Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Modeling mass balance of glaciers in the Coast Mountains of British Columbia under the influence of future climate change Thesis Proposal (Draft) Raju Aryal NRES, PhD Program University of Northern BC December, 2008 1.0 Introduction During the last 100 years, the province of British Columbia has experienced warming consistent with trends seen around the globe. Average annual temperature warmed by 0.6ºC on the coast, 1.1ºC in the interior, and 1.7ºC in northern BC over a period of 1895 to 1995 and it is projected to increase by 1oC to 4oC during 21st century (BC Ministry of Water, Land and Air Protection, 2002) (Figure 1). Trend in temperature rise can be seen both in instrumental records as well as in proxies particularly after Little Ice Age maximum in glacier extent. Climate warming is therefore believed to be the main cause for recent glacier recession in BC. In the Canadian Cordillera, glacier recession has been associated with unusually warm mean annual air temperatures and a reduction in winter snowfall since 1976 (Moore and Demuth, 2001; Demuth and Keller, 2006). DeBeer and Sharp (2007) report a net loss of glacier area in the southern Canadian Cordillera through a comparison of historical aerial photography with contemporary Landsat 7 ETM+ imagery. Place glacier in southern British Columbia has been experiencing negative mass-balance since measurement began in 1965 (Moore and Demuth, 2001). The terminus positions of Helm Glacier in southwestern BC and Illecillewaet Glacier in southeastern BC both receded by more than 1,100 metres from 1895 to 1995 (Figure 2). Wedgemont Glacier near Whistler, BC has retreated hundreds of metres in the past two decades alone BC Ministry of Water, Land and Air Protection, 2002). 1 Figure 1. Change in Average Temperature 1895-1995 in BC (°C per century) Figure 2. Changes in glacier terminus of Illecillewaet and Helm glacier in BC, 1895-1995 There has been increasing concern over the loss of glacier ice reserves in BC following climate warming in the region. Glacial meltwater feeds many mountain streams and rivers in BC. Glacier meltwater constitutes a major portion of river flows of these rivers 2 especially during summer when rainwater contribution to the river flow is at a minimum. Although only a small percentage of the total annual runoff of higher-order streams and rivers in the Cordillera is contributed by glacial melt, mountain glaciers tend to moderate interannual variability in stream-flow and help to maintain higher runoff volume during extreme warm and dry periods (Fountain and Tangborn, 1985; Hopkinson and Young, 1998). A decrease in the late-summer flow of glacier-fed rivers throughout British Columbia has been observed, indicating that most glaciers here have already passed the phase of warming induced runoff increases (Stahl and Moore, 2006) suggesting that we might be in a period of diminishing glacier runoff resulting from decrease in glacier covered area. Analysis of historical data revealed significant changes in the stream flow of the upper Similkameen River in BC over the last 30 years (Leith and Whitfield, 1998) (Figure 3). Long term late summer stream flow record downstream of Place Glacier suggests a negative trend in total runoff. This is attributed to significant depletion of firn ice prior to 1965, such that the dominant effect of glacier changes was a reduction in ice area, resulting in decreased meltwater production (Moore and Demuth, 2001). Figure 3. Changing stream flow pattern of Upper Similkameen River in BC, 1971-1995 (Source: Leith and Whitfield, 1998) Decrease in meltwater contribution to stream flow is likely to have an adverse impact on water resources of BC. Hydropower generation, water supply for municipal and industrial 3 uses, agriculture, fisheries etc. are some of the key areas likely to be affected most. Beside changes in flow pattern, glacier retreat is likely to change the temperature of some streams and rivers. These changes, along with other climate-driven changes to hydrological systems, will likely have significant impacts on freshwater and estuarine ecosystems and on aquatic species. They will affect other biological systems and human activities that depend on water. To formulate future water management policies in response to the growing challenges associated with climate change, it is important to conduct climate change impact studies on glaciers of the Coast Mountains. Therefore, the present research aims to assess the impact of climate change on glaciers of the Coast Mountains of western Canada. It is believed that the results from this study will help formulating future water management policies in the region in consideration of future climate change. 2.0 Objective The main objective of this research is to simulate the future mass balance of selected glaciers in the Coast Mountains following climate warming using a Glacier Mass Balance (GMB) model forced by downscaled coarse resolution temperature and precipitation fields at monthly time resolution. While doing so, the research aims to achieve the following goals: i. Develop a model to downscale coarse resolution temperature and precipitation fields to the glacier scale ii. Develop high resolution monthly temperature and precipitation fields for climate change impact studies in the Coast Mountains iii. Develop methods to transfer the simulated mass balance results from selected glaciers to other glaciers in the Coast Mountains 3.0 Study Area Two geographical regions within western Canada have been selected for this research-the Coast Mountains and the Rocky Mountains. 4 Tiedemann glacier is the main glacier in the Coast Mountains selected for this research. Tiedemann glacier is situated on the Mt. Waddington range at the lee (eastern) side of the southern Coast Mountains of British Columbia (Fig. 4). Its mean coordinate is 51°19’N; 124°54’W and has a length of about 24 km (Canadian Glacier Inventory Projecthttp://cgip.wetpaint.com/). Elevation of the glacier ranges from 700 m to 3,800 m. With an area of 63 km2 (Ommanney, 2002), it is one of the more prominent valley glaciers on the east side of the Waddington Range. The ablation area of this glacier is partially covered by debris. Tiedemann glacier is one of the model glacier selected for climate change impact studies by Western Canadian Cryospheric Network (WC2N), a network of research institutions that is studying the past and future impact of climate change on glaciers of western Canada and its implication to freshwater ecosystem. This research will mainly focus on modeling the mass balance of Tiedemann glacier. The ablation areas of some of the larger glaciers in the Coast Mountains are covered with some amount of debris. To simulate the mass balance of glaciers in the Coast Mountains, it is important to understand the influence of debris on ice ablation underneath. In situ measurement of melt rate of bare ice as opposed to melt rate of debris covered ice would give a coefficient of melt which can be included in GMB model for simulating mass balance of debris covered areas. Dome glacier, situated inside Jasper National Park in the province of Alberta (Figure 4), has been selected for taking in situ measurement of thermal properties of debris material and melt rate of ice under the influence of debris. It is one of many glaciers originating from the Columbia Icefield and lies adjacent to Athabasca glacier (52o12.1'N; 117o18.1'W). The glacier has an area of 5.92 km2 out of which 2.16 km2 is covered by debris. Its total length is 5.7 km and elevation ranges from 1980 m at the bottom to 3200 m at the highest point (Canadian Glacier Inventory Project- http://cgip.wetpaint.com/) . The glacier is classified as outlet valley glacier. Dome glacier is preferred over other debris covered glaciers due to ease of accessibility. 5 Figure 4. Location map of Dome glacier in the Rockies (upper right) and Tiedemann glacier in southern Coast Mountains (lower right) 4.0 Data The research will use point observations, interpolated gridded observed datasets, reanalysis datasets and GCM future climate scenarios covering a domain between 128oW-120oW and 48.5oN-53.5oN. These datasets will be considered at monthly temporal resolution and model output will also have same temporal resolution. NCEP reanalysis data These data have been widely used in North America as well as in Europe to calibrate models for downscaling GCM future climate projections. The dataset provides global 6 meteorological fields at 6-hour intervals from 1948 to present. It is based on T62 global model and has grid resolution of 2.5o×2.5o and has 28 vertical levels (Kalnay et al., 1996). The data are available from the NCEP web site (http://www.cdc.noaa.gov/cdc/data.ncep.reanalysis.html). This dataset includes different variables such as temperature, precipitation, SLP, wind, humidity, geopotential height and many more at different pressure levels. The present research will use the NCEP dataset from 1961-2001 for ‘predictor’ variables to calibrate and validate the SD model. These data require some processing before they can be used for calibrating the SD model. This is discussed in detail in methodology section. PRISM based ClimaetBC dataset PRISM (parameter-elevation regressions on independent slopes model) is a regression based model that uses point observations, a digital elevation model (DEM) as well as other spatial datasets to interpolate sparse observations onto a regular grid (Daly et al. 2002) PRISM is based on the assumption that the distribution of temperature and precipitation in a localized region is determined by elevation. Observations from many parts of the world show the altitudinal variations of temperature and precipitation to approximate a linear form. In PRISM, a simple linear climate-elevation regression is employed for each DEM grid cell to vertically extrapolate the climate fields. Linear regression is chosen over other nonlinear methods mainly because altitudinal variation of climate fields often exhibits linear form. In addition to this, linear functions are more stable when extrapolating beyond the elevational range of data and also they can be easily manipulated to compensate for any inadequacies in data. (PRISM Guide Book; http://www.prism.oregonstate.edu/docs/index.phtml)” Monthly precipitation and temperature datasets for part of North America, at 4 km resolution, prepared using the PRISM model, are available from the web site of the PRISM group at Oregon State University. 7 However, the 4 km grid resolution is not suitable for climate impact studies and other ecological modeling in mountain areas. The PRISM climate estimates are based on each tile’s average elevation. According to Wang et al. (2006), PRISM tile elevation of a particular location in mountain areas may differ by up to 1200 m from the actual elevation of that site. This may lead to significant biases in PRISM predicted climate fields. Recently, Wang et al. (2006) presented a method to develop scale–free, high-resolution climate data for western Canada by combining an interpolation technique and elevation adjustment to existing PRISM dataset. The result showed a significant improvement over the original PRISM data both in terms of spatial resolution and prediction precision. This method has been integrated into the computer software called “Climate BC” developed by the Centre for Forest Conservation Genetics (CFCG) at the University of British Columbia (http://www.genetics.forestry.ubc.ca/cfcg/climate-models.html#data). ClimateBC generates the output at any required resolution and for any historical or future duration. As far as historical data are concerned, ClimateBC has downscaled climate fields at two different temporal resolutions. The first dataset includes seasonal and annual climate variables for BC, Yukon and parts of Alberta obtained by downscaling PRISM climate normals for 1961 to 1990. The second dataset has monthly temporal resolution, which is developed by using monthly PRISM climate fields for the period of 1901 to 2001 for the same domain. An example of this scale free data set corresponding to temperature normals for 1961-1990 is shown in Figure 5. ClimateBC monthly temperature and precipitation fields will be used as an observed dataset while developing a model for downscaling GCM climate projections. 8 Figure 5. Scale-free mean annual temperature in BC for reference period 1961-1990 (upper) and for future (lower) derived from the PRISM dataset (Wang et al., 2006) Historical Weather Data There are a number of weather stations in the Coast Mountains operated by Environment Canada, BC Ministry of Highways and BC Ministry of Forests and Range (Figure 6). These stations collect different meteorological parameters at hourly and daily temporal resolution. These data will form an important resource for comparing with the ClimateBC temperature and precipitation datasets to ascertain whether the latter can be used as a replacement to the observed data for developing models for climate downscaling. Data for the period of 1961-2001 will be considered for the present research. Some of the required observed data have already been stored at the UNBC HPC data archive. The remaining 9 data will be acquired from the relevant agencies. Besides this, the climate record observed at the Tiedemann glacier will also be used to determine their closeness to gridded ClimateBC dataset. Figure 6. Climate station locations within study area GCM data sets Present day GCMs are Atmosphere-Ocean Coupled Global Climate Model that simulate past and future climates under different scenarios of greenhouse gas emissions. There are several CGMs developed by different climate research institutions worldwide. Performance of these models varies from model to model because each has its own physics. To determine the range of uncertainly in model output, climate change impact studies often consider outputs from different models under different scenarios of greenhouse gas emission. The present research will consider ensembles of projected climate fields developed using several GCMs under different scenarios of greenhouse gas emissions. Recent IPCC work (IPCC, 2007) used several GCMs under a range of emission scenarios to develop ensembles of future climate scenarios. Downscaling will be applied to temperature and precipitation outputs from 4 models in the IPCC archive: (1) CGCM3.1 10 developed by Canadian Center for Climate Modeling and Analysis, ECHAM5 of The Max Planck Institute for Meteorology, HadCM3 developed by Hadley Centre for Climate Prediction and Research and PCM of National Center for Atmospheric Research. Three simulations from each model will be considered for downscaling: 20th century simulation forced by historic conditions with variations of anthropogenic greenhouse gases, anthropogenic and natural aerosols, and solar output, and two forced by future emissions scenarios, the SRES A2 and B1. The historical and future time period of model simulations vary from one model to another. These time periods typically range from the middle of 17th century to the end of 24th century. For the present research 20th century simulation data will be considered for the period of 1961-1990 while the future simulation corresponding to SRES A2 and B1 scenarios will be considered for the period of 2001-2100. The historical GCM simulation will be used to correct the bias from the future GCM simulations. Historical and future simulations of climate variables from models mentioned above are available from the IPCC Data Distribution Centre (http://www.ipcc-data.org/). Details of these GCM models are provided in Table1. RAMS regional climate data One of the research groups within WC2N is employing dynamic downscaling technique to obtain high resolution climate fields for the Coast Mountains to be used in GMB model. This group is using the RAMS meso-scale model at 8 km resolution to downscale regional climate fields from 1979 to 2008. Downscaled variables include temperature, precipitation, relative humidity, incoming solar radiation, albedo, wind and many more for different pressure levels. Besides the PRISM dataset, these data will also be used as predictands to develop a model for downscaling global climate projections to the local scale. Field data Field data used for this study include both primary and secondary data. Primary data to be collected from the site are: weather data, mainly air temperature, obtained from Automatic Weather Stations at different elevations along the Tiedemann glacier, melt rate of ice under various debris thickness, debris thicknesses and distribution on glacier surface. Secondary 11 data will include historical mass balance records and climate records for calibrating and validating GMB model. A description of data to be used for this research is given in Table 1. Table 1 Description of data to be used for the research Description Parameters NCEP-global Temperature, meteorological fields precipitation, PRISM Duration 1961-2001 wind, Temporal resolution resolution 2.5o×2.5o 6 lat/lon Daily Source hourly/ National Centers of Environmental insolation, humidity, Protection geopotential height, (www.cdc.noaa.go SLP at 500, 850 and v/cdc/data.ncep.rea 1000 hPa nalysis.html) based Temperature ClimateBC dataset Spatial and 1961-2001 <200m Monthly Precipitation Centre for Forest Conservation Genetics, UBC (http://www.geneti cs.forestry.ubc.ca/c fcg/climatemodels.html) Historical Weather Temperature Data and 1961-2001 precipitation Point Daily observation Environment Canada, BC Ministry of Forests, Ministry of Highways Historical mass Winter, summer and Variable balance record <200 m seasonal Variable Daily, IPCC monthly Distribution net mass balance Place glacier Bridge glacier Sentinel glacier GCM future climate Temperature scenarios and 2001-2100 under precipitation SRES A2 and B1 Centre 12 Data CGCM3.1 2.8o×2.8o ECHAM5 1.9o×1.9o HadCM3 2.5 ×3.75 PCM 2.8o×2.8o GCM 20th o century Temperature climate (http://www.ipccdata.org/) o and 1961-1990 precipitation CGCM3.1 o o o o 2.8 ×2.8 ECHAM5 1.9 ×1.9 HadCM3 2.5 ×3.75 PCM 2.8o×2.8o DEM o Area-altitude Recent Daily, IPCC monthly Distribution Centre (http://www.ipcc- o data.org/) <200 m UNBC GIS data distribution In situ measurements resource Temperature and Summer surface melt rate of 2008 bare and of Point and observation Hourly/ Direct Daily measurement debris 2009 at Tiedemann glacier covered glacier ice, (continuous heat Data and Dome glacier conduction measurement through debris layer, at least for 15 meteorological days) elements RAMS downscaled Temperature, dataset precipitation, 1979-2008 8 km wind, Hourly/ WC2N research Daily group seasonal Current research sensible and latent heat fluxes, incoming goal radiation, outgoing longwave radiation, albedo etc Air photo Vertical surface 2008-??? changes 13 <50 m ?? 5.0 Methodologies This research will mainly comprise 3 parts: Part I: Development of high resolution meteorological fields for the Coast Mountains Part II: Field measurements Part III: Mass Balance modeling 5.1 Part I: Development of high resolution meteorological fields Climate change impact studies on glaciers require the equivalent of point climate observations and are highly sensitive to fine-scale climate variations that are parameterized in global or regional climate models. This is especially true in regions of complex topography, coastal or inland locations similar to those found in the Coast Mountains of western Canada. This part of the research will focus on using Statistical Downscaling (SD) techniques for downscaling global meteorological fields to the glacier scale for assessing the impact of climate change on glaciers of the Coast Mountains. SD methods will compliment the similar research work being conducted within WC2N, which is employing a dynamical downscaling technique at 8 km resolution using the meso-scale meteorological model (RAMS) for 1979-present. Statistical downscaling SD is an alternative approach to Regional Climate Model (RCM0 for large scale climate downscaling. This approach of climate downscaling involves developing a statistical model, which links large-scale climate variables (or ‘predictors’) to regional and local variables (or ‘predictands’). SD approach is based on assumptions that the regional climate is a direct result of interaction between large scale climate state and regional/local physiographic features such as topography, land-sea distribution and land use (Von Storch, 1995, 1999). Once a model is calibrated, the large-scale output of a GCM simulation is fed into this statistical model to estimate the corresponding local or regional climate characteristics. Unlike dynamical methods, statistical methods do not require a large amount of computational resources, and they can therefore be easily applied to output from different GCM experiments. Another advantage is that they can be used to provide site14 specific information, which can be critical for many climate change impact studies. However, being an empirical model, SD does not explicitly describe the physical processes limiting their wider applicability. In addition, the major theoretical weakness of SD lies in the fundamental assumption that they are based on-that the statistical relationships developed for the present day climate remain valid for future climates- an assumption which is often not verifiable. This limitation also applies to the physical parameterization of dynamical models. As mentioned above, SD involves developing quantitative relationships between largescale atmospheric variables (predictors) and local surface variables (predictands). The most common form has the predictand as a function of predictor(s), but other types of relationships have also been used. For example, between predictors and the statistical distribution parameters of the predictand (Pfizenmayer and Von Storch, 2001) or the frequencies of extremes of the predictands (Katz et al., 2002) There are a number of techniques used in SD of large scale climate variables. Not all techniques perform well for all regions and for all datasets. The choice of SD techniques therefore depends upon their performance in simulating the observed climate, especially their ability to reconstruct the observed variance. Therefore, rather than focusing on a single SD technique with limited number of predictors and predictands, the present research aims to use different SD techniques with different combinations of predictors and predictands (in terms of types of variable used and their temporal resolution). This method produces a number of SD models; but all of these models might not be suitable to be used for downscaling GCM fields. Only models which reproduce observed fields with a reasonable degree of accuracy in terms of the absolute value and the variance will be selected for downscaling GCM projected temperature and precipitation. This will produce ensembles of downscaled temperature and precipitation fields for a particular location. Ensembles are widely used in climate change impact studies mainly because these help understand the range of uncertainty present in the result due to uncertainty in model parameter and input data. These ensembles can be averaged together to improve the accuracy of projected fields at the glacier scale. 15 Since the main goal of this research is to downscale the GCM future temperature and precipitation scenarios to the glacier scale for GMB simulation, choice of GCM model for getting future temperature and precipitation scenarios is very important. However, there is no strong basis for selecting the outputs from one model and rejecting that from others because all GCM outputs inherit some degree of uncertainty. To determine the range of uncertainty in model output, the present research will consider ensemble temperature and precipitation outputs from 4 different GCMs under SRES A2 and B1 emission scenarios. GCMs are discussed in detail in the following chapters. Predictors NCEP data (or ‘predictor’ variables) will be acquired from the NCEP web site for the regions covering 128oW-120oW and 48.5oN-53.5oN and for a period of 1961-2001. Maximum and minimum temperature, precipitation, wind, insolation, geopotential height at the 500 hPa, 850 hPa and 1000 hPa levels and sea level pressure (SLP) are some of the predictors widely used for downscaling. Widemann et al. (2003) investigated several combinations of predictors and found that large-scale precipitation from the global model is a robust predictor for Pacific Northwest precipitation. Sea level pressure can be included as a secondary predictor to capture the effects of interaction of atmospheric circulation and the topography. Similarly, the study claims that the large-scale surface air temperature is a robust predictor for regional temperature. In contrast to precipitation, there was found to be little additional skill in including a circulation parameter, so this single predictor is sufficient (Salathe et al., 2007). The decision on choice of predictor variables cannot be made unless they are tested independently. Therefore, the present research will consider all potential predictor variables and test each of them independently to determine which variable best reproduce the observed temperature and precipitation at the local scale. These predictor variables will be considered for 500 hPa, 850 hPa and 1000 hPa covering the entire elevation range of Tiedemann glacier. The mean of three levels gives the average of the predictor variables for the entire glacier. Instead of using raw NCEP data directly, downscaling work mostly uses the normalized data for calibrating the SD models. Therefore, normalization of all the potential predictor 16 variables will be performed over a period of 1961-1990 before using them for calibrating the SD model. Prior to normalization, these datasets will be interpolated to individual GCM grid resolutions whose simulations are to be downscaled. Since the present research aims to use output from 4 different GCMs, predictor variables for calibrating the SD model will be slightly different from one model to another as a result of which there will be 4 different SD models corresponding to 4 different GCMs. Some of the potential predictor variables interpolated to CGCM2 grid resolution and normalized with respect to their 1961-1990 means and standard deviations, are available from the website of Canadian Climate Change Scenario Network (CCCSN, http://www.ccsn.ca/index-e.html). Predictands Monthly temperature and precipitation will be the two predictand variables used for calibrating the SD model. Three different sets of predictand variables will be used: 1) the scale free ClimateBC dataset (Wang et al. 2006), 2) the RAMS model output for the Coast Mountains (personal communication, Bruce Ainslie, WC2N), and 3) point observation data will be used with NCEP predictor variables to calibrate the SD model. The SD model corresponding to predictand variables (ClimateBC, RAMS or point observation) having the highest skill to specify the local climate will be selected for downscaling future climate projection. Similar to predictor variables, predictand variables are also normalized over the period of 1961-1990. Environment Canada, the BC Ministry of Highways and the BC Ministry of Forests and Range operate a number of weather stations in BC. These stations collect different meteorological parameters at hourly and daily time resolution. Daily maximum temperature, minimum temperature, mean temperature and total daily wet day precipitation (>0.25mm/day) in the vicinity of Tiedemann glacier will be taken from these sources. These data will be taken for a period of 1961-2001 which corresponds to the period of NCEP predictor variables considered. These data will be subjected to preliminary analysis in which they will be adjusted for inhomogeneities caused by non-climate factors, such as station relocation and change in observing practice. 17 The RAMS dataset for 1979-present, which is available in 8 km spatial resolution, will be further interpolated to the DEM resolution (<200 m) before using them with NCEP fields for model calibration. The time scale of the SD will be monthly. All hourly and daily data will be converted to monthly means before using them for downscaling. Several downscaling techniques will be employed to evaluate their skill in determining the local temperature and precipitation fields from their large scale values. Techniques which best specify the local climate both in terms of the absolute value and the observed variance will be selected for downscaling global fields. Some of the downscaling techniques to be employed for the present research are discussed below. Statistical DownScaling Model (SDSM) Standard software, such as Statistical DownScaling Model (SDSM) is available for downscaling of large scale fields (Wilby et al., 2002). SDSM calculates statistical relationships, based on multiple linear regression techniques, between large-scale (the predictors) and local (the predictand) climate. Lines et al. (2006) successfully employed SDSM to downscale the temperature and precipitation fields in Atlantic Canada. SDSM is a hybrid of multiple regression and stochastic downscaling methods. Observed data sets (predictands) are first regressed against a ‘selection’ of climate predictor(s) to develop regression equations. SDSM is said to be calibrated when the regression coefficients, explained variance, and standard error are within acceptable limits for each regression model. SDSM can be downloaded free of charge from the SDSM UK website (http://wwwstaff.lboro.ac.uk/~cocwd/sdsm.html). The methodology is fully described in the SDSM ‘User Manual’, by Wilby et al. (2001), which can be downloaded form the same web site. Goldstein et al. (2004) and Barrows et al. (2004) showed that, of several statistical downscaling models, SDSM produced optimal results. Different predictors necessary as inputs to SDSM are also available 18 from CCCSN web site (http://www.ccsn.ca/The_Network/The_Network-e.html). Relevant information is also available from the web site of Canadian Climate Impacts Scenarios at the University of Victoria (http://www.cics.uvic.ca/scenarios/index.cgi?Scenarios) Linear downscaling method Huth (1999) evaluated several linear downscaling methods to determine their ability to specify the local climate at different locations in central Europe from the NCEP reanalysis dataset. He examined three linear methods of statistical downscaling: (1) canonical correlation analysis (CCA) pre-filtered by principal component analysis (PCA); (2) singular value decomposition (SVD) and (3) multiple linear regression (MLR). Three different MLR models were considered: (1) stepwise screening of principal components (PCs) (‘stepwise regression’), (2) MLR on PCs without screening, that is, all PCs being forced to enter the model (‘full regression’), and (3) stepwise screening of gridded values (‘pointwise regression’). The pointwise regression proved to be the best method of those considered whereas SVD method appeared to be the worst (Huth, 1999). However, Widmann et al. (2003) used SVD technique to successfully downscale precipitation fields over northwestern United States using NCEP data. Based on the result of the past studies, the present research plans to employ SVD and MLR methods for downscaling large scale temperature and precipitation field. Local scaling method Some studies have used a downscaling method referred to as ‘local scaling’ to downscale precipitation and temperature in western North America (Widmann et al., 2003; Salathe, 2005). Radic and Hock (2006) successfully applied a local scaling method to downscale global projection of temperature and precipitation to Storglaciären in Sweden for assessing climate change impacts on glacier mass balance. The local scaling method for precipitation downscaling simply multiplies the large-scale simulated precipitation at each local gridpoint by a seasonal scale factor (Widmann et al., 2003). The scaling factor is derived during a fitting period to remove the long-term bias between the large-scale simulated precipitation and the observed precipitation at that gridpoint. The downscaled monthly mean precipitation is given by: 19 Pi (t ) Pi ,c (t ) Pobs Pc i 1,...........,12 (1) where Pi,c is monthly precipitation sum from the climate model for the duration of simulation (t; e.g. t = 2001 to 2100), Pobs and Pc are mean precipitation from observation and climate model, respectively, averaged over the baseline period. Surface air temperature is downscaled in a similar way to precipitation. For temperature the adjustment is additive; thus, the downscaled monthly mean surface temperature is given by (Salathe, 2005) : Ti (t ) Ti ,c (t ) (Ti ,obs Ti ,c ) i 1,...........,12 (2) where Ti,c is monthly temperature for the ith month from the climate model for the duration of simulation (t; e.g. t = 2001 to 2100), Ti ,c and Ti ,obs are mean temperature from climate model and observation, respectively, for the ith month averaged over a chosen baseline period Beside point observations as predictands, ClimateBC dataset and RAMS fields will also be used as equivalent to observed data for downscaling GCM future projections directly using above mentioned equations. Model calibration and validation It is a general practice to divide the historical dataset into two different time periods to use them separately for model calibration and validation. For the present research, the monthly averaged predictand variables (ClimateBC, point observation or RAMS) and relevant predictor variables covering a period from 1961-2001 will be first divided into two parts: 1961-1981 and 1981-2001, each covering a period of 20 years. The first 20 years dataset will be used for calibrating the SD model while the remaining 20 years data will be used to validate the model output. Once the model is calibrated using first 20 years (1961-1981) data, model skill validation will be performed by forcing the model with relevant NCEP 20 predictor variables for the remaining 20 years period (1981-2001) to obtain downscaled temperature and precipitation fields at the glacier scale. The downscaled result will be compared with both point observations and ClimateBC dataset for corresponding period and for corresponding locations. Skill testing will be performed for all variants of SD models developed using different downscaling techniques and different combinations of predictors and predictands. Downscaling GCM simulations GCM output always inherits some degree of uncertainty and should be used cautiously while using them in climate change impact studies. To overcome this problem, rather than considering outputs from single model and emission scenarios, climate researchers always prefer considering outputs from different GCM models under a range of greenhouse gas emission scenarios, also known as ensemble results. This research will therefore consider temperature and precipitation outputs from several GCMs under different emission scenarios. Downscaling will be applied to temperature and precipitation simulations from 4 different GCMs: (1) CGCM3.1 developed by Canadian Centre for Climate Modeling and Analysis, ECHAM5 of The Max Planck Institute for Meteorology, HadCM3 developed by Hadley Centre for Climate Prediction and Research and PCM of National Center for Atmospheric Research. Summary of these models are provided in Table 2. Three simulations will be considered for each model: one simulation forced by the 20th century historic conditions and two future simulations forced by IPCC SRES A2 and B1 scenarios. 20th century simulations will be considered for the period 1961-1990 while for future simulation a period of 2001-2100 will be considered. GCM outputs will be considered from all grid cells within the research domain. 21 Table 2 GCM models and scenarios to be used in the proposed research Models Scenarios Period CGCM3.1 Historic 1961-1990 SRES A2 2001-2100 SRES B1 2001-2100 Historic 1961-1990 SRES A2 2001-2100 SRES B1 2001-2100 Historic 1961-1990 SRES A2 2001-2100 SRES B1 2001-2100 Historic 1961-1990 SRES A2 2001-2100 SRES B1 2001-2100 ECHAM5 HadCM3 PCM The downscaled GCM simulations do not always agree with the actual observation; they are either negatively or positively biased. Before applying downscaled GCM results for climate change impact assessment, bias correction must be performed. One of the commonly used approaches is to compare the historic simulations with the observed values for the corresponding period, determine differences between to data series and correct the future projections by adding or subtracting the difference from the corresponding data series of future climate fields. Downscaled 20th century temperature and precipitation for the period 1961-1990 will be compared with the corresponding ClimateBC dataset. Errors will be determined for each data value which is then applied to downscaled GCM future temperature and precipitation to correct potential biases. The bias corrected future temperature and precipitation fields will then be used to force GMB model for future projection of glacier mass balance in the Coast Mountains. Procedures are discussed in detail under methodology section in the following chapters. 22 5.2 Part II: Field work Tiedemann glacier The distribution of debris properties in the ablation area, especially debris thickness, is a key input variable when simulating mass balance of Tiedemann glacier. In addition to this, surface air temperature at different elevation bands at or/and near the glacier surface and glacier surface temperature are other variables of interest. Some surface air temperature data recorded at different elevation on the glacier (Figure 7) and surface air temperature recorded along the left lateral moraine at different elevation is already available. This measurement work will be continued in the future. Although these variables do not enter into the GMB model directly they are still useful in validating short term downscaled climate fields on Tiedemann glacier. Given the enormous size of the glacier and dangerous surface conditions, it is not possible to cover the entire glacier for taking measurements of different parameters of interest. This field work will therefore focus only on lower part of the Tiedemann glacier, approximately about 5-8 km up from the glacier terminus. In addition to this, existing Automatic Weather Station near the glacier will be regularly maintained to obtain continuous weather data (Figure 8). This station started collecting data since 2005 but there has been significant data loss due to frequent damage of station following heavy winter snowfall. Tiedemann Temperatures (2006) 30.00 T1 (506 m) T2 (852 m) 20.00 15.00 T3 (1045 m) 10.00 T4 (1352 m) 5.00 T5 (1838 m) 0.00 T6 (3020 m) -5.00 Date 23 8-Oct 3-Oct 28-Sep 23-Sep 18-Sep 13-Sep 8-Sep 3-Sep 29-Aug 24-Aug 19-Aug 14-Aug 9-Aug 4-Aug 30-Jul 25-Jul 20-Jul -10.00 15-Jul Temperature (o C) 25.00 Figure 7. Daily(?) surface air temperature recorded at different elevation in Tiedemann glacier during the summer of 2006 20.00 Air Temperature ( o C) 15.00 10.00 5.00 0.00 -5.00 -10.00 -15.00 6/1/2007 6/15/2007 6/29/2007 7/13/2007 7/27/2007 8/10/2007 8/24/2007 9/7/2007 6/1/2007 6/15/2007 6/29/2007 7/13/2007 7/27/2007 8/10/2007 8/24/2007 9/7/2007 5/4/2007 5/18/2007 4/6/2007 4/20/2007 3/9/2007 3/23/2007 2/9/2007 2/23/2007 1/26/2007 1/12/2007 -20.00 Time 4.50 4.00 Snow Depth (m) 3.50 3.00 2.50 2.00 1.50 1.00 0.50 5/4/2007 5/18/2007 4/6/2007 4/20/2007 3/9/2007 3/23/2007 2/23/2007 2/9/2007 1/26/2007 1/12/2007 0.00 Time Figure 8. Automatic Weather Station at Tiedemann glacier (left). Daily mean values of air temperature (upper right) and snow depth (lower right) recorded at the station. Dome glacier Dome glacier in the Rocky Mountains has been selected for taking measurement concerning the influence of debris on melt rate of ice underneath. The measurement work will include taking daily records of melt rate of ice (in terms of ice thickness) under variable debris thickness and from bare ice surface. Based on the data collected from this measurement, a ratio of melt rates corresponding to particular debris thickness and bare ice surface will be determined. This ratio gives a Coefficient of Melt (CM) for that particular debris thickness. CM will be determined for all the debris thicknesses considered and each CM value will be averaged over the period of measurements to obtain a single CM value for a particular debris thickness. Corresponding CM values will be applied to GMB model 24 when simulating summer mass balance (ablation) of debris covered part of the glacier. The first field experiment at Dome glacier commenced in mid-August 2008 during which measurements were taken for 12 days. A measurement site was prepared near the terminus of the glacier. The measurement site consisted of an Automatic Whether Station equipped with sensors to measure air temperature, humidity, solar radiation (incoming and reflected), net radiation, wind, thermistors (4) to record temperature at different debris depths and plots to take daily measurements of melt rate under 0 cm (bare ice), 5 cm, 10 cm, 15 cm, and 20 cm of debris thicknesses (Figure 9). Preliminary analysis of some of the recorded data is shown in figure 10. Results show that for debris thickness greater than 10 cm, melt rate does not change significantly unless there is an abrupt change in meteorological condition. This can be seen in Figure 10 where melt rates from 10 cm and 15 cm almost converge with each other. To further improve the accuracy of CM, a detail long term measurement of melt rate under range of debris thicknesses is planned during summer of 2009. This field measurement will be a continuation of 2008 measurements but the measurement of melt rate will be extended to more debris thickness with increased duration and frequency of measurements. Measurements will be conducted for the debris thicknesses of 0 cm, 2 cm, 5 cm, 7 cm, 10 cm, 12 cm, 15 cm and 17 cm and 20 cm so that CM can be determined for a range of debris thicknesses. These data can also be used to develop a relationship between melt rates and the debris thicknesses. Total duration of this experiment will be at least 20 days and the temporal resolution of measurement will be increased from daily to twice daily. 25 Figure 9. Automatic Weather Station (left) and in situ measurement of melt rate under variable debris thickness at the Dome glacier during August 2008. 26 100.0 90.0 Bare Ice Daily Surface Lowering (mm) 80.0 70.0 60.0 5 cm debris layer 50.0 40.0 10 cm debris layer 30.0 15 cm debris layer 20.0 10.0 0.0 8/15/2008 8/16/2008 8/17/2008 8/18/2008 8/19/2008 8/20/2008 8/21/2008 8/22/2008 Date 18.0 Daily Air Temperature (oC) 16.0 14.0 12.0 10.0 8.0 6.0 4.0 2.0 0.0 8/13/2008 8/14/2008 8/15/2008 8/16/2008 8/17/2008 8/18/2008 8/19/2008 8/20/2008 8/21/2008 Date Figure 10. Daily surface lowering (equivalent ice melt) on bare ice and on debris of variable thicknesses (top). Lowering is consistent with daily air temperatures measured for the same time period (bottom). 5.3 Part III: Mass balance modeling Mass-balance models are widely used in many parts of the world to assess the glacier’s response to climate parameters (Braithwaite and Olesen, 1985; Tangborn, 1999; Hock and Holmgren, 2005; Radic´ and Hock, 2006; Hock et al., 2007; Bhatt et al., 2007). Massbalance models have also been widely used in modeling the response of glaciers to future climate change (Oerlemans et al., 1998 and 2005; Radic´ and Hock, 2006). In general, two 27 categories of GMB model exist: one that is based on an energy balance approach (Hock and Holmgren, 2005) and the other on an empirical temperature-index model (Hock, 2003). GMB models based on energy balance approach are physically based models which derive melt as the residual in the energy-balance equation. Being physically based model, they often require detailed data input which is one of the limitations of this model. On the other hand, temperature-index models do not require detailed data input for melt simulations, but their modeling skill is often less reliable due to their empirical nature. Hock et al. (2007) intercompared the sensitivity of mass-balance projections to the choice of mass-balance models. They applied five different mass-balance models to one glacier in Northern Sweden for mass-balance simulation, namely, (1) zero-dimensional temperatureindex regression model, (2) elevation-dependent temperature-index regression model, (3) distributed temperature-index model including potential direct solar radiation, (4) elevation-dependent simplified energy-balance model, and (5) distributed energy-balance model. Of the results from all the models, the temperature-index model performed well in reproducing glacier mass-balance components. de Woul and Hock (2005) applied a simple degree day approach in Arctic glaciers and ice caps to estimate static mass balance sensitivity to temperature and precipitation change. Radic´ and Hock (2006) successfully applied the same approach to simulate future mass balance of Storglaciären in Sweden. Tangborn (1997, 1999); Tangborn and Rana (2000); Bhatt et al. (2007) and Zhang et al. (2007a, b) used the Precipitation-Temperature-Area-Altitude (PTAA) model to produce mass-balance hindcasts of a few glaciers in Northwestern America and Himalaya. The use of temperature index model is justified from the fact that the confidence level of GCM projected temperature and precipitation is relatively higher than other projected parameters, such as global solar radiation, and therefore the future mass balance projection developed using Temperature Indexed (TI) model is likely to be more realistic than mass balance projection made with distributed models. The performance of TI model varies from one region to another because the individual model is configured to suit the topography and climate of a particular location. Therefore, a good deal of model calibration and parameterization is always desired before using TI model in a particular region. 28 The present research will use a temperature index model following de Woul and Hock (2005) and Radic´ and Hock (2006). Summer mass balance, bs, and winter mass balance, bw, are modeled by: t2 bs s aiTi s i t1 t2 ai =1, Ti > 0 , bw w ai Pi w , i t1 (3) a i =0, Ti ≤ 0 ai =1, Ti < T0r/s a i =0, Ti ≥ T0r/s (4) where Ti is the downscaled air temperature (oC). α and β are coefficients derived from linear regression between measured summer mass balances (bs) and positive degree-day sums (ΣaiTi) over the entire mass balance year, and between measured winter mass balances (bw) and annual sums of daily precipitation (ΣaiPi) with air temperatures below the threshold temperature T0r/s that discriminates rain from snowfall. The monthly ClimateBC dataset will be used to linearly regress bs and bw of the entire glacier with ΣaiTi and ΣaiPi respectively. α and β will be assumed to be constant throughout the glacier. bs and bw will be estimated for every DEM grid cell keeping α and β constant. Summer mass balance from debris covered area The lower part of the ablation area of Tiedemann glacier is covered by debris. The effect of debris is to inhibit the melt of underlying ice, thus considerably affecting summer mass balance. To address this problem, a field experiment was conducted at the debris covered Dome glacier in the Rocky Mountains during the middle of August 2008. A detailed measurement is planned again during the summer of 2009. The August 2008 experiment included conducting direct measurement of melt rates of ice under variable debris thicknesses and under the condition of bare ice. Observation of different meteorological elements along with measurements of spatial distribution of size, thicknesses and type of 29 debris was also conducted. The field methodology and the method to determine Coefficient of Melt (CM) are already discusses in section 5.2. For every DEM grid over Tiedemann glacier containing the debris, the summer mass balance will be estimated by scaling Eq. 3 with the CM for corresponding debris thickness obtained from the experiment conducted at the Dome glacier. A detailed survey of debris thicknesses and debris distribution on the ablation area of Tiedemann glacier is therefore required. Calibrating model parameters and output validation One of the limitations of the model discussed above is that it needs to be calibrated based on historic temperature, precipitation and mass balance data, thus hampering direct transferability to other glaciers. Validation of model output is equally important to confidently use the model for future projection of glacier response to climate change. Model calibration for Tiedemann glacier is not possible because no long term continuous historical mass balance data exist for this glacier. Short term intermittent mass balance records, dating back to 1980’s and 1990’s, have been found to exist for this glacier. However, these data may not be suitable for model calibration and validation because of their intermittent nature. As an alternative to this, present research plans are to calibrate and validate the model in glacier with similar geographic settings. Place glacier has been continuously monitored since 1965 and has the longest mass balance record (Stanley, 1975). Many researchers have used this mass balance record for assessing climate-glacier relationship in the past (Moore and Demuth, 2001; Rasmussen and Conway, 2004). Place glacier is located in the same mountain range, has similar exposure and is separated by an aerial distance of only 200 km. Bridge glacier and Sentinel glaciers are two more glaciers in the Coast Mountains with long term mass balance record. Both of these glaciers are also situated in the southern Coast Mountains in reasonably close proximity to the Tiedemann glacier. It is assumed that all of these glaciers respond with climate variables in similar fashion. Mass balance records for Bridge and Sentinel are available since the 1960’s but they are not as continuous as the records from Place glacier. Characteristics of these glaciers and the duration of mass balance records available are listed in Table 2. Table 2 Characteristics and available mass balance records for selected glaciers in the Coast Mountains 30 Name Lat/Lon Area Length Lowest/highest Mass balance (km2) (km) elevation records available (m a.s.l) Tiedemann glacier 51°19’N/ 63.0 24.0 700/3800 124°54’W Place glacier 50°25.3'N/ Summer of ‘81, 1 ‘89, ’90, ‘91 3.8 ~3.0 1850/2600 2 83.0 ~16.0 1400/2900 2 Since 1965 122°36.0'W Bridge glacier 50°49.4'N 1970-1980 /123°33.0'W Sentinel glacier 49°53.6'N/ 2 1.8 1970-1980 122°58.9'W Sources: Canadian Glacier Inventory Project (http://cgip.wetpaint.com/); 2Ommanney, 2002; 1BC Hydro Report, Google Earth 2008. To calibrate GMB model given in equation (3) and (4), the time series of mass balance records from these glaciers will be regressed with seasonal ClimateBC temperature and precipitation record for the corresponding period. Regression will be performed with available mass balance records for all the glaciers. Prior to calibration, the available mass balance records and corresponding ClimateBC temperature and precipitation records will be divided into two parts: 1961-1981 and 1981-2001. The first part of the records will be used to calibrate the model while the second part will be used to validate the model. Once the model calibration is completed, the best fit GMB model for a particular glacier will be selected and forced with ClimateBC monthly temperature and precipitation for 1981-2001. Glacier mass balance hindcasts thus developed will be compared with historical mass balance records of the corresponding periods to test the model skill. If the model is able to reproduce the measured mass balance of the selected glacier to an acceptable degree of accuracy, we assume that the model is capable of simulating future mass balance of other glaciers in the Coast Mountains well. 31 From September 2008, photogrammetric method of mass balance measurement has been initiated on Tiedemann glacier. This involves taking repeated aerial photos of glaciers, at least once a year. An estimation of glacier surface elevation change will be made based on time series of aerial photographs which can then be used to estimate the mass balance indirectly. However, a few years of data are required to make a meaningful estimation of mass balance. Projection of future glacier mass balance Once the GMB model is able to simulate the observed mass balance to an acceptable degree of accuracy, it will be used to project the future mass balance of selected glaciers in the Coast Mountains in response to future projection of temperature and precipitation. The model will be forced by downscaled GCM projection of temperature and precipitation fields under SRES A2 and B2 emission scenarios in monthly time scale for a period extending from 2001 to 2100. An ensemble of mass balance projection will be developed corresponding to different temperature and precipitation fields downscaled using different SD models and different sets of outputs obtained from 4 different GCMs and two different emission scenarios. This will help determine the range of uncertainly in simulated mass balance fields which may result due to uncertainty in SD models, GCM projections and NCEP reanalysis data. Mass balance ensembles will be developed for every individual year and averaged for every 10 years starting from 2001 until 2100. A complete process of climate downscaling and mass balance modeling is schematically shown in Figure 11 below. 32 Figure 11. Schematic diagram showing complete process involved in simulating future mass balance 6.0 Potential contribution and uniqueness of the research The Coast Mountains of BC have the largest concentration of mountain glaciers in Canada. In British Columbia alone, the total glacier area exceeds 30,000 km2. Depletion of the glacier ice reserve from the Coast Mountains following climate change is a major concern because many larger rivers in the region derive a sizable portion of their flows from glacier melt runoff. The importance of glacier melt on river runoff varies from river to river. For a river basin with a large glacier covered area in its headwater, the glacier melt contribution is often very large. Even for a river basin with a small glacier covered area, the glacier meltwater can be very important for sustaining low flow conditions in rivers during the dry season when water demand is highest. For much of BC, glaciers play an important role in supplying communities with freshwater for irrigation, drinking, hydroelectric power generation and industrial production. The freshwater supply from glacier melt is also extremely important for sustaining aquatic ecosystems. Power generation during summer 33 months in BC depends heavily on glacier and snowmelt runoff. About 85% of the electricity produced in British Columbia comes from hydropower plants. (Web page: BC Ministry of Energy, Mines and Petroleum Resources). In some regions of BC, glacier meltwater forms an important part of water supplies for communities during the dry season when rainwater contribution to surface runoff is minimal. There is a widespread social, economic and environmental implication of glacier loss in BC. The proposed research will improve our present understanding regarding the response of glaciers to future climate change. It will provide key information to government authorities to formulate water management policies following the changes in glacier ice reserve in future. Water manager can use the results from this research to effectively manage dwindling water supply in future. Runoff modeling is an important tool often used by water manager to manage the river water supplies. The ensemble of mass balance projection obtained from this research can be applied to runoff models to develop scenarios of possible changes in runoff patterns in future which can be very helpful for managing the water resources. As many economic sectors such as tourism, industry, hydropower and agriculture depend on freshwater supply for their sustenance, effective management of this precious resource is a most for sustaining the regional economy. A good management of water resource is equally important to sustain the aquatic ecosystem and wildlife habitat which together keep the environment in good shape. This research will also open up new avenues of different researches such as estimating the contribution to sea level rise due to accelerated melting of glaciers in the Coast Mountains. Since climate downscaling is a major component of this research, it will provide high resolution climate fields, which can then be used for different ecological modeling and climate change impact studies in the Coast Mountains. Similarly, there has not been enough research to understand how debris covered glaciers respond to climate warming although they make up a sizable portion of total glacier area in the Coast Mountains. This research will provide scientific information regarding how debris covered glaciers responds to temperature change and how they influence the rate of ablation of glacier ice underneath. This will help understand the effects of debris on melt water contribution to the rivers downstream from the perspective of future temperature rise. 34 Research results will not make any sense unless it is made public to the relevant scientific communities to gain the general acceptance. Therefore, the present research aims to publish at least 3 papers in international, peer reviewed journals. The following is the tentative list of areas within this research with the potential to develop into good papers: Influence of debris on glacier melt SD of large scale climate fields in the Coast Mountains Projection of future GMB using downscaled GCM fields The proposed research is unique because: It is for the first time future projection of mass balance is going to be made on glaciers in the Southern Coast Mountains of BC following the future climate warming. Although climate downscaling work has been done in this region, no research has been found to exist which uses the downscaled results for the climate change impact studies on glaciers. Despite the fact that a sizable area of glaciers in the Coast Mountains are covered with some amount of debris in their ablation areas, no mass balance modeling work in this region found to have included the effects of debris while simulating the mass balance. In fact, there has been no research in western Canada to understand the process of glacier ice melt under the layer of debris. This research will take the influence of debris into account when simulating glacier mass balance. Research schedule Date June-Aug 2008 Major tasks Reconnaissance visit to Dome glacier, field measurements at Dome glacier Sep–Dec 2008 Analysis of collected data, presentation of results at the WC2N meeting, acquisition of relevant data from different sources, continue working with thesis proposal preparation, defend thesis proposal 35 Jan-April 2009 Continue collecting data from various sources, analyze and manage collected data set, prepare dataset for DS experiments May-Aug 2009 Begin climate DS work, model calibration using combination of predictors and predictands variables, field work at Dome glacier and Tiedemann glacier Sep-Dec 2009 Continue working with DS, develop ensembles of DS fields, validation of downscaled fields, manage DS results Jan-April 2010 Calibration of GMB model, simulation of historical mass balance, validate simulated results, simulation of future glacier mass balance May-Aug 2010 Preparing research papers, work towards paper publication Sep 2010-??? Thesis writing, thesis defense Funding This is a part of the research under Canadian Cryospheric Network (WC2N), a network of research institutions which is studying the past and future impact of climate change on glaciers of western Canada and its implication to freshwater ecosystems. The network is funded by Canadian Foundation of Climate and Atmospheric Sciences (CFCAS). The research is being conducted under the supervision of Prof. Peter L. Jackson of University of Northern British Columbia. Tentative Chapter Titles 1. Introduction 1.1 Context 1.2 Aims and objectives 1.3 Rationale 1.4 Thesis outline 2. Literature review 2.1 Climate change and glacier fluctuation 36 2.1.1 Anthropogenic climate change 2.1.2 Glaciers and climate change 2.1.2 Worldwide glacier fluctuation 2.1.3 Glacier fluctuation in western Canada 2.2 Implication of glacier recession 2.2.1 Environmental aspect 2.2.2 Economic aspect 2.2.3 Human dimensions 2.3 Climate change impact assessment 2.3.1 Climate downscaling 2.3.2 Glacier mass balance modeling 3. Methodology 3.1 Statistical climate downscaling 3.1.1 Predictor 3.1.2 Predictand 3.1.3 Statistical DownScaling Model (SDSM) 3.1.4 Linear downscaling method 3.1.5 Local scaling method 3.2 Glacier melt under the debris layer 3.2.1 Direct measurement of melt rate 3.2.2 Measurement of debris properties 3.2.3 Coefficient of melt 3.2.4 Modeling melt under debris layer 3.3 Glacier mass balance modeling 3.3.1 Temperature index model 3.3.2 Model calibration 3.3.3 Projection of future mass balance 37 4. Results and discussions 4.1 Climate downscaling 4.2 Glacier melt under the debris layer 4.3 Mass balance simulation 5. Conclusions and recommendations References References BC Ministry of Water Land and Air Protection, 2002. Indicators of Climate Change for British Columbia 2002. Barrows, E.B., B. Maxwell and P. Gachon (Eds). 2004. Climate variability and Change in Canada: Past, Present and Future, ACSD Science Assessment Series No. 2, Meteorological Service of Canada, Environment Canada, Toronto, Ontario, 114p. Bhatt, U.S., J. Zhang, C. S. Lingle, L. M. Phillips and W. V. Tangborn. 2007. Examining glacier massbalances with hierarchical modeling approach. Computing in Science and Engineering, Co-published by the IEEE CS and the AIP © 2007 IEEE. Braithwaite, R. J. and O.B. Olesen. 1985. Ice ablation in West Greenland in relation to air temperature and global radiation. Zeitschrift für Gletscherkunde und Glazialgeologie. 20 (1984), 155-168. Daly, C., W. P. Gibson, G. H. Taylor, G. L. Johnson and , P. Pasteris. 2002. A knowledge-based approach to the statistical mapping of climate. Climate Research, 22, 99–113. Demuth, M.N. and R. Keller. 2006. An assessment of the mass-balance of Peyto Glacier (1966–1995) and its relations to recent and past-century climatic variability. In Demuth, M.N., D.S. Munro and G.J. Young, eds. Peyto Glacier: one century of science. Saskatoon, National Hydrological Research Institute, 83–132. (NHRI Scientific Report 8.) de Beer, C., and M.J. Sharp. 2007. Recent glacier retreat within the southern Canadian Cordillera, Ann. Glaciol., 46, 215-221. de Woul, M. and R. Hock. 2005. Static mass-balance sensitivity of Arctic glaciers and ice caps using a degree-day approach. Ann. Glaciol., 42, 217–224. Fountain, A.G. and W. Tangborn. 1985. The effects of glaciers on streamflow variations. Water Resour. Res., 21(4), 579–586. 38 Hock, R. 2003. Temperature index melt modelling in mountain areas. J. Hydrol., 282 (1–4), 104–115. Hock, R. and B. Holmgren. 2005. A distributed surface energy balance model for complex topography and its application to Storglaciären, Sweden. J. Glaciol., 51(172), 25–36. Hock, R., V. Radic and M. D. Woul. 2007. Climate sensitivity of to Storglaciären, Sweden: an intercomparison of mass-balance models using ERA-40 re-analysis and regional climate model data. Ann. Glaciol., 46, 342–348. Hopkinson, C. and G.J. Young. 1998. The effect of glacier wastage on the flow of the Bow River at Banff, Alberta, 1951–1993. Hydrol. Process., 12(10–11), 1745–1762. Huth, R. 1999. Statistical downscaling in central Europe: evaluation of methods and potential predictors. Climate Researh,, 13, 99-101. IPCC, 2007: Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437–471. Katz, R.W., M.B. Parlange and P. Naveau. 2002. Statistics of extremes in hydrology. Advances in Water Resources, 25, 1287-1304. Leith, R.M.M. and P. H. Whitfield. 1998. Evidence of climate change effects on the hydrology of streams in south-central B. C. Canadian Water Resources Journal, 23(3), 219-230. Lines, G.S, M. Pancura and C. Lander. 2006. Building climate change scenarios of temperature and precipitation in Atlantic Canada using the statistical downscaling model (SDSM). Science Report Series 2005-9, Meteorological Service of Canada, Atlantic Region. Moore, R.D. and M.N. Demuth. 2001. Mass-balance and streamflow variability at Place Glacier, Canada, in relation to recent climate fluctuations. Hydrol. Process., 15(18), 3472–3486. Oerlemans, J. and 10 others. 1998. Modelling the response of glaciers to climate warming. Climate Dyn., 14(4), 267–274. Oerlemans, J. and 8 others. 2005. Estimating the contribution from Arctic glaciers to sea-level change in the next 100 years. Ann. Glaciol., 42, 230–236. Ommanney, C.S. 2002. History of glacier investigation in Canada. In Williams, R.S. Jr and J.G. Ferrigno, eds. Satellite Image Atlas of Glaciers of the World, U.S. Geological Survey Professional Paper 1386–J–1, Pub. United States Government Printing Office, Washington, 27-82. Pfizenmayer, A and H. Von Storch. 2001. Anthropogenic climate change shown by local wave conditions in the North Sea. Climate Research, 19, 15-23. Radic´, V. and R. Hock. 2006. Modeling future glacier mass-balance and volume changes using ERA-40 reanalysis and climate models: sensitivity study at Storglacia¨ren, Sweden. J. Geophys. Res., 111(F3), F03003. (10.1029/2005JF000440.) Rasmussen, L.A. and H. Conway. 2004. Climate and Glacier Variability in Western North America. J. Climate., 17, 1804-1815. 39 Satathe, E.P Jr, P.W. Mote and M.W Wiley. 2007. Review of scenario selection and downscaling methods for the assessment of climate change impacts on hydrology in the United States pacific northwest. Int. J. Climatol.,, 27, 1611-1621. Salathé, E. P. 2005, Downscaling simulations of future global climate with application to hydrologic modelling, Int. J. Climatol., 25, 419– 436. Stahl, K. and D. Moore. 2006. Influence of watershed glacier coverage on summer streamflow in British Columbia, Canada. Water Resour. Res., 42(6), W06201. (10.1029/2006WR005022.) Stanley, A.D. 1975. Snow and Ice-Symposium-Neiges et Glaces (Proceedings of the Moscow Symposium, August 1971; Actes du Colloque de Moscou, août 1971): IAHS-AISH Publ. No. 104, 1975. Tangborn, W. 1997. Using low-altitude meteorological observations to calculate the mass-balance of Alaska’s Columbia Glacier and relate it to calving and speed. Byrd Polar Res. Cen. Rep. 15, 141–161. Tangborn, W.V. 1999. A mass-balance model that uses low-altitude meteorological observations and the area-altitude of a glacier. Geografiska Annaler, December 1999, Proceedings of the Tarafala Mass-balance Workshop. Tangborn, W.V. and B. Rana. 2000. Mass-balance and Runoff of the Partially Debris-Covered Langtang Glacier, Nepal. Proc. Debris- Covered Glaciers Workshop, Int’l Assoc. Hydrological Sciences, 2000, pp. 99– 108. Von Storch, H. 1999. On the use of “inflation” in statistical downscaling. Journal of Climate, 12, 3505-3506. Von Storch, H. 1995. Inconsistencies at the interface of climate impact studies and global climate research. Meteorol. Zeitschrift, 4, 72-80. Wang, T., A. Hamann, D. Spittlehouse, and S.N. Aitken. 2006. Development of scale-free climate data for western Canada for use in resource management. International Journal of Climatology, 26(3), 383-397. Widmann, M., C.S. Bretherton and E.P. Satathe Jr. 2003. Statistical precipitation downscaling over the northwestern United States using numerically simulated precipitation as a predictor Journal of Climate, 16, 799-816 Wilby, R.L., C.W. Dawson and E.M. Barrow. 2001. SDSM User Manual- A Decision Support Tool for the Assessment of Regional Climate Change Impacts. http://wwwstaff. lboro.ac.uk/~cocwd/sdsm.html Wilby, R.L., C.W. Dawson and E.M. Barrow.2002. SDSM - a decision support tool for the assessment of regional climate change impacts. Environmental and Modeling Software, 17, 145-157. Zhang, J., U. S. Bhatt, W. V. Tangborn, and C. S. Lingle. 2007a. Climate downscaling for estimating glacier mass-balances in northwestern North America: Validation with a USGS benchmark glacier, Geophys. Res. Lett., 34, L21505, doi:10.1029/ 2007GL031139. Zhang, J., U. S. Bhatt, W. V. Tangborn, and C. S. Lingle. 2007b. Response of glaciers in northwest North America to future climate change: an atmosphere/glacier hierarchical modeling approach. Ann. Glaciol., 46, 283–290. 40