicative of noise sources along the Atlantic seaboard. Preliminary results indicate shield-like conditions in most areas, but reduced crustal velocities beneath the Hudson Bay basin.
Investigating Anisotropy in the Upper Mantle Using Shear Wave Splitting From the Manitoba Teleseismic Array
Shear wave splitting parameters have been determined for a large portion of the Archean Superior Province in Ontario. An anomalous zone was previously discovered in the western Superior (WS) west of 86 deg. W, where the split time is significantly greater than in the eastern Superior. The high split times in the WS were believed to be the result of the alignment of anisotropic fabrics in the lithosphere and asthenosphere or a strong fossil anisotropy preserved in the sub continental lithosphere. The stations of the Manitoba teleseismic array (MTA) are distributed throughout the WS covering its western edge beneath the Phanerozoic cover in southern Manitoba, and the northwestern boundary of the Superior with the Trans Hudson Orogeny in northern Manitoba. The southern stations of the MTA show a distinct east to west decrease in delay time from 1.38-0.38 s while the fast axis direction remains constant around 66 degrees. A much different response was viewed in the north where the split time remained constant around 1 s but the fast axis direction varied significantly from 35-79 degrees. Anisotropy in the north is interpreted to be the result of vertical coherent deformation of the lithosphere because of the similarities in varying fast axis and geological structures. The southern stations display a decrease in measured anisotropy, which could be the result of a change in the structure or thickness of the lithosphere or possibly a deviation from the alignment of asthenospheric and lithospheric anisotropic fabrics. Our findings were able to confine the anomalous zone of high split time in the WS from 86-98 deg. W.
Hydrology of two Hillslope Streams, Polar Bear Pass, Nunavut, Canada
Polar Bear Pass (PBP) is a large low-gradient wetland which spans east-west across the central part of Bathurst Island, in the Canadian High Arctic (75o44'N, 98o25'W). It is about 20 km long and about 4 km wide, dotted by ponds, two large lakes and is bordered by low-lying hills with an elevation of about 90 m. Valleys and incised streams ranging from single order to higher dissect the hills and act to channel meltwater, rainwater and nutrients into the low-lying wetland. While the biology of PBP is well known, its hydrology and future sustainability in response to climate warming is not clear. In 2007, we investigated the hydrology of channel snow in a single-order stream near base camp and in 2008; we expanded our study to include a 2nd order hillslope stream. In this study, we investigate the seasonal hydrology of two hillslope streams (small-0.2 km2, large-4.2 km2) from the end of winter to freeze-back. Eventually, we want to assess the importance of these lateral water sources in the resilience of this High Arctic wetland. Snowcover was thin in 2008 but summer precipitation was higher than normal for this polar desert environment (> 90 mm). Air temperatures were lower than in 2007 but above average for the area. The two hillslope watersheds depict nival regimes where snowmelt is a major component of the streamflow pattern, with diurnal pulses driven by meteorologic conditions (e.g. radiation receipt, Ta > 0oC). In the post- snowmelt period, streamflow was maintained in the large basin but not in the smaller one. The large basin had zones within its tributaries with thick organic material, ground ice and thin frost tables (ca. 0.40 m), which likely limited infiltration and water storage, prolonging streamflow. The small basin was largely devoid of vegetation and had a deep frost table (ca. 0.7 m). Streamflow did not occur until basin storage (e.g. soil moisture) was satisfied. Consequently, seasonal water budgets and runoff ratios for the two hillslope watersheds were different. This study suggests that in the post-snowmelt period, basin characteristics (soils, vegetation, and ground ice) and their location in a hillslope stream network can play an important role in prolonging either flow, or limiting streamflow through absorption and storage of water.
Parameterization of Organic-covered Permafrost Soils in Land Surface and Hydrological Models
Close to one-third of the earth's surface is underlain with permafrost and much of the permafrost terrain is covered with a surface organic layer of various depths. The need to improve mathematical representation and parameterization of cold region processes in land surface and hydrological models have been well recognized in recent decades. However, progress has been hindered by (a) the complexity and variability of the soil system associated with thawing/freezing processes and organic cover and (b) the shortage of high quality field data due to the technical and logistic difficulties imposed by the harsh environments. Large variations exist in the parameterizations of thermal and hydrological processes in current land surface and hydrological models. Many of them were developed and validated in soil and climate conditions different from those in permafrost regions. In this study, efforts have been made to examine the most important thermal and hydraulic parameterizations and their effects on the simulations of ground thawing/freezing and infiltration/runoff processes against detailed measurements obtained at six field sites in Canada's discontinuous permafrost region. The tested parameterizations include (a) three methods for thermal conductivity, (b) three methods for hydraulic conductivity and soil water retention, (c) three methods for unfrozen water content, (d) six algorithms for thawing/freezing simulation and (e) five algorithms for infiltration simulation. The field sites cover various vegetation types including boreal forest, alpine tundra and wetland peat plateau. The soil organic depth ranges from 0.0 to 3 m. Field data used include daily meteorological variables (solar radiation, air temperature, relative humidity, precipitation and wind speed), daily liquid soil water content and soil temperature at various depths, and daily snow depth. Total soil water content (frozen and liquid) was monitored using twin probe gamma attenuation at three sites. Ground thawing and freezing depths, snow melt, infiltration and runoff were estimated from the above measurements. Methods of different parameterization were compared using identical inputs and evaluated against the same set of observed outputs. The main conclusions of this study include: (a) de Vries' method is recommended to parameterize the thermal conductivity in permafrost soils; (b) the van Genuchten method provides the best fit to the observed soil hydraulic curves, but has the most parameters. while the Brooks and Corey method and Clapp and Hornberger method achieve similar results for most soil moisture conditions as long as appropriate parameters were chosen; (c) the segmented linear function is the simplest to be parameterized with limited available data, while the water potential-freezing point depression equation is recommend for coupled thermal and moisture simulations; (d) a numerical model with an apparent heat capacity treatment gives the most accurate simulation of ground thawing/freezing depths in all the tested sites; and (e) no current infiltration scheme is valid for all the infiltration situations in organic-covered permafrost soils as different methods must be employed in different infiltration stages.
Isotope Hydrology of Arctic Tundra Lakes in a Region Impacted by Permafrost Disturbance
A projected "hot spot" of climate warming and development is the Mackenzie River Delta region, Northwest Territories, Canada. The upland tundra areas within the Mackenzie Gas Project development area north of Inuvik contain thousands of small lakes and ponds with poorly defined ephemeral drainage that are underlain by thick permafrost and ice-rich sediments for which the basic water balance controls are not fully understood. Natural retrogressive thaw slumps are common along lakeshores and the rapid drainage of ice-rich permafrost-dammed lakes has been occurring. Ongoing oil/gas exploration activities and infrastructure construction may result in terrain disturbance and localized degradation of permafrost, while climate change may increase the magnitude and frequency of thermokarst processes. These disturbed lakes are believed to act as historical analogues for the future effects of climate change on the hydrology, geochemistry, and aquatic ecology of small tundra lake catchments in the continuous permafrost zone of northwestern Canada. Environment Canada initiated an integrated research program in 2005 with the overall goal of improving our understanding of hydro-ecological processes in freshwater aquatic ecosystems affected by shoreline slumping vs. pristine lakes. Limited catchment studies have examined water-balance parameters (e.g., precipitation, evaporation, and surface flows) for tundra lakes in the development area. Enrichment of oxygen-18 (18O) and deuterium (2H) stable isotopes in surface waters have been shown to be useful indicators of water balance variations in remote permafrost regions of Canada where hydroclimatic information is very limited. In particular, information on evaporation: inflow (E/I) ratios and residence times would provide useful information for estimating appropriate water withdrawals from lakes within the proposed development area. A key question is "does permafrost slumping impact the hydrology of tundra lakes via catchment area enlargement and/or enhanced snow accumulation?" The objective of this paper is to present preliminary isotope hydrology findings from i) seasonal surveys in a pair of representative lake catchments and ii) annual synoptic surveys in >60 lakes (shoreline slumping vs. pristine lake catchments) located along the proposed Mackenzie Valley Gas Pipeline.
Development of an Energy-based Runoff Generation Model in Arctic Tundra Regions
Recently, the demand for improved predictive models of runoff in arctic tundra regions has increased, given uncertainties regarding the future availability of northern freshwater resources related to climate change and expanding resource development. Hydrological characteristics of the arctic tundra are substantially different from those of temperate regions in which traditional runoff generation theories were based. Therefore, theories that apply specifically to the arctic tundra need to be incorporated into a robust runoff model to improve predictions for this region. In addition, the topographically based contributing areas described by source area concepts in the model can be combined with or superseded by an energy-based contributing area. The goal of this project is to design an energy-based runoff generation model to simulate characteristics influencing the aerodynamic and radiation regimes at the surface, including surface roughness, slope aspect and angle in tundra landscapes. Through the model, the frost table topography can be derived, which is a critical step toward predicting the rate and direction of flow. As most arctic tundra catchments in Canada are ungauged, understanding the impact of observed climate warming and unprecedented resource extraction activities can only be achieved through improved conceptualization of hydrological processes. Thus, an energy-based runoff generation model may lead to improved predictions of streamflow in both the present and future.
Investigating Short-term Sustainability of Shallow Water Features Within the Canadian Sub- Arctic; A Churchill, Manitoba Case Study (1947-2008)
Within the sub-arctic environment, shallow ponds occupy between 15-50% of the landscape and strongly contribute to regional hydrological and energy budgets. Recent remote sensing studies have mainly focused on the sustainability of small water features within Alaska and Siberia; while little is known regarding the Canadian sub-arctic. The overall objectives of this study were to (a) quantify the current distribution and size of ponds within the Churchill region, (b) calculate surface area changes between 1947 and 2008 from a selected number of ponds and to (c) determine the influence of inter- and intra-seasonal hydrological variability on the quality of remote sensing change detection studies within the region. This study focused on detecting surface area distribution and change within 120 surveyed ponds over a 100kmē study area using SPOT, Landsat and air photo images (7 images total) between 1947 and 2008. Results from the imagery analyses were linked to a regional climate analysis and simple water balance model in order to determine if climate and hydrological changes could have impacted the quality of change detection results over the 61 year period. Although inter-annual variability has been observed, no significant long-term changes were revealed with regards to precipitation over the past 61 years. In 2008, ponds exhibited a mean water depth of 29cm and a mean surface area of 11,000mē. An examination of pond bathymetry, field data (subset of 30 ponds) and seasonal/annual hydrological fluctuations (based on simple water balance modeling) indicate the significant variance of seasonal pond hydrologic storage; which can impact the quality of short-term pond sustainability detection. This work suggests that remote sensing studies should be paired with climatic analysis when examining the sustainability of surface water features.
Paleoclimate Reconstruction at Lamanai, Belize Using Oxygen-Isotope Tropical Dendrochronology
Tropical dendrochronology can be complicated because many trees growing in these areas lack distinct visible annual rings. However, the oxygen-isotope composition of wood growing in tropical regions can provide a record of seasonal fluctuations in the amount of precipitation even when visible rings are absent. Variations in the oxygen-isotope compositions of cellulose as the trees grow can be related to the relative timing of wet and dry seasons and used to identify periods of drought. In this study, the oxygen-isotope composition was determined for cellulose extracted from living trees at the site of Lamanai, Belize to assess the variation in oxygen-isotope values that result from heterogeneity within individual tree rings and seasonal fluctuations in amount of precipitation. In temperate regions, the latewood rings that form during periods of reduced growth are traditionally selected for oxygen-isotope analysis of cellulose because their oxygen-isotope compositions are more directly influenced by climate and precipitation during the growing season. However, in tropical isotope dendrochronology, when visible rings are present, detailed sampling of both the light coloured earlywood and the denser latewood is required. At Lamanai, a seasonal signal was evident in the oxygen- isotope composition of the cellulose when tree rings were sectioned in very small increments (approximately every mm), sub-sampling both earlywood and latewood. However, the visible rings did not always correspond with minimum or maximum oxygen-isotope values. As a result, the amplitude of the oxygen-isotope signal obtained by considering only latewood samples is smaller than that obtained from fine-increment sampling. Hence, the oxygen-isotope values of latewood samples alone did not provide accurate data for climate reconstruction. Multiple series of latewood samples extracted from different cross-sections of the same tree did not consistently show the same trends in oxygen isotope values, which can differ by up to 2 permil around the circumference of the same ring. This indicates that even when visible rings are present in tropical trees, the rings may not be annual or continuous. However, the amplitude of variation in the oxygen-isotope values of cellulose from both early and latewood can be related to seasonal signals across the modern tree rings. These signals will be compared to the oxygen-isotope composition of tree ring cellulose extracted from a wood sample excavated from an ancient tomb at the site of Lamanai to assess the preservation of the cellulose- isotope signal in this artefact. If similar oxygen-isotope patterns are preserved in ancient cellulose they can be used as a proxy to determine past climate conditions, such as those experienced by the ancient Maya populations in Belize.
Oxygen Isotope of Phytoliths in Modern Wetland Plants and the Application to Paleoclimate Reconstruction
Because the oxygen-isotope composition of phytoliths in modern plants is controlled primarily by the oxygen-
isotope composition of source water, temperature and relative humidity, isotope analyses of ancient phytoliths
extracted from soils have the potential to reveal paleoclimate information. A controlled-temperature, growth-
chamber experiment was conducted to determine the relationships among temperature, relative humidity, soil
water evaporation, plant-water isotope composition and oxygen-isotope composition of phytoliths in cattails
and horsetails. Typha, a cattail species that grows in wetland conditions, and Equisetum, a horsetail
species that prefers dry soils, were each grown in four separate chambers at 15, 20, 25 and 30 degrees
Celsius. The oxygen- and hydrogen-isotope compositions of watering water, soil water, vapour in the growth
chambers and plant water from the leaves and stems were analyzed throughout the eight-month long artificial
growing season. The oxygen-isotope compositions of phytoliths extracted from the transpiring tissues of the
plants at the end of the growing season were also analyzed.
The results show that the oxygen-isotope composition of phytoliths is strongly correlated with oxygen isotope composition of average plant water from late growing season, rather than plant water extracted earlier in the season or from the stem, leaf-base or leaf- apex alone. As the temperature increases, the oxygen-isotope fractionation between phytoliths and plant water decreases. This trend is comparable to the oxygen-isotope thermometer equation developed by Shahack- Gross et al. (1996: Geochim. Cosmochim. Acta 60, 3949-3953), but the separation in oxygen-isotope values between silica and water at a given temperature is about 4 ‰ lower. This discrepancy is likely from uncertainties in the oxygen-isotope value of leaf water at the site of phytolith precipitation, which varied over the growing season as a result of fluctuations in relative humidity and the oxygen-isotope composition of atmospheric water vapour in the chamber. Measured plant water oxygen-isotope values displayed variations of up to 37 ‰ along the length of a single Equisetum plant, and average plant water oxygen-isotope values for this species varied up to 10 ‰ over the growing season. Because the transpiring tissues of most plants produce the largest amount of phytoliths, which are ultimately preserved in the underlying soil, it is imperative to understand the how the variability of the oxygen-18 enrichment in the water of transpiring tissues is reflected in the oxygen-isotope composition of the phytoliths in order to use this method for successful paleoclimate reconstruction.
High-Resolution Gridded Hydroclimatic Reconstruction Over the Canadian Prairies and the Connection to Large-Scale Circulation
A network of precipitation sensitive tree-ring chronologies extending throughout the western Prairie region and the Northwest Territories highly correlates to the climate moisture index of precipitation-potential evapotranspiration (P-PET), thus, capturing the long-term hydroclimatic variability of the region. We present a 500 year seasonal hydroclimatic reconstruction for the prairie region of western Canada over a 0.5o X 0.5 o gridded historical climate model from 1901 to 2000 (McKenney et al. 2006). Transfer functions were developed using forward selection regression techniques to reconstruct the large-scale moisture index. Climate variation in North America has been linked (teleconnection) to sea surface temperature anomalies in the tropical and extra-tropical regions of the Pacific Ocean, ENSO and PDO, respectively, and the Atlantic Ocean, AMO. The influence of ENSO on inter-annual variability is strong but is also modulated by the quasi-periodic decadal and multi-decadal phases of PDO and, to a lesser extent, AMO. Identifying these oscillatory modes in both observed and reconstructed climate records and teleconnection indices provides a better physical explanation of the hydroclimate variability of this region, particularly the climatic conditions associated with in- phase or anti-phase modes. Understanding and explaining natural variability and the effect of global warming on the natural climate cycle, ultimately depends on relating large-scale circulation to wet and dry cycles, the drivers of precipitation and drought. A series of circulation indices associated with extreme hydroclimatic events have been developed and reconstructed to identify different phases of the dominant climate modes. These reconstructions ultimately provide the baseline for putting future climate scenarios of large-scale circulation and drought scenarios into historical context.
Runoff Regime Prediction: a Water Balance Approach for High Elevation Basins
Runoff regime, intended as the curve of the mean monthly runoff values, is a valuable information for the control of water management strategies. The increasing competition between different water uses and the recognized need for a sustainable resources exploitation require an improvement in the runoff prediction procedures also for regions, as the European Alps, historically characterized by abundant water resources. Moreover, threats of global temperature increase include specific modifications of the regime, that can affect water uses based on runoff regularity. For mountain catchments, however, runoff prediction is made complex by the topographic heterogeneities, the dynamics of snow accumulation and melt, and the high spatial precipitation variability, generally poorly described by the measurements networks. In this paper we present a parsimonious (i.e. characterized by a limited number of parameters) water balance model conceived for mountain basins that uses widely available input data (mean monthly temperature and precipitation). The model works on a monthly scale but we introduce a statistical representation of the within- month temperature variability to take into account the effects of the above- and below-zero frequency on the snow accumulation. In fact, the modeling scheme contains a specific snow accumulation and melt module that allows to partition precipitation into snow and rainfall and to assess snowmelt volumes based on a temperature threshold mechanism. Runoff is obtained as the sum of two contributes: net precipitation (rainfall minus evapotranspiration) and snowmelt. A parametric procedure to correct precipitation measurements due to snow undercatch is also proposed. To assess the performances of the model we apply it to a sample of catchments located in the Western Italian Alps and we compare the simulated and observed runoff regimes in terms of mean absolute error. The model proves able to describe the timing and shape of the regime for most of the basins, with good performances also for very high elevation watersheds. Discrepancies, however, exist between the observed and reconstructed runoff volume when the observed precipitation regimes are affected by evident errors (e.g., mean annual precipitation smaller than mean annual runoff). In these cases, a correction factor is applied to the precipitation regime to guarantee the closure of the basin water balance. Due to the parsimonious representation of runoff regime formation mechanism, the model allows one to investigate, over a large scale, the role of climatic variables on water availability and runoff timing in mountain basins. Moreover, the modest requirements in terms of input data and parameter calibration make the model suitable for applications in ungauged basins.
Potential of Multiple Dendroclimatic Proxies for the Prairies?
Concern to establish the nature and rate of climatic changes, should serve to reinforce our determination to understand similar details of the 'natural' (i.e. non-anthropogenic) variability of climate. Dendrochronology offers great potential for studying climatic and environmental variability at local and regional levels, because of the wide geographical distribution of suitable sites, high temporal resolution (annual or even seasonal), continuous and relatively long, absolutely-dated, well replicated and environmentally sensitive (i.e. accurate) characteristics of tree rings. Patterns within the annual rings of Quercus (oak) species, suggest that environmental factors influence the size and density of vessels within the ring, either by acting as a limiting factor for growth or through fine tuning of the wood structure to environmental factors. The purpose of this study is to investigate growth responses (annual, early- and late-wood widths) of Q. macrocarpa to regional climatic variability affecting the Canadian Prairies. Results indicate that annual ring widths, as well as early- and late- wood chronologies from southeastern Saskatchewan capture regional signals related to moisture and drought conditions. Correlations suggest that late-wood measurements are more strongly representative of annual ring-widths, than are early-wood widths, and can therefore be applied for investigating seasonal fluctuations in climatic data. Correlations with precipitation and PDSI values indicate that annual, early- and latewood chronologies are useful proxies for investigating large scale climatic fluctuations, and present the opportunity for further investigation of the effects of indices that represent major modes of climate variability, such as the effects of El Niño/Southern Oscillation (ENSO) and Pacific Decadal Oscillation (PDO) patterns that are thought to influence climate within the Prairie region. This study is novel in terms of sub-annual analysis of tree-rings in a region that previously lacked dendrochronological research.
Trends in Groundwater Levels in the Canadian Prairies
Groundwater could be an increasingly important water supply in the Canadian interior with global warming and declining summer runoff; however, not enough is known about the behaviour of groundwater under climatic variability. A network of over 33 groundwater wells is analyzed in order to document trends in groundwater levels. Groundwater wells are spread through the three Prairie Provinces (Alberta, Saskatchewan, and Manitoba) with median monthly groundwater level records spanning up to 40 years. The aquifers are mostly in sand and sandstone which make them highly sensitive to climatic variations. In addition, these wells have not been affected by human activities such as pumping. The Mann-Kendall non parametric test and the trend free pre-whitening (TFPW) approach for correcting the first order autocorrelation are used to detect significant trends on annual and seasonal groundwater levels. Results of the application of the Mann-Kendall test show that 67% of the wells have a trend statistically significant at p < 0.05 and most of these are decreasing. They also suggest that groundwater levels in north central areas show either no or decreasing trend, in contrast, groundwater levels in southern areas are dominated by increasing trend. The spatial distribution of trends in annual groundwater levels show a spatial coherence with trends in evaporation during the warm season documented in other studies, however, no spatial coherence has been detected between trends in annual groundwater levels and annual precipitation.
Results from the Mayson Lake Hydrological Processes Study 2008 Summer Field Season
The Mayson Lake Hydrological Processes Study area is located in the southern interior of British Columbia ∼ 60 km NNW of the City of Kamloops, British Columbia on the Thompson-Bonaparte Plateau (51.2° N, 120.4° W; 1260 m a.m.s.l.). During the summer of 2008 a series of projects were carried out in preparation for a larger, more detailed study of the impact forest disturbance and subsequent re-growth has on hydrological processes. Results from the 2008 field season suggest that canopy interception loss of rainfall in a mixed lodgepole pine (Pinus contorta var. latifolia Dougl.) - hybrid spruce (Picea glauca (Moench) Voss. x engelmanni Perry x Engelm.) - subalpine fir (Abies lasiocarpa (Hook.) Nutt.) stand, where pines were at the grey - attack stage of mountain pine beetle (Dendroctonus ponderosae Scolytidae) infestation, is comparable to healthy mature stands, but significantly greater (α = 0.05) than that from the burned stand. Canopy interception loss, throughfall and stemflow for 14 events totalling 50.1 mm were found to be 41.2, 58.7, and 0.1 % of rainfall, respectively. Near-surface (surface to 20 cm depth) soil moisture depletion was determined using weekly TDR measurements at 32 points in each plot during a two-month dry- down period (June 16 -August 18) in which only 30.8 mm of rain fell. Soil depletion was found to be ∼ 2.6 times greater from juvenile stands than from a clear-cut, while in the beetle infested stands soil depletion averaged ∼ 1.6 times greater than in the clear-cut. Assuming no deep drainage past a depth of 20 cm or lateral throughflow out of the study plots, actual evapotranspiration (AET) was estimated at 53.2 ± 4.0 mm from the clear-cut during the dry-down period, while from two healthy juvenile stands AET was estimated at 87.1 ± 7.0 and 87.8 ± 4.0 mm. In two beetle infested forests AET during the dry-down period was estimated at 63.4 ± 5.0 and 69.8 ± 3.2 mm. The larger AET losses from the juvenile stands compared to the clear-cut is probably a consequence of transpiration from fast growing stocked pines, and, to a lesser extent canopy interception losses, while the greater AET from dead/declining forests compared to the clear cut is probably a result of relatively high interception losses from the forest canopy as well as transpiration from the understory. Stemflow, although negligible in the mature forest, was found to be an important point source of water from juvenile pine stands, especially for trees with basal diameters < 8 cm. These small trees had an average seasonal funneling ratio of 19.6 ± 6.6 (α = 0.05), while larger, put still healthy pines (ranging from 8.2 to 16.3 cm in diameter) had an average funneling ratio of 3.3 ± 2.2 (α = 0.05). The maximum funneling ratio observed during the study period was 79.7 (rainfall = 8.6 mm, tree basal diameter = 4.6 cm).
Coupling Evapotranspiration and Watershed Storage to Assess the Impact of Forest Disturbance on Low Flows
Low flows are important for water-supply planning and design, and maintenance of quantity and quality of water for irrigation, recreation, and fish and wildlife conservation. There have been concerns recently that climate warming and land cover changes due to an unprecedented pine beetle epidemic in British Columbia, Canada, may cause a deterioration of water quantity during low flow periods and at certain times may become a hazard to ecosystem and to water management schemes. A study to characterize the sensitivity of the low flow regimes was performed for several mainly forested catchments located within the Fraser River basin. Here, summer low flows are maintained through the release of water from groundwater and riparian storage, lakes and wetlands, but are reduced by high evapotranspiration rates in the catchments. Since evapotranspiration in British Columbia accounts around 40% of the precipitation, the first part of this work was focused on the assessment of the relationship between the potential evapotranspiration (PET) and the actual evapotranspiration (AET) for undisturbed and disturbed landscapes which is expected to influence the hydrological behavior during the low-flow season. Through its influence on evapotranspiration, forest age appears to play an important role in the water balance. The second part of the study implemented a forest age dependent calculation of AET into a parsimonious water balance model, which was applied to simulate the sensitivity of the flow regimes of 15 non regulated watersheds to changes after the beginning of the pine beetle epidemic at a large scale. The model input was derived from disaggregated gridded 30-year climate normals. Since the geologic and topographic properties are first order controls on water storage and release of the examined catchments a framework for regionalization of these properties into ungauged catchments was developed. Furthermore, the interaction between forest disturbance and evapotranspiration may help to predict the magnitude and timing response of low flows -among others- to environmental changes as well as the temporal scales of biogeochemical cycling.
Quantifying the Hydrologic Impacts of Mountain Pine Beetle and Salvage Harvest in the Fraser River Basin, British Columbia, Canada
The province of British Columbia (B.C.), Canada is currently experiencing the largest mountain pine beetle (MPB) outbreak ever recorded in North America. The most recent surveys indicate that widespread mortality of pine trees has occurred in over 10 million hectares of forest (an area roughly the size of Iceland) and the outbreak continues to kill mature pine in the province. The epicenter of the current outbreak is in the Fraser River drainage basin (230,000 km2), where roughly 8 million hectares of forest has been affected, approximately 35% of the drainage area. Due to the massive area of the current (and projected) infestation and associated salvage harvest operations, the potential exists for widespread and significant local and regional hydrologic impacts within the basin. However, the scale and physiographic, climatic and topographic heterogeneity of the Fraser River basin precludes both direct observation and extrapolation of hydrologic impacts observed from a limited number of stand-level and small-basin experiments. As a result, the Variable Infiltration Capacity (VIC) hydrology model has been used to quantify these hydrologic impacts within the Fraser River basin. The VIC model is a spatially-distributed macro-scale hydrology model that has been applied at a resolution of 1/16-degree (approximately 27-32 km2, depending upon latitude) and used to quantify streamflow impacts for 60 sub-basins ranging in area from 330 to 230,000 km2. The local and regional sensitivity of streamflow to MPB and harvest disturbance has been assessed using a set of seven scenarios. These include a pre-infestation baseline (c. 1995), a current (c. 2007) forest cover and five hypothetical scenarios of increasing disturbance severity (from baseline) ranging progressively from 100% kill of mature pine plus 0%, 25%, 50%, 75%, and 100% harvest (by area) of killed pine. All scenarios are forced with meteorological boundary conditions interpolated from station data collected during 1915 to 2006. Results of the modeling effort are preliminary and currently focus on impacts to the annual maximum peak flow regime. As anticipated peak flow impacts becomes more severe as the severity of disturbance increases. For instance, relative changes (from baseline) in the magnitude of the 20-year peak flow event range over the studied basins from no change to a) 8% for current forest cover, b) 8% for 100% beetle-kill (0% harvesting), c) 46% at 25% harvesting, d) 91% for 50% harvesting, e) 130% for 75% harvesting, and f) 172% for 100% harvesting of dead pine. Work is ongoing to explore the mechanisms controlling basin sensitivity to MPB and harvest disturbance, but current results suggest that basin sensitivity is related to several factors including sub-basin scale, local climate, extent and type of forest cover, and topography. Ongoing work involves exploring impacts to other aspects of the flow regime (i.e. annual and seasonal yield, low flow), further exploring the mechanisms of hydrologic change at both the regional and local scales and developing potential indices and predictors of hydrologic sensitivity.
Drought Analysis for River Basins, Using the Hydrological Model SIMGRO
Drought is a recurring and worldwide phenomenon, with spatial and temporal characteristics that vary significantly from one region to another. Drought has major impacts on society and affects among others the environment and the economy. Impacts are likely to increase with time as societies demands higher services for water and the environment. This will even be more pronounced in the coming decades with the projected climate change, i.e. droughts are becoming more severe in large parts of the world. The prediction of droughts is an essential part of impact assessment for current and future conditions, as part of integrated land and water management. An important question is how changes in meteorological drought will propagate into hydrological droughts in terms of changes in the groundwater system or in the river flow. The objective of our study is to develop and test tools that quantify the space-time development of droughts in a river basin. The spatial aspect of a hydrological drought (spatially-distributed recharge and groundwater heads), in a river basin brings different challenges with respect to describing the characteristics of a drought, such as: onset, duration, severity and extend. We used the regional hydrological model SIMGRO as a basis to generate the necessary data for the drought analysis. SIMGRO is a distributed physically-based model that simulates regional transient saturated groundwater flow, unsaturated flow, actual evapotranspiration, sprinkler irrigation, stream flow, groundwater and surface water levels as a response to rainfall, reference evapotranspiration, and groundwater abstraction. The model is used within the GIS environment Arc-View, which enables the use of digital data, such as soil map, land use, watercourses, as input data for the model. It is also a tool for analysis, because interactively data and results can be presented, as will be shown. Droughts in different hydrological variables (recharge, groundwater heads, river flow) are identified by applying the fixed threshold concept to spatially-distributed simulated time series. The method captures the development of both the duration and the severity for the area in a drought. For the analysis we applied the model to the Taquari river basin (about 106.000 km2), which is situated in the Pantanal region, the upper part of the Paraguay River Basin, Brazil. The question we will address is: how does a hydrological drought develop and what are the spatial characteristics and what are the underlying mechanisms. Examples of the analysis will be shown that aim at a better understanding of the process involved which are essential; to assess the vulnerability of river basins for hydrological droughts.
Temporal variation of the water budget in Central Southwest Asia
Water budget terms for Central Southwest Asia (CSWA) have been computed using National Centre for Environmental Prediction (NCEP) reanalysis data for 60-years from 1948-2007 and European Centre for Medium-Range Weather Forecasts (ERA-40) reanalysis for the 44-year period from 1958-2001. First, observational rainfall data from the Pakistan Meteorological Department (PMD) and the Iran Meteorological Organization (IMO) were compared with Global Prediction Climate Centre (GPCC) version-4 data. GPCC rainfall data appear close to the observational data of the region. GPCC rainfall data have then been used for comparison with both NCEP and ERA-40 reanalysis data sets and it was found that NCEP is closer to GPCC than ERA-40 for this region. The domain under study is located from 45-75E and 25-40N including parts of Iran, Afghanistan, Pakistan, Iraq and Kazakhstan. For realistic results only land areas has been used in the study. In the ERA-40 reanalysis data, evapotranspiration may have been overestimated in the system noting that evaporation exceeds rainfall. NCEP computes more precipitation than evaporation with moisture flux into the area providing the balance. This suggests that the ERA-40 data set is not suitable for computing water budget terms in this particular region. Moisture flux convergence is computed by using four daily readings (00, 06, 12 and 18z) up to 300 hPa. Finally, Principle Component Analysis (PCA) techniques are used to divide the domain into six areas and we discuss the temporal variation of rainfall for each area. The main contribution of rainfall in most of the areas is due to western disturbance weather systems moving from west to east. Also the area at the extreme southeast is getting rain though southwest monsoon in the month of July August. It is observed that rainfall decreases significantly after 1980 as a result of a dedcrease in moisture flux convergence.
Hydroclimatic controls on late-summer low flow in British Columbia, Canada
For catchments in the Pacific Northwest of North America the summer period is typically dominated by low flows associated with the relatively warm, dry summers in that region. In recent years streamflow during this period has been critical in terms of water use and fisheries, and there is increasing concern regarding how future climate change with potentially more severe summer droughts may affect late-summer flows. In addition, it is feared that earlier snowmelt timing in spring causes longer recession periods and consequently lower streamflow during summer. This study examined the sensitivity of late summer flows of 153 unregulated rivers over the period 1976-2003 to hydroclimatic influences by fitting regression models using simultaneous and lagged variables of precipitation and timing and magnitude of the snowmelt. To assess potential additional influences of long-term storage changes the residuals were tested for trends and serial correlation. A decrease in September flows across most of the region is broadly consistent with a decline in September precipitation. The most important control on August flows for all streamflow regimes is August precipitation. Lagged variables including July precipitation and the previous winter's precipitation are also positively but more weakly related to August streamflow. Rain-dominated and hybrid catchments tended to have positive trends in their residuals in August, suggesting an increasing trend in groundwater storage. Snow-melt dominated catchments showed no tendency to trends but the runs test detected a substantial number of stations with non-random residuals. This regional study greatly improves the understanding of the hydroclimatic influences on extreme summer low flow in the region. Results suggest that (the less predictable) summer climate is a larger influence than previously assumed. They also help to identify where seasonal snow processes or long- term groundwater storage should be considered in model development, forecasting, and prediction of climate change impacts.
Impact of drought on surface albedo in Canadian Prairie observed from Terra- MODIS
A new technology was developed at the Canada Centre for Remote Sensing (CCRS) for generating Canada wide clear-sky surface albedo data based on observations from MODIS sensor onboard TERRA satellite. The data include all seven MODIS land bands (B1-B7) mapped at 250m spatial resolution and 10-day temporal interval from year 2000 through 2008. The new product presents an important spatial enhancement as well as an improved retrieval of water fraction and snow characteristics relative to the standard MODIS archival products. The regional data for the entire Canadian Prairie region are extracted and aggregated for different ecozones, such as north to south, the boreal transition, aspen parkland, moist mixed grassland, and mixed grassland etc. The preliminary results indicate that in comparison to normal summer conditions (2006-2008), the albedo for the drought years (2000-2003) summer increases up to 20 percent in the visible band (B1) and decreases as low as 10 percent in the near infrared band (B2). In the shortwave infrared band (B6) where a large absorption by leaf water occurs, the albedo increases as much as 15 percent for the drought years due to less leaf water content. The derived Normalized Difference Vegetation Index (NDVI), which represents a density of healthy vegetation, drops dramatically (up to 30 percent) for the drought period of 2000-2003. Among the different ecozones, the grassland shows the largest response to droughts while the boreal zone shows the least. Further applications of this product include mapping of snow cover (fraction and grain size), the fraction of absorbed photo-synthetically active radiation (fAPAR), ecosystem productivity, water and energy budget, as well as impact of various disturbances, such as wildfires, and long term climate induced trends. This work was conducted at the Canada Centre for Remote Sensing (CCRS), Earth Sciences Sector of the Department of Natural Resources Canada as part of the Project J35 of the Program on "Enhancing Resilience in a Changing Climate". This work was also supported by the Canadian Space Agency under the Government Related Initiative Program (GRIP) and the Canadian IPY program. The MODIS data files were acquired from the NASA Distributed Data Archive Center (DAAC).
Trends of Drought in the Canadian Prairies
The Canadian Prairies is a drought-prone region and severe drought events in this major agricultural region were among the most costly disasters for Canada. Results from several recent global studies suggest that the Canadian Prairies is among the regions which have exhibited the strongest rising trends in drought over the second half of last century. The purposes of the present study are to further examine and to better understand these results for Prairie droughts. In particular, different drought indices, including the Palmer Drought Severity Index (PDSI) and Standardized Precipitation Index (SPI) derived from thoroughly corrected and homogenized station data and soil moisture simulated by forcing the VIC hydrological model with station data are used to assess the trends. The results will be compared to those from the global studies. In addition, we will also attempt to attribute the causes of detected drought trends in the region by using optimal detection techniques and global climate model simulation results as well as water and energy budget analysis.
Hydrological Modelling of a Canadian Prairie Wetland Basin
The eastern Canadian Prairies are a region of cropland, pasture, woodland and wetlands. The region is characterized with many poor and internal drainage systems and large surface water storage terms, so watersheds here have proven challenging to hydrological models that assume good drainage to a stream. The cold climate means that snow redistribution, snowmelt and infiltration to frozen soils are important in regulating runoff generation. The Cold Regions Hydrological Modelling platform (CRHM) is an assembly system to create physically based, flexible, object oriented models. It was used to develop a prairie hydrological model to simulate the hydrological cycling in prairie watersheds and eventually the impact of land use and wetland change on hydrology. Smith Creek Basin (∼445 km2) was divided into five sub- basins, and a modelling feature - 'Groups' or representative basins was applied to model these sub-basins. Within each sub-basin, seven hydrological response units (HRUs): fallow, stubble, grassland, river channel, open water, woodland, and wetland were derived from supervised classification of SPOT 5 imagery. Physically based modules were sequentially assembled in CRHM and applied to all HRUs to simulate hydrological processes, including redistribution of snow by wind, snowmelt, infiltration, evaporation, soil moisture balance, wetland storage and runoff routing. Almost all parameters were set from values determined by remote sensing or field observations; however calibration was used to determine upland depressional storage capacity which could not be measured. Model performance in simulating snow accumulation during the winter of 2007-08 and the subsequent spring freshet was evaluated. Results show the model had generally good performance in estimating snow accumulation, with Root Mean Square Difference (RMSD) ranging from 1.8 mm to 7.9 mm for fallow, stubble, open water and woodland HRUs. Grassland, river channel and wetland HRUs had moderately large values of RMSD, ranging from 6.4 mm to 18.1 mm. The model estimated springtime basin streamflow fairly well; RMSD and Model Bias (MB) were 0.16 m3/s and -0.22, respectively. The model was also tested on an extremely wet year, 1995, and a drought year 2002. In the wet year, wetland storage was nearly full at the time of snowmelt; in the drought year wetland storage was severely depleted. The model showed satisfactory performance in both years though parameter uncertainty was affected by the lack of intense field observations.
Linking Hydrology and Water Resources Management in the IP3 Network
IP3, or Improved Processes and Parameterisation for Prediction in Cold Regions (http://www.usask.ca/ip3), is a research network devoted to studying hydrological science in cold regions. IP3 is funded by the Canadian Foundation for Climate and Atmospheric Sciences for 2006-2010. The network has approximately 80 scientists and students who conduct field work in 10 highly-instrumented research basins in western and northern Canada and incorporate improved understanding of cold regions hydrological processes into a suite of predictive models. This presentation will highlight several efforts and partnerships that have been developed over the last 2 years between IP3 hydrologists and water resources managers, primarily development and dissemination of IP3's Cold Regions Hydrological Model (CRHM) and Environment Canada's Modélisation Environnementale Communautaire - Surface and Hydrology (MESH) modelling framework. To effect direct communication between hydrological scientists and water resources managers, IP3 has hosted or co-sponsored numerous workshops and other meetings that have brought these two groups together for the purpose of identifying current and future needs of the water resources management community. Expressed needs to be addressed by IP3 include improving the models to predict variables such as hydrological outcomes of landscape change, improving information and data management to enhance availability and usefulness of collected data, and improving outreach to raise awareness in the water resources community of the most recent science and tools available for application. IP3 will continue to pursue links with water resources managers to optimally coordinate the network's scientific outcomes with the needs of water managers now and in the future.
Hydrologic Model Development and Calibration: Contrasting a Single- and Multi-Objective Approach for Comparing Model Performance
Hydrologic model calibration aims to find a set of parameters that adequately simulates observations of watershed behavior, such as streamflow, or a state variable, such as snow water equivalent (SWE). There are different metrics for evaluating calibration effectiveness that involve quantifying prediction errors, such as the Nash-Sutcliffe (NS) coefficient and bias evaluated for the entire calibration period, on a seasonal basis, for low flows, or for high flows. Many of these metrics are conflicting such that the set of parameters that maximizes the high flow NS differs from the set of parameters that maximizes the low flow NS. Conflicting objectives are very likely when different calibration objectives are based on different fluxes and/or state variables (e.g., NS based on streamflow versus SWE). One of the most popular ways to balance different metrics is to aggregate them based on their importance and find the set of parameters that optimizes a weighted sum of the efficiency metrics. Comparing alternative hydrologic models (e.g., assessing model improvement when a process or more detail is added to the model) based on the aggregated objective might be misleading since it represents one point on the tradeoff of desired error metrics. To derive a more comprehensive model comparison, we solved a bi-objective calibration problem to estimate the tradeoff between two error metrics for each model. Although this approach is computationally more expensive than the aggregation approach, it results in a better understanding of the effectiveness of selected models at each level of every error metric and therefore provides a better rationale for judging relative model quality. The two alternative models used in this study are two MESH hydrologic models (version 1.2) of the Wolf Creek Research basin that differ in their watershed spatial discretization (a single Grouped Response Unit, GRU, versus multiple GRUs). The MESH model, currently under development by Environment Canada, is a coupled land-surface and hydrologic model. Results will demonstrate the conclusions a modeller might make regarding the value of additional watershed spatial discretization under both an aggregated (single-objective) and multi-objective model comparison framework.
Forecasting Hydrologic Model Prediction Errors During Calibration
Computational budget is a limiting factor in the calibration of computationally intensive hydrologic models. In a conventional calibration, a solution is rejected or accepted based on the model prediction error over the entire calibration time period. But in transient or continuous hydrologic models, there is a relationship between the errors in the early stages of the model simulation and the total prediction error over the entire calibration time period. We propose an approach to model calibration that exploits this relationship by forecasting hydrologic model prediction errors before simulating the model over the entire calibration period. For each model parameter set to be evaluated during calibration, prediction errors are monitored and accrued as the hydrologic model simulation proceeds. If it is forecasted that a given model simulation will not outperform some threshold error value (e.g. the current best solution), the hydrologic model simulation is terminated early; a strategy known as 'model pre-emption'. The value of the model pre-emption strategy is demonstrated for automatic calibration (optimization) case studies using the MESH hydrological model in terms of the potential reduction of computation time required for model calibration. The MESH (version 1.2) model, currently under development by Environment Canada, is a coupled land-surface and hydrological model. Results are evaluated for MESH applied to the Wolf Creek Research basin (located approximately 15 km south of Whitehorse, Yukon, Canada) and Reynold's Creek Watershed (located in Idaho, USA). An analysis of the search history of a previous automatic calibration experiment, performed using the dynamically dimensioned search (DDS) algorithm, suggests that the model pre-emption approach can significantly reduce the computational cost of calibration with absolutely no change to the calibration quality/result.