Introduction

HESS

Hydrology and Earth System Sciences

HESS

Hydrol. Earth Syst. Sci.

1607-7938

Copernicus Publications

Göttingen, Germany

10.5194/hess-21-2163-2017

Inter-comparison of daily precipitation products for large-scale hydro-climatic applications over Canada

Wong

Jefferson S.

jefferson.wong@usask.ca

https://orcid.org/0000-0002-6793-5017

Razavi

Saman

Bonsal

Barrie R.

Wheater

Howard S.

Asong

Zilefac E.

https://orcid.org/0000-0001-7086-6764

1Global Institute for Water Security and School of Environment and Sustainability, University of Saskatchewan, 11 Innovation Blvd, Saskatoon, SK, S7N 3H5, Canada 2Environment and Climate Change Canada, 11 Innovation Blvd, Saskatoon, SK, S7N 3H5, Canada

Jefferson S. Wong (jefferson.wong@usask.ca)

20April2017

21 4 21632185 28September2016 13October2016 28February2017 16March2017

This work is licensed under a Creative Commons Attribution 3.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by/3.0/

This article is available from https://hess.copernicus.org/articles/21/2163/2017/hess-21-2163-2017.html

The full text article is available as a PDF file from https://hess.copernicus.org/articles/21/2163/2017/hess-21-2163-2017.pdf

A number of global and regional gridded climate products based on multiple data sources are available that can potentially provide reliable estimates of precipitation for climate and hydrological studies. However, research into the consistency of these products for various regions has been limited and in many cases non-existent. This study inter-compares several gridded precipitation products over 15 terrestrial ecozones in Canada for different seasons. The spatial and temporal variability of the errors (relative to station observations) was quantified over the period of 1979 to 2012 at a 0.5∘ and daily spatio-temporal resolution. These datasets were assessed in their ability to represent the daily variability of precipitation amounts by four performance measures: percentage of bias, root mean square error, correlation coefficient, and standard deviation ratio. Results showed that most of the datasets were relatively skilful in central Canada. However, they tended to overestimate precipitation amounts in the west and underestimate in the north and east, with the underestimation being particularly dominant in northern Canada (above 60∘ N). The global product by WATCH Forcing Data ERA-Interim (WFDEI) augmented by Global Precipitation Climatology Centre (GPCC) data (WFDEI [GPCC]) performed best with respect to different metrics. The Canadian Precipitation Analysis (CaPA) product performed comparably with WFDEI [GPCC]; however, it only provides data starting in 2002. All the datasets performed best in summer, followed by autumn, spring, and winter in order of decreasing quality. Findings from this study can provide guidance to potential users regarding the performance of different precipitation products for a range of geographical regions and time periods.

Introduction

The availability of accurate data, especially precipitation, is essential for understanding the climate system and hydrological processes since it is a vital element of the water and energy cycles and a key forcing variable for driving hydrological models. Reliable precipitation measurements provide valuable information for meteorologists, climatologists, hydrologists, and other decision makers in many applications, including climate and/or land use change studies (e.g. Cuo et al., 2011; Huisman et al., 2009; Dore, 2005), agricultural and environmental research (e.g. Zhang et al., 2012; Hively et al., 2006), natural hazards (e.g. Taubenböck et al., 2011; Kay et al., 2009; Blenkinsop and Fowler, 2007), and hydrological and water resources planning (e.g. Middelkoop et al., 2001; Hong et al., 2010). With respect to land-surface hydrology, the increasing sophistication of distributed hydrological modelling has urged the requirement of better and more reliable gridded precipitation estimates at a minimum, daily temporal resolution. Before incorporating precipitation measurements, quantifying their uncertainty becomes an essential prerequisite for hydrological applications and is increasingly critical for potential users, who are left without guidance and/or confidence in the myriad of products for their specific hydrological problems over different geographical regions. This study attempts to address this issue by comparing and examining the error characteristics of different types of gridded precipitation products and assessing how these products perform geographically and temporally over Canada.

Precipitation measurements and their limitations

With technological and scientific advancements over the past 3 decades, tremendous progress has been made in the various methods of precipitation measurement, each one with its own strengths and limitations. Rain gauges provide the direct physical readings with relatively accurate measurements at specific points. However, such measurements are subject to various errors arising from wind effects (Nešpor et al., 2000; Ciach, 2003), evaporation (Strangeways, 2004; Mekis and Hogg, 1999), undercatch (Yang et al., 1998; Adam and Lettenmaier, 2003; Mekis and Hogg, 1999), and instrumental problems. Moreover, rain-gauge measurements are often spatially interpolated, which may not capture the true spatial variability of precipitation fields due to sparse gauge networks. Ground-based radar measurements can estimate precipitation over a relatively large area (radius of 200 to 300 km) but are prone to inaccuracies as a result of beam spreading, curvature of the Earth, and terrain blocking (Dinku et al., 2002; Young et al., 1999), and errors in the rain rate–reflectivity relationship, range effects, and clutter (Jameson and Kostinski, 2002; Villarini and Krajewski, 2010). Development of satellite-based precipitation estimates, such as the Global Precipitation Measurement (GPM) mission (Hou et al., 2014), has provided excellent spatial coverage but also contain inaccuracies resulting primarily from temporal sampling errors, instrumental errors, and algorithm errors (Nijssen and Lettenmaier, 2004; Gebremichael et al., 2005). Recognizing the limitations in the various precipitation observation methods, a number of attempts to combine information from multiple sources have been undertaken (Xie and Arkin, 1996; Maggioni et al., 2014; Shen et al., 2010). Reanalysis data provide an alternative source of precipitation estimates by assimilating all available data (rain-gauge stations, aircraft, satellite, etc.) into a background forecast physical model. However, accuracies in reanalysis precipitation are dependent on the specific analysis-forecast systems and the choice of physical parameterizations (Betts et al., 2006). Numerical climate models including Atmosphere–Ocean General Circulation Models (AOGCMs) and Regional Climate Models (RCMs) offer another potential source of precipitation estimates, as well as future precipitation simulations. Precipitation estimates from climate models, however, remain relatively coarse in resolution and often produce systematic bias due to imperfect model conceptualization, discretization, and spatial averaging within grid cells (Teutschbein and Seibert, 2010; Xu et al., 2005).

Scope and objectives

Numerous previous evaluation efforts among the precipitation products have been limited into three groups of inter-comparison of (1) satellite-derived products (e.g. Adler et al., 2001; Xie and Arkin, 1995; Turk et al., 2008), (2) reanalysis data (e.g. Janowiak et al., 1998; Bosilovich et al., 2008; Betts et al., 2006; Bukovsky and Karoly, 2007), and (3) climate model simulations (e.g. Covey et al., 2003; Christensen et al., 2007; Mearns et al., 2006, 2012). Despite the aforementioned efforts, few studies have conducted a detailed inter-comparison among different types of precipitation products. Gottschalck et al. (2005) compared seasonal total precipitation of several satellite-derived, rain-gauge-based, and model-simulated datasets over contiguous United States and showed the spatial root mean square error of seasonal total precipitation and mean correlation of daily precipitation between each product and the impacts of these errors on land-surface modelling. Additionally, Ebert et al. (2007) examined 12 satellite-derived precipitation products and four numerical weather prediction models over the United States, Australia, and northwestern Europe and found that satellite-derived estimates performed best in summer and model-induced ones were best in winter. However, a number of questions regarding the reliability of the precipitation products remained in doubt, including to what extent do the users have the knowledge about the error information associated with all these different types of precipitation products, how do the error distribution of precipitation products vary by location and season, and which product(s) should the users have more confidence for their regions of interest. Answering these questions is therefore a crucial first step in quantifying the spatial and temporal variability of the precipitation products so as to better understand their reliability as forcing inputs in hydrological modelling and other related studies.

Given the emergence of various products derived from different methods and sources (Tapiador et al., 2012), accuracy comparison studies of precipitation products have been reported over several regions; examples include the globe (e.g. Gebregiorgis and Hossain, 2015; Adler et al., 2001; Tian and Peters-Lidard, 2010), Europe (e.g. Frei et al., 2006; Chen et al., 2006; Kidd et al., 2012), Africa (e.g. Dinku et al., 2008; Asadullah et al., 2008), North America (e.g. Tian et al., 2009; West et al., 2007), South America (e.g. Vila et al., 2009), and China (e.g. Shen et al., 2010; Wetterhall et al., 2006). However, less attention has been paid to high-latitude regions such as Canada, where a considerable proportion of precipitation is in the form of snow (Behrangi et al., 2016). In many regions of Canada, precipitation-gauge stations are sparsely distributed and the information required for hydrological modelling may not be available at the site of interest. This is especially true in northern areas (north of 60∘ N) and over mountainous regions where precipitation-gauge stations are usually 500 to 700 km apart or at low elevations (Wang and Lin, 2015). Meanwhile, the decline and closure of manual observing precipitation-gauge stations further reduced the spatial coverage and availability of long-term precipitation measurements (Metcalfe et al., 1997; Mekis and Hogg, 1999; Rapaic et al., 2015). Of additional concern, the observations for solid precipitation (snow, snow pellets, ice pellets, and ice crystals) and precipitation phase (liquid or solid) changes make accurate measurement of precipitation more difficult and challenging, and the measurement errors have been found to range from 20 to 50 % for automated systems (Rasmussen et al., 2012). The Meteorological Service of Canada has implemented a network of 31 radars (radar coverage at full range of 256 km) along southern Canada (see Fortin et al., 2015b; Fig. 1 for spatial distribution). Yet, the shortcomings of using the radar data are 2-fold: (1) many areas of the country (north of 60∘ N) are not covered by this network and (2) the implementation of the network began in 1997 and thus did not have sufficient lengths of data for any long-term hydro-climatic studies. The availability, coverage, and quality of precipitation-gauge measurements are thus obstacles to effective hydrological modelling and water management in Canada. However, the availability of several global and regional gridded precipitation products, which provide complete coverage of the whole country at applicable time- and spatial scales, may provide a viable alternative for regional- to national-scale hydrological applications in Canada.

In total, 15 terrestrial ecozones of Canada with numerical codes indicating regions from 1 (Arctic Cordillera) to 15 (Hudson Plain). Big (a total of 145) and small (a total of 137) white dots are the extracted precipitation-gauge stations from the Canadian adjusted and homogenized precipitation datasets of Mekis and Vincent (2011) for the period of 1979 to 2012 and 2002 to 2012 respectively. Black dots are major cities in Canada.

Given the aforementioned, this study aims to (1) inter-compare various daily gridded precipitation products against the best available precipitation-gauge observations; and (2) characterize the error distributions of different types of precipitation products over time and different geographical regions in Canada. Such inter-comparison will in turn help assess the performance of the precipitation products over specific climatic/hydrological regions.

The rest of this paper is organized as follows: a brief description of the study area and precipitation data is provided in Sects. 2 and 3. The methodology for evaluating precipitation products against the precipitation-gauge station observations is described in Sect. 4. Results and discussion are provided in Sects. 5 and 6 respectively, with a summary and conclusion following in Sect. 7.

Study area

Canada, which covers a land area of 9.9 million km2, extends from 42 to 83∘ N latitude and spans between 141 to 52∘ W longitude. With substantial variations over its land mass, the country can be divided into many regions according to aspects such as climate, topography, vegetation, soil, geology, and land use. The National Ecological Framework for Canada classified ecologically distinct areas with four hierarchical levels of generalization (15 ecozones, 53 ecoprovinces, 194 ecoregions, and 1021 ecodistricts from the broadest to the smallest) (Ecological Stratification Working Group, 1996; Marshall et al., 1999). Similarly, the Standard Drainage Area Classification was developed to delineate hydrographic areas to cover all the land and interior freshwater lakes of the country with three levels of classification (11 major drainage areas, 164 sub-drainage areas, and 974 sub-sub-drainage areas) (Brooks et al., 2002; Pearse et al., 1985). The precipitation comparisons in this study incorporated both the ecological and hydrological delineations. This involved classifying the Canadian land mass into 15 ecozones for the main study (Fig. 1) and 14 major drainage areas (the Arctic major drainage area was further divided into Arctic and Mackenzie, whereas the St. Lawrence major drainage area was further split into St. Lawrence, Great Lakes, and Newfoundland). Results are based on the ecozone classification, while those based on drainage areas are reported in the Supplement.

Precipitation data Precipitation-gauge station observations

In Canada, climate data collection is coordinated by the federal government, which is made available by the National Climate Data Archive of Environment and Climate Change Canada (NCDA). These data provide the basis for all available quality controlled climate observations. There are a total of 1499 precipitation-gauge stations (as of 2012) across Canada. However, given the frequent addition and subtraction of climate stations, these numbers have greatly varied through time with peak reporting in the 1970s followed by a general decline to the present (see Hutchinson et al., 2009; Figs. 1 and 2 for details). Furthermore, the existing precipitation observations are often subject to various errors, with gauge undercatch being of significant concern (Mekis and Hogg, 1999). To account for various measurement issues, Mekis and Hogg (1999) first produced the Adjusted and Homogenized Canadian Climate Data (AHCCD) including adjusted daily rainfall and snowfall values and Mekis and Vincent (2011) then updated the data for a subset of 464 stations over Canada. The data extend back to 1895 for a few long-term stations and run through 2014. As a result of adjustments, total rainfall amounts were of the order of 5 to 10 % higher in southern Canada and more than 20 % in the Canadian Arctic when compared to the original observations. Adjustments to snowfall were even larger and varied throughout the country. These adjusted values are widely considered as better estimates of actual precipitation and therefore have been used in numerous analyses (e.g. Nalley et al., 2012; Shook and Pomeroy, 2012; Wan et al., 2013; Asong et al., 2015). Given the lack of an adjusted daily gridded precipitation product for Canada, the AHCCD station precipitation is considered to be the best available data for Canada and thus is used as the benchmark for all gridded precipitation product comparisons.

Precipitation products used in this study.

Dataset Full name Type Spatial Temporal Duration Coverage Reference resolution resolution ANUSPLIN Australian National University Spline Station-based interpolated 300 arcsec 24 h 1950–2013 Canada Hutchinson et al. (2009) (∼ 0.0833∘/∼ 10 km) CaPA Canadian Precipitation Analysis Station-based multiple-source 10 km 6 h 2002–2014 North America Mahfouf et al. (2007) (∼ 0.0833∘) Princeton Global dataset at the Princeton University Reanalysis-based multiple-source 0.5∘ 3 h 1901–2012 Global Sheffield et al. (2006) (∼ 50 km) WFDEI [CRU] Water and Global Change Forcing Data methodology Reanalysis-based multiple-source 0.5∘ 3 h 1979–2012 Global Weedon et al. (2014) applied to ERA-Interim [Climate Research Unit] (∼ 50 km) WFDEI [GPCC] Water and Global Change Forcing Data methodology Reanalysis-based multiple-source 0.5∘ 3 h 1979–2012 Global Weedon et al. (2014) applied to ERA-Interim [Global Precipitation (∼ 50 km) Climatology Centre] NARR North American Regional Reanalysis Reanalysis-based multiple-source 32 km 3 h 1979–2015 North America Mesinger et al. (2006) (0.3∘) PCIC Pacific Climate Impacts Consortium Station-driven global circulation model (GCM) 300 arcsec 24 h Historical: 1950–2005 Canada Pacific Climate Impacts Consortium; (∼ 0.0833∘/∼ 10 km) Projected: 2006–2100 University of Victoria (2014) NA-CORDEX North America COordinated Regional climate GCM-driven RCM 0.22∘ 3 h Historical: 1950–2005 North America Giorgi et al. (2009) Downscaling EXperiment (25 km) Projected: 2006–2100

Global circulation models (GCMs) chosen in the Pacific Climate Impacts Consortium (PCIC) dataset.

PCIC Full name Country Statistical downscaling method GFDL-ESM2G_BCCAQ Geophysical Fluid Dynamics Laboratory USA Bias correction constructed analogues with quantile mapping reordering GFDL-ESM2G_BCSD Earth System Model 2G Bias correction spatial disaggregation HadGEM2-ES_BCCAQ Hadley Global Environmental Model 2 UK Bias correction constructed analogues with quantile mapping reordering HadGEM2-ES_BCSD – Earth System Bias correction spatial disaggregation CanESM2_BCCAQ Second generation Canadian Earth Canada Bias correction constructed analogues with quantile mapping reordering CanESM2_BCSD System Model Bias correction spatial disaggregation MPI-ESM-LR_BCCAQ Max-Planck-Institute Earth System Model Germany Bias correction constructed analogues with quantile mapping reordering MPI-ESM-LR_BCSD running on low resolution Bias correction spatial disaggregation

GCMs–RCMs chosen in the North America COordinated Regional climate Downscaling EXperiment (NA-CORDEX) dataset.

NA-CORDEX Full name Global Circulation Model (GCM) Regional Climate Model (RCM) CanESM2 – CanRCM4 Second generation Canadian Earth System Model Fourth generation Canadian Regional Climate Model CanESM2 – CRCM5_UQAM Fifth generation Canadian Regional Climate Model MPI-ESM-LR – CRCM5_UQAM Max-Planck-Institute Earth System Model running Fifth generation Canadian Regional Climate Model MPI-ESM-LR – RegCM4 on low resolution Fourth generation Regional Climate Model

Gridded precipitation products

In total, 7 precipitation datasets were chosen for assessment based on the following criteria: (1) a complete coverage of Canada, (2) minimum of daily temporal and 0.5∘ (∼ 50 km) spatial resolutions, (3) sufficient length of data (> 30 years) for long-term study including recent years up to 2012, and (4) representing a range of sources/methodologies (e.g. station based, remote sensing, model, blended products). Table 1 summarizes these datasets, including their full names and original spatial and temporal resolutions for the versions used.

Station-based product – ANUSPLIN

Hutchinson et al. (2009) used the Australian National University Spline (ANUSPLIN) model to develop a dataset of daily precipitation, and daily minimum and maximum air temperature over Canada at a spatial resolution of 300 arcsec (0.0833∘ or ∼ 10 km) for the period of 1961 to 2003. All available NCDA stations (that ranged from 2000 to 3000 for any given year during this period) were used an input to the gridding procedure. To retain maximum spatial coverage, the smaller number of stations in AHCCD were not incorporated (i.e. only unadjusted archive values were used). Interpolation procedures included incorporation of tri-variate thin-plate smoothing splines using spatially continuous functions of latitude, longitude, and elevation. Hopkinson et al. (2011) subsequently extended this original dataset to the period 1950 to 2011. The Canadian ANUSPLIN has now further been updated to 2013 and has recently been used as the basis of “observed” data for evaluating different climate datasets (e.g. Eum et al., 2012) and for assessing the effects of different climate products in hydro-climatological applications (e.g. Eum et al., 2014; Bonsal et al., 2013; Shrestha et al., 2012a).

The percentage of reliability, calculated by Eq. (1), of each precipitation dataset in four seasons for the period of 1979 to 2012 (a), 2002 to 2012 (b), and 1979 to 2005 (c) across Canada. The higher the percentage, the more reliable the precipitation dataset. Different colours represent different precipitation products, with magenta representing the whole PCIC datasets and cyan representing the whole NA-CORDEX datasets. The full names of the precipitation products are provided in Tables 1, 2, and 3.

Station-based multiple-source product – CaPA

In November 2003, the Canadian Precipitation Analysis (CaPA) was developed to produce a dataset of 6-hourly precipitation accumulation over North America in real time at a spatial resolution of 15 km (from 2002 onwards) (Mahfouf et al., 2007). Data were generated using an optimum interpolation technique (Daley, 1993), which required a specification of error statistics between observations and a background field (e.g. Bhargava and Danard, 1994; Garand and Grassotti, 1995). For Canada, the short-term precipitation forecasts from the Canadian Meteorological Centre (CMC) regional Global Environmental Multiscale (GEM) model (Cote et al., 1998a, b), in its regional configuration, were used as the background field with the rain-gauge measurements from NCDA as the observations to generate an analysis error at every grid point. CaPA become operational at the CMC in April 2011, with updates in the statistical interpolation method (Lespinas et al., 2015) and increase of spatial resolution to 10 km in October 2012. The assimilation of quantitative precipitation estimates from the Canadian weather radar network is also used as an additional source of observations in November 2014 (Fortin et al., 2015b). Since November 2016, data from 33 US radars near the border are also assimilated, in addition to that of the 31 Canadian radars. With its continuous improvement and different configurations, CaPA has been employed in Canada for various environmental prediction applications (e.g. Eum et al., 2014; Fortin et al., 2015a; Pietroniro et al., 2007; Carrera et al., 2015). However, the study period of these applications only start in 2002.

Reanalysis-based multiple-source products – Princeton, WFDEI, and NARR Princeton

The Terrestrial Hydrology Research Group at the Princeton University initially developed a dataset of 3-hourly near-surface meteorology with global coverage at 1.0∘ spatial resolution (∼ 120 km) from 1948 to 2000 for driving land-surface models and other terrestrial systems (Sheffield et al., 2006). This dataset (called hereafter “Princeton”) was constructed based on the National Centers for Environmental Prediction-National Center for Atmospheric Research (NCEP-NCAR) reanalysis (2.0∘ and 6-hourly) (Kalnay et al., 1996; Kistler et al., 2001), combined with a suite of global observation-based data including the Climatic Research Unit (CRU) monthly climate variables (New et al., 1999, 2000), the Global Precipitation Climatology Project (GPCP) daily precipitation (Huffman et al., 2001), the Tropical Rainfall Measuring Mission 3-hourly precipitation (Huffman et al., 2002), and the NASA Langley Research Center monthly surface radiation budget (Gupta et al., 1999). With the inclusion of additional temperature and precipitation data (e.g. Willmott et al., 2001), Princeton has been updated and is currently available with two versions. This study used the 1901–2012 experimental version at 0.5∘ at daily time steps. Studies employing Princeton to examine different hydrological aspects have been carried out over different parts of Canada (e.g. Wang et al., 2013, 2014; Kang et al., 2014).

WFDEI

To simulate the terrestrial water cycle using different land-surface models and general hydrological models, the European Union Water and Global Change (WATCH) Forcing Data (WFD) were created to provide datasets of sub-daily (3- and 6-hourly) and daily meteorological data with global coverage at 0.5∘ spatial resolution (∼ 50 km) from 1901 to 2001 (Weedon et al., 2011). Similar to Princeton, the WFD were derived from the 40-year European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-40) (1.0∘ and 3-hourly) (Uppala et al., 2005) and combined with the CRU monthly variables and the Global Precipitation Climatology Centre (GPCC) monthly data (Rudolf and Schneider, 2005; Schneider et al., 2008; Fuchs, 2009). The WATCH Forcing Data methodology applied to the ERA-Interim (WFDEI) dataset has further been developed covering the period of 1979 to 2012 (Weedon et al., 2014). The WFDEI used the same methodology as the WFD, but was based on the ERA-Interim (Dee et al., 2011) with higher spatial resolution (0.7∘). As for the WFD, the WFDEI had two sets of rainfall and snowfall data generated by using either CRU or GPCC precipitation totals. Both sets of data were used in this study (hereafter known as WFDEI [CRU] and WFDEI [GPCC] respectively). To date, specific studies using the WFDEI related to Canada have been limited to the investigation of permafrost changes in the Arctic regions (e.g. Chadburn et al., 2015; Park et al., 2015, 2016).

NARR

With the aim of evaluating spatial and temporal water availability in the atmosphere, the North American Regional Reanalysis (NARR) was developed to provide datasets of 3-hourly meteorological data for the North America domain at a spatial resolution of 32 km (∼ 0.3∘) covering the period of 1979 to 2003 as the retrospective system and is being continued in near real time (currently up to 2015) as the Regional Climate Data Assimilation System (Mesinger et al., 2006). The components in generating NARR included the NCEP-DOE reanalysis (Kanamitsu et al., 2002), the NCEP regional Eta Model (Mesinger et al., 1988; Black, 1988), and the Noah land-surface model (Mitchell et al., 2004; Ek et al., 2003), as well as the use of numerous additional data sources (see Mesinger et al., 2006; Table 2). For hydrological modelling in Canada, Choi et al. (2009) found that NARR provided reliable climate inputs for northern Manitoba while Woo and Thorne (2006) concluded that NARR had a cold bias resulting in later snowmelt peaks in subarctic Canada. In addition, Eum et al. (2012) identified a structural break point in the NARR dataset beginning in January 2004 over the Athabasca River basin due to the assimilation of station observations over Canada being discontinued in 2003.

GCM statistically downscaled products – PCIC

The Pacific Climate Impacts Consortium (PCIC), which is a regional climate service centre at the University of Victoria, British Columbia, Canada, has offered datasets of statistically downscaled daily precipitation and daily minimum and maximum air temperature under three different Representative Concentration Pathway (RCP) scenarios (RCP 2.6, 4.5, and 8.5) (Meinshausen et al., 2011) over Canada at a spatial resolution of 300 arcsec (0.833∘ or ∼ 10 km) for the historical and projected period of 1950 to 2100 (Pacific Climate Impacts Consortium; University of Victoria, 2014). These downscaled datasets were a composite of 12 global circulation model (GCM) projections from the Coupled Model Inter-comparison Project Phase 5 (CMIP5) (Taylor et al., 2012) and the ANUSPLIN dataset. The historical 1950 to 2005 period of the ANUSPLIN was used for bias-correction and downscaling of the GCMs. Two different methods were used to downscale to a finer resolution (Werner and Cannon, 2016). These included bias correction spatial disaggregation (BCSD) (Wood et al., 2004) following Maurer and Hidalgo (2008) and bias correction constructed analogues (BCCA) with quantile mapping reordering (BCCAQ), which was a post-processed version of BCCA (Maurer et al., 2010). The ensemble of the PCIC dataset has currently been used in studying the hydrological impacts of climate change on river basins mainly in British Columbia (e.g. Shrestha et al., 2011, 2012b; Schnorbus et al., 2014) and Alberta (e.g. Kienzle et al., 2012; Forbes et al., 2011) in Canada. In this study, only four GCMs with two respective statistical downscaling methods were chosen for comparison (see Table 2 for details). The choice of the four GCMs was to match those available in the NA-CORDEX dataset (see next section for details).

GCM-driven RCM dynamically downscaled products – NA-CORDEX

Sponsored by the World Climate Research Programme, the COordinated Regional climate Downscaling EXperiment (CORDEX) over the North America domain (NA-CORDEX) provides dynamically downscaled datasets of 3-hourly or daily meteorological data over most of North America (below 80∘ N) at spatial resolutions of 0.22 and 0.44∘ (∼ 25 and ∼ 50 km) under RCP 4.5 and 8.5 for the historical (1950–2005) and future (2006–2100) periods (Giorgi et al., 2009). Drawing from the strengths of the North American Regional Climate Change Assessment Program (Mearns et al., 2012), a matrix of six GCMs from the CMIP5 driving six different RCMs was selected to compare and characterize the uncertainties of RCMs and thus provided climate scenarios for further impact and adaption studies. Current studies using NA-CORDEX datasets were mainly focused on evaluating the model performance of different GCM-driven RCM simulations over North America (e.g. Lucas-Picher et al., 2013; Martynov et al., 2013; Separovic et al., 2013). In this study, two GCMs and three RCMs were chosen for comparison due to the availability of the NA-CORDEX dataset (see Table 3 for details).

Methodology Pre-processing

Due to the different spatial and temporal resolutions of the various precipitation products, the first step was to re-grid each onto a common 0.5∘ × 0.5∘ resolution to match the lowest-resolution dataset. It was acknowledged that re-gridding products onto a common spatial resolution might introduce more errors or uncertainties and the number of interpolation steps should be minimized. However, the main focus of this study was to inter-compare various gridded precipitation products using precipitation-gauge station data as a reference/benchmark but not to assess the individual accuracy of each product against the reference dataset. Therefore, upscaling to a common resolution provided a direct and more consistent inter-comparison. Such methodology was consistent with similar studies in the literature (e.g. Janowiak et al., 1998; Rauscher et al., 2010; Kimoto et al., 2005). All data were accumulated to daily timescale for comparison. Two common time spans were selected since CaPA covered a shorter time frame compared to the rest of the products: (1) long-term comparison from January 1979 to December 2012 with the exclusion of CaPA (from January 1979 to December 2005 for PCIC, and NA-CORDEX as the historical period of the datasets ends in 2005), and (2) short-term comparison from January 2002 to December 2012 when CaPA data are available. Daily values were summed over the four standard seasons (spring: March-April-May – MAM; summer: June-July-August – JJA; autumn: September-October-November – SON; and winter: December-January-February – DJF) to inter-compare the precipitation products at a seasonal scale.

To identify the most consistent gridded dataset corresponding to different seasons and regions, comparisons of each dataset with direct precipitation-gauge station data from the aforementioned AHCCD were carried out. For the period of 1979 to 2012, only 169 of the original 464 stations across Canada were available. This drastic drop was due to 271 stations ending before or after early 2000s and 23 not having a complete year for 2012. Subsequently, any of the 169 stations where the percentage of missing values exceeded 10 % during the study period were also eliminated. This resulted in 145 and 137 stations across Canada for long-term and short-term comparison respectively (see Fig. 1 for locations). Note that most of the stations are located in southern Canada with only 15 stations above 60∘ N.

Gridded-based precipitation estimates at the coordinates of the precipitation-gauge stations were then extracted by employing an inverse-distance square-weighting method (Cressman, 1959), which has been used to interpolate climate data for simple and efficient applications (Eum et al., 2014; Shen et al., 2001). This method assumes that an interpolated point is solely influenced by the nearby gridded points based on the inverse of the distance between the interpolated point and the gridded points. The interpolations were carried out on an individual ecodistrict basis and were based on both the number of precipitation-gauge stations and number of 0.5∘ × 0.5∘ grid cells within the ecodistrict in question. For instance, when a single precipitation-gauge station was located within an ecodistrict, the value of the interpolated point was calculated by using all of the gridded points within that ecodistrict. When two or more precipitation-gauge stations were within the same ecodistrict, their interpolated values were calculated by using the same numbers of gridded points but with different weightings based on inverse distance. In the case where an ecodistrict contained one grid cell, no weighting was used and the interpolated value was equal to the nearest grid point.

Comparison of probability distributions using Kolmogorov–Smirnov test

A two-sample, non-parametric Kolmogorov–Smirnov (K–S) test was used to compare the cumulative distribution functions (CDFs) of gridded precipitation products with the AHCCD. The null hypothesis (H0) was that the two datasets came from same population. For each season, monthly total precipitation data were used to avoid commonly known issues of numerous zero values in the daily precipitation data that might affect significance. The K–S test was repeated independently for all precipitation-gauge stations at 5 % significance level (α=0.05). A measure of reliability (in percent) was calculated based on counting the number of stations that do not reject the null hypothesis (any p-values greater than 0.05) over the total number of stations (145 and 137 stations in long-term and short-term comparison respectively), as shown in Eq. (1). %of reliability=number of stations that supportH0total number of precipitation-gauge stations⋅100

Distributions of p value of the K–S test in four seasons for the period of 1979 to 2012 (long-term comparison without CaPA). Note that the numbers of precipitation-gauge stations in each ecozone are different (see Table 4). The p values of regions 6 to 9, and 13 to 14 (R6–R9 and R13–R14), which are more than or equal to 10 stations, were only shown for illustration in box-and-whisker plots with bottom, band (black thick line), and top of the box indicating the 25th, 50th (median), and 75th percentiles respectively.

Distributions of p value of the K–S test in four seasons for the period of 2002 to 2012 (short-term comparison with the inclusion of CaPA). Note that the numbers of precipitation-gauge stations in each ecozone are different (see Table 4). The p values of regions 6, 8 to 9, and 13 to 14 (R6, R8–R9, and R13–R14), which are more than or equal to 10 stations, were only shown for illustration in box-and-whisker plots with bottom, band (black thick line), and top of the box indicating the 25th, 50th (median), and 75th percentiles respectively.

Comparison of gridded precipitation data using performance measures

Since the generation of the climate model-based precipitation products (PCIC dataset and NA-CORDEX dataset) only preserved the statistical properties without considering the day-by-day sequencing of precipitation events in the observational record, these two datasets were excluded from the following comparison, which only focused on the station-based and reanalysis-based gridded products. In particular, these products were assessed in their ability to represent the daily variability of precipitation amounts in different ecozones by four performance measures: percentage of bias (PBias), root mean square error (RMSE) (Erms), correlation coefficient (r), and standard deviation ratio (σG/σR), as shown by Eqs. (2) to (5) respectively. PBias;s=∑iNGi-Ri∑iNRi⋅100,Erms;s=∑iNGi-Ri2N,rs=∑iNGi-G‾Ri-R‾∑iNGi-G‾2∑iNRi-R‾2,σG/σRs=∑iNGi-G‾2N∑iNRi-R‾2N, where s is the season, G and R are the spatial average of the daily gridded precipitation product and the reference observation dataset (precipitation-gauge stations) respectively, G‾ and R‾ are the daily mean of gridded precipitation product and point station data over the time spans (1979–2012 and 2002–2012) respectively, i is the ith day of the season, and N is the total numbers of day in the season. These four performance measures examined different aspects of the gridded precipitation products, with PBias for accuracy of product estimation, RMSE for magnitude of the errors, r for strength and direction of the linear relationship between gridded products and precipitation-gauge station data, and σG/σR for amplitude of the variations.

Results Reliability of precipitation products

The percentage of reliability of each precipitation dataset during every season for the periods of 1979 to 2012 and 2002 to 2012 across Canada is shown in Fig. 2. The higher the percentage, the more reliable the precipitation dataset in question. In general, for long-term comparison (Fig. 2a), WFDEI [GPCC] provided the highest percentage of reliability for the individual seasons (from spring to winter: 72.5, 81.4, 70.3, and 50.3 %) while NARR had the lowest percentage (24.8, 45.5, 27.6, and 11.7 %). Therefore, in spring, WFDEI [GPCC] is not significantly different for 72.5 % of the 145 precipitation-gauge stations, whereas for NARR it is only 24.8 %. ANUSPLIN is second in spring and summer (56.6 and 73.1 %) and WFDEI [CRU] in autumn and winter (63.4 and 45.5 %).

Regarding the PCIC ensembles, the different GCMs provided a range of reliabilities for the individual seasons (Fig. 2c). MPI-ESM-LR performed the best in summer (70.2 %) and CanESM2 in autumn (45.5 %). GFDL-ESM2G generally gave more reliable estimates in spring and winter (57.4 and 41.7 %). Overall, the performance of MPI-ESM-LR (52.0 %) was the best among the GCMs, followed by GFDL-ESM2G (50.1 %), CanESM2 (47.8 %), and HadGEM2 (36.2 %). In terms of statistical downscaling methods, the BCCAQ was on average slightly better than BCSD (49.5 % versus 44.0 %) with the former having a greater similarity in spring and summer as opposed to autumn and winter. These small differences therefore suggest that both methods are similar. With respect to the NA-CORDEX ensembles, the CRCM5 RCM gave the most reliable estimates in summer and autumn regardless of the GCM used. CanRCM4 had the best reliability in spring (49.4 %), whereas RegCM4 had the poorest reliability in spring and summer (24.4 and 34.0 %). Overall, the reliability of MPI-ESM-LR (44.7 %) was better than that of CanESM2 (42.5 %) regardless of the RCMs used, whereas the reliability of CRCM5 (43.6 %) was the best among the RCMs, followed by CanRCM4 (41.2 %), and RegCM4 (32.5 %). It should also be noted that in all cases, the gridded station-based and reanalysis-based products outperformed the climate model-simulated products.

With regard to the short-term comparison (Fig. 2b), ANUSPLIN showed better performance in summer with 94.1 % of reliability among the 137 precipitation-gauge stations while CaPA indicated better skill in winter with 68.6 % of reliability. Again, WFDEI [GPCC] in general provided the most consistent and reliable estimates with over 65 % of reliability in all four seasons. It is interesting to note that for the most part there is a higher percentage of reliability in short-term period compared to long-term period. Reasons for this are not clear but can be partly attributed to the fact that the power of K–S test (i.e. the probability of rejecting the null hypothesis when the alternative is true) decreases with the number of samples.

Figures 3, 4, and 5 display the seasonal distributions of p values using the K–S test for long-term and short-term comparison respectively. Due to the uneven distribution of precipitation-gauge stations across Canada, the number of stations in each ecozone are different (Table 4), with no stations in region 1 (Arctic Cordillera), and regions 2 to 5, 10, 12, and 15 have less than 10 stations. As a result, regions having more than or equal to 10 stations (6 to 9 and 13, 14) were only shown in box-and-whisker plots for illustration. Different colours in the figures corresponded to the various precipitation products. The higher the p values (> 0.05) in each ecozone (represented by a thick black line in box-and-whisker plots towards 1 in y axis in Figs. 3, 4, and 5), the more confidence we attribute to each gridded precipitation datasets in that ecozone.

Number of precipitation-gauge stations within each ecozone.

Region (ecozone) Number of precipitation-gauge stations 1979–2012 2002–2012 1 Arctic Cordillera 0 0 2 Northern Arctic 4 4 3 Southern Arctic 1 1 4 Taiga Plain 2 2 5 Taiga Shield 4 5 6 Boreal Shield 31 29 7 Atlantic Maritime 10 9 8 Mixedwood Plain 18 16 9 Boreal Plain 14 14 10 Prairie 9 7 11 Taiga Cordillera 1 0 12 Boreal Cordillera 6 6 13 Pacific Maritime 15 15 14 Montane Cordillera 28 26 15 Hudson Plain 2 3 Total 145 137

From 1979 to 2012 (Fig. 3), the consistency of each type of precipitation products is explored by assessing the median of the p values. Overall, all the precipitation products showed very low reliability and consistency in winter among these ecozones and in every season in regions 13 and 14 (Pacific Maritime and Montane Cordillera) as the medians were close to zero, despite a couple of locations having a higher chance of the same CDFs as in the precipitation-gauge station data. The WFDEI [GPCC] dataset provided the highest consistency in the remaining three seasons except for region 7 (Atlantic Maritime) where ANUSPLIN showed higher medians (0.51 and 0.46) than WFDEI [GPCC] (0.42 and 0.42) in spring and autumn respectively. Noticeably, NARR provided the lowest median among the reanalysis-based datasets in all four seasons in regions 6 to 8 but gave fairly consistent estimates in regions 9 and 10, especially in summer in region 9 (Boreal Plain) where it came second after WFDEI [GPCC]. The medians of Princeton were similar with those of ANUSPLIN on average in these regions except for summer in which ANUSPLIN offered higher medians than Princeton. WFDEI [CRU] generally showed consistent estimates among these ecozones with medians well above 0.05 except for region 7 (Atlantic Maritime) in spring and autumn. From 1979 to 2005 (Fig. 5), the PCIC ensembles and the NA-CORDEX ensembles showed different degrees of consistency among their GCM members with generally higher p values using BCCAQ method than BCSD method in spring and summer regardless of GCMs in the PCIC datasets. CanESM2 was generally having higher consistency and reliable estimates than MPI-ESM-LR in spring and summer but opposite case in autumn in the NA-CORDEX ensembles. In addition, almost all the precipitation products had lower chance of having same CDFs as the precipitation-gauge stations in ecozones above 60∘ N (regions 2 to 5, 11, and 12) (figure not shown).

Performance measures (accuracy (PBias), magnitude of the errors (RMSE), strength and direction of relationship between gridded products and precipitation-gauge stations (r) and amplitude of the variations (σG/σR)) of each type of gridded precipitation products when evaluating against the precipitation-gauge station data over Canada in four seasons for the time period of 2002 to 2012.

Performance measure Season Precipitation product ANUSPLIN Princeton WFDEI [CRU] WFDEI [GPCC] NARR CaPA PBias (%) Spring -14.2 -12.9 3.1 1.0 5.7 0.7 Summer -9.3 -4.7 2.6 0.8 -1.3 -4.4 Autumn -16.1 -16.0 -3.1 -2.7 -9.3 -1.3 Winter -19.9 -22.4 -3.3 -1.2 -11.9 -8.6 Annual -14.7 -13.6 -1.3 -1.4 -5.7 -4.2 RMSE (mm day-1) Spring 2.39 5.30 3.68 3.64 3.42 2.70 Summer 3.41 7.18 5.33 5.12 5.17 3.74 Autumn 3.00 6.76 4.82 4.70 4.46 3.35 Winter 2.70 5.24 3.95 3.98 3.61 3.05 Annual 3.00 6.33 4.61 4.51 4.35 3.34 r (–) Spring 0.78 0.16 0.53 0.55 0.55 0.72 Summer 0.78 0.13 0.45 0.49 0.46 0.73 Autumn 0.80 0.18 0.53 0.56 0.55 0.75 Winter 0.76 0.17 0.51 0.53 0.54 0.70 Annual 0.79 0.17 0.50 0.54 0.51 0.74 σG/σR (–) Spring 0.72 1.04 0.91 0.95 0.75 0.83 Summer 0.76 0.97 0.80 0.84 0.75 0.82 Autumn 0.74 1.02 0.91 0.95 0.72 0.85 Winter 0.64 0.97 0.96 1.06 0.63 0.72 Annual 0.74 0.99 0.86 0.92 0.72 0.82

Distributions of p value of the K–S test in four seasons for the period of 1979 to 2005 (long-term comparison of PCIC and NA-CORDEX). Note that the numbers of precipitation-gauge stations in each ecozone are different (see Table 4). The p values of regions 6 to 9, and 13 to 14 (R6–R9, and R13–R14), which are more than or equal to 10 stations, were only shown for illustration in box-and-whisker plots with bottom, band (black thick line), and top of the box indicating the 25th, 50th (median), and 75th percentiles respectively.

For the shorter time period of 2002 to 2012 (Fig. 4), CaPA showed the highest consistency in winter in regions 6, 8, 9, and 13, whereas ANUSPLIN was the highest in summer in regions 8, 13, and 14, echoing the results found in Fig. 2. However, the reliability and consistency of CaPA in summer was not particularly high, especially in regions 8 and 13 where the medians were approaching zero. In addition, in ecozones above 60∘ N, similar performances were seen among the precipitation products in the period of 2002 to 2012 as compared with the long-term performance.

Daily variability of precipitation (station- and reanalysis-based products)

The accuracy (PBias), magnitude of the errors (RMSE), strength and direction of the relationship between gridded products and precipitation-gauge station data (r), and amplitude of the variations (σG/σR) are shown in Figs. 6 and 7 for the period of 1979 to 2012. In general, the gridded precipitation products that agree well with the precipitation-gauge station data should have relatively high correlation and low RMSE, low bias and similar standard deviation (light grey or dark grey squares in Figs. 6 and 7).

Portrait diagram showing the accuracy (PBias) (a) and amplitude of the variations (σG/σR) (b) of each type of gridded precipitation products when evaluating against the precipitation-gauge station data in each ecozone (regions 1 to 15) in four seasons for the time period of 1979 to 2012. Each column indicates one gridded precipitation product and each row represents one ecozone with numerical code corresponding to region shown in Fig. 1. White indicates that no data are available due to no precipitation-gauge stations existing in that region.

Portrait diagram showing magnitude of the errors (RMSE) (a), and strength and direction of relationship between gridded products and precipitation-gauge stations (r) (b) of each type of gridded precipitation product when evaluating against the precipitation-gauge station data in each ecozone (regions 1 to 15) in four seasons for the time period of 1979 to 2012. Each column indicates one gridded precipitation product and each row represents one ecozone with numerical code corresponding to the region shown in Fig. 1. White indicates that no data are available due to no precipitation-gauge stations existing in that region.

In terms of accuracy (Fig. 6a), all precipitation products tended to generally overestimate total precipitation in regions 12 to 14, whereas region 14 (Montane Cordillera) had the overall highest positive PBias for the individual seasons (from spring to winter: > 20.9, > 6.24, > 14.4, and > 26.8 %). On the other hand, all products mostly underestimated the precipitation amounts in regions 3 to 6, 9, and 10. This was especially worse in region 3 (southern Arctic) where the underestimation of precipitation amounts for the individual seasons were > -22.6, > -2.2, > -10.2, and > -28.1 % respectively. In particular, ANUSPLIN was associated with a generally negative PBias for all the ecozones in four seasons, except for regions 12 (Boreal Cordillera) and 14 (Montane Cordillera). The accuracy of ANUSPLIN was the worst in winter, with underestimation of precipitation amounts ranging from -7.8 % in region 13 (Pacific Maritime) to -38.7 % in region 3 (southern Arctic). WFDEI [CRU] and WFDEI [GPCC] had similar performances across different regions. They performed particularly well in summer in regions 2 to 9 where the accuracy was within -4.6 to 4.2 %. With the exception of regions 13 and 14, Princeton and NARR generally provided the overall largest and second largest underestimation of precipitation amounts across different ecozones. NARR performed the worst in regions 7 (Atlantic Maritime) and 8 (Mixedwood Plain) where the precipitation amounts for the individual seasons were underestimated by > -42.0, > -33.1, > -38.8, and > -59.7 %.

When examining the magnitude of errors (Fig. 7a), all products showed very high magnitude of errors in regions 6 to 8, and 13 while region 13 (Pacific Maritime) had the greatest RMSE for the individual seasons (from spring to winter: > 5.35, > 3.74, > 7.82, and > 8.24 mm day-1). Specifically, ANUSPLIN showed generally better correspondence with precipitation-gauge station data, providing the overall lowest RMSE across ecozones in four seasons (2.50, 3.24, 2.79, and 2.45 mm day-1) with the only exception in spring in region 15 (Hudson Plain). Moreover, referring to Fig. 7b, ANUSPLIN had the overall highest r across ecozones in four seasons (0.75, 0.78, 0.80, and 0.74). On the contrary, Princeton had the worst performance in both magnitude of errors and correlation with observations irrespective of ecozone or season, with the grand RMSE and r of 5.65 mm day-1 and 0.17 respectively. The performances of WFDEI [CRU], WFDEI [GPCC], and NARR were in between ANUSPLIN and Princeton and they shared similar RMSE and r across different regions and seasons. The resulting values of the RMSE metric in regions 7 (Atlantic Maritime) and 13 (Pacific Maritime) tended to be larger than that of other ecozones. However, the other metrics such as PBias and r showed better performance in these regions. This suggests that higher RMSE values can be mainly attributed to the fact that precipitation amounts are higher in the maritime regions.

Regarding the amplitude of variations (Fig. 6b), all datasets generally had variations that were much smaller than precipitation-gauge station data in regions 3, 4, and 11 in four seasons. In particular, ANUSPLIN and NARR were consistently having too little variability across different ecozones, especially in winter in which σG/σR ranged from 0.41 in region 15 (Hudson Plain) to 0.76 in region 13 (Pacific Maritime). WFDEI [CRU] and WFDEI [GPCC] had the most similar standard deviations as that of precipitation-gauge station data in regions 5 to 8 in autumn and winter, while Princeton estimated σG/σR the best in regions 4 to 10 in summer. However, Princeton had much larger variability in regions 12 to 14 in spring and regions 6 to 8 in autumn.

Concerning the short-term comparison (Table 5), CaPA performed the best in spring and autumn in terms of accuracy, with the lowest positive PBias of 0.7 % and the lowest negative PBias of -1.3 % respectively. The performance of CaPA generally resembled that of ANUSPLIN regarding the magnitude of errors and correlation with observations, which were the second lowest RMSE for the individual seasons (from spring to winter: 2.70, 3.74, 3.35, and 3.05 mm day-1) and the second highest r (0.72, 0.73, 0.75, and 0.70) respectively. Despite its better performances in RMSE and r, CaPA was generally not able to capture satisfactorily the amplitude of variations, with consistently lower values in four seasons (0.83, 0.82, 0.85, and 0.72). However, CaPA showed more skill compared to ANUSPLIN (0.72, 0.76, 0.74, and 0.64) and NARR (0.75, 0.75, 0.72, and 0.63). In addition, the five gridded products in the long-term comparison performed similarly in the period of 2002 to 2012, with ANUSPLIN having the lowest annual RMSE and the highest annual r of 3.00 mm day-1 and 0.79, and Princeton being the worst again with the highest annual RMSE and lowest annual r of 6.33 mm day-1 and 0.17 respectively.

Discussion

The preceding has provided insight into the relative performance of various gridded precipitation products over Canada relative to gauge measurements over different seasons and ecozones. Results showed that there is no particular product that is superior for all performance measures although some datasets are consistently better. Based on the performances, one could broadly characterize the station- and reanalysis-based precipitation products into four groups: (1) ANUSPLIN and CaPA with negative PBias, low RMSE, high r, and small σG/σR; (2) WFDEI [CRU] and WFDEI [GPCC] with relatively small PBias, high RMSE, fair r, and similar standard deviation; (3) Princeton with negative PBias, high RMSE, low r, and a mixture of large and small σG/σR; and (4) NARR with negative PBias, high RMSE, fair r, and small σG/σR. Among the reanalysis-based gridded products, Princeton performed the worst in all seasons and regions in terms of minimizing error magnitudes (Figs. 8 and 9). Princeton was especially poor in winter (Fig. 8) and showed significant underestimation in regions above 60∘ N (Fig. 9). This could be due to the use of the NCEP-NCAR reanalysis as the basis to generate the dataset, which have been shown to be less accurate than NCEP-DOE reanalysis (used in NARR) and ERA-40 reanalysis (used in WFD) (Sheffield et al., 2006). The better performance of NARR in capturing the timings and amounts of precipitation compared to Princeton was probably because NCEP-DOE reanalysis was a major improvement upon the earlier NCEP-NCAR reanalysis in both resolution and accuracy. However, the overall reliability of NARR was among the poorest mainly because of non-assimilation of gauge precipitation observations over Canada from 2004 onwards, as reported by Mesinger et al. (2006). ANUSPLIN and CaPA performed well in capturing the timings and minimizing the error magnitudes of the precipitation, despite their general underestimation across Canada (PBias ranging from -7.7 % (region 13) to -40.7 % (region 3) and -2.0 % (region 15) to -17.1 % (region 8) in the period of 2002 to 2012) (Fig. 9) and too little variability (grand σG/σR of 0.72 and 0.80 of the same period). This was not surprising given that the generation of the products was based on the unadjusted precipitation-gauge stations where the total rainfall amounts were increased after adjustment (Mekis and Vincent, 2011). WFDEI [CRU] and WFDEI [GPCC], on the other hand, performed well in estimating the accuracy and amplitude of variations, but not the timings and error magnitudes of the precipitation. This could probably be due to the positive bias offsetting the negative bias resulting in small mean bias, but was picked up by RMSE that gives more weight to the larger errors. The larger errors could result from a mismatch of occurrence of precipitation in the time series, as reflected by the fair correlation coefficients (grand r of 0.52 and 0.50 for WFDEI [CRU], 0.54 and 0.53 for WFDEI [GPCC], for time periods of 1979 to 2012 and 2002 to 2012 respectively).

Scatter plots showing absolute PBias (x axis) versus RMSE (y axis) of each precipitation dataset in four seasons and the entire year for the period of 1979 to 2012 (a) and 2002 to 2012 (b). Each hollow circle represents one ecozone and the solid stars indicate the overall average across ecozones.

Bar graphs showing the annual accuracy (PBias) (first row) and magnitude of the errors (RMSE) (second row) of each precipitation dataset for the period of 1979 to 2012 (a, c) and 2002 to 2012 (b, d) in different ecozones. The white bar shows the scale of the bars with the number beside it indicating the value of the bar.

By matching the statistical properties of the adjusted gauge measurements at monthly timescale, one could establish the confidence in using the climate model-simulated products for long-term hydro-climatic studies. Comparing the overall reliability of the PCIC and NA-CORDEX datasets, it was found that for the individual seasons the PCIC ensembles (spring, summer, and winter: 54.0, 64.7, and 35.7 %) outperformed the NA-CORDEX ensembles (39.1, 45.0, and 31.3 %) except in autumn when the NA-CORDEX ensembles (45.5 %) provided slightly higher reliability than the PCIC ensembles (45.2 %). The better reliability of the PCIC datasets could be due to the use of ANUSPLIN to train the GCMs and thus the statistical properties of the downscaled outputs are guided by those of the ANUSPLIN. Similarly, for ecozones where more than 10 precipitation-gauge stations could be found (regions 6 to 9, 13 and 14), the PCIC ensembles (reliability ranging from 35.7 to 64.4 %) also outperformed the NA-CORDEX ensembles (from 17.2 to 61.6 %). This would suggest that the PCIC ensembles may be the preferred choice for long-term climate change impact assessment over Canada, although further research is required.

The evaluations of this comparison were impacted by the spatial distribution of adjusted precipitation-gauge stations (Mekis and Vincent, 2011), which were assumed to be the best representation of reality owing to efforts in improving the raw archive of the precipitation-gauge stations. However, the major limitation of this dataset was the number of precipitation-gauge stations that could be used for comparison. As aforementioned, due to temporal coverage not encompassing the entire study period and not having a complete year for 2012, over half of the precipitation-gauge stations were discarded from the analysis. Although the locations of the remaining stations covered much of Canada, there are only one or a few stations located in some of the ecozones (e.g. region 3 to 5, 11, and 15). Even in region 10 (prairie) there are only nine precipitation-gauge stations for analysis. While the reliability of different types of gridded products could be tested in these ecozones, the consistency of the performance of each gridded product could not be established due to small sample sizes.

In addition, results from the above analysis should be interpreted with care because the precipitation-gauge station data are point measurements while the gridded precipitation products are areal averages, of which the accuracy and precision of the estimates can be very different given the non-linear responses of precipitation (Ebert et al., 2007). When comparing point measurements and areal-average estimates, fundamental challenges occur because of the sampling errors arising from different sampling schemes and errors related to gauge instrumentation (Bowman, 2005). It is therefore difficult to have perfect spatial matching between point measurements (gauge stations) and areal-averaged estimates (gridded products) (Sapiano and Arkin, 2009; Hong et al., 2007). However, in the absence of a sufficiently dense precipitation-gauge network in Canada, the options for assessing different gridded products are limited. The only gridded product that is basically representing areal averages of precipitation (via interpolation) based on ground observations is ANUSPLIN. As aforementioned (see Sect. 3.2.1), this product has its own limitations and may not be qualified to be considered as the “ground truth”. Therefore, ANUSPLIN is also included in the pool of gridded products to be evaluated. Notwithstanding the issues, using the selected gauge measurements would remain the best way for the evaluation of the multiple gridded products because the set of gauges used had been adjusted (e.g. for undercatch) and are the most accurate source of information on precipitation in Canada (although small with limited spatial coverage). Also, given that all the gridded products are compared against this common set of station observations, it is assumed that the bias that the difference between point and areal data introduces into the analysis is consistent for all the products. Therefore, given the current data situation, the preceding methods could be used for comparing the performance of different daily gridded precipitation products.

Conclusion

A number of gridded climate products incorporating multiple sources of data have recently been developed with the aim of providing better and more reliable measurements for climate and hydrological studies. There is a pressing need for characterizing the quality and error characteristics of various precipitation products and assessing how they perform at different spatial and temporal scales. This is particularly important in light of the fact that these products are the main driver of hydrological models in many regions, including Canadian watersheds where precipitation-gauge network is typically limited and sparse. This study was conducted to inter-compare the probability distributions of several gridded precipitation products and quantify the spatial and temporal variability of the errors relative to station observations in Canada, so as to provide some insights for potential users in selecting the products for their particular interests and applications. Based on the above analysis, the following conclusions can be drawn:

In general, all the products performed best in summer, followed by autumn, spring, and winter in order of decreasing quality. The lower reliability in winter is likely the result of difficulty in accurately capturing solid precipitation.

Overall, WFDEI [GPCC] and CaPA performed best with respect to different performance measures. WFDEI [GPCC], however, may be a better choice for long-term analyses as it covers a longer historical period. ANUSPLIN and WFDEI [CRU] also performed comparably well, with considerably lower quality than WFDEI [GPCC] and CaPA. Princeton and NARR demonstrated the lowest quality in terms of different performance measures.

Station-based and reanalysis-based products tended to underestimate total precipitation across Canada except in southwestern regions (Pacific Maritime and Montane Cordillera) where the tendency was towards overestimation. This may be the due to the fact that the majority of precipitation-gauge stations are located at lower altitudes, which might not accurately reflect areal precipitation due to topographic effect.

In southern Canada, WFDEI [GPCC] and CaPA demonstrated their best performance in the western cold interior (Boreal Plain, prairie, Montane Cordillera) in terms of timing and magnitude of daily precipitation.

In northern Canada (above 60∘ N), the different products tended to moderately (ranging from -0.6 to -40.3 %) and, in some cases, significantly (up to -60.3 % in Taiga Cordillera) underestimate total precipitation, while reproducing the timing of daily precipitation rather well. It should be noted that this assessment was based on only a limited number of precipitation-gauges in the north.

Comparing the climate model-simulated products, PCIC ensembles generally performed better than NA-CORDEX ensembles in terms of reliability and consistency in four seasons across Canada.

In terms of statistical downscaling methods, the BCCAQ method was slightly more reliable than the BCSD method across Canada on the annual basis.

Regarding GCMs, MPI-ESM-LR provides the highest reliability, followed by GFDL-ESM2G, CanESM2, and HadGEM2. With respect to RCMs, CRCM5 performed the best regardless of the GCM used, followed by CanRCM4, and RegCM4.

The findings from this analysis provide additional information for potential users to draw inferences about the relative performance of different gridded products. Although no clear-cut product was shown to be superior, researchers/users can use this information for selecting or excluding various datasets depending on their purpose of study. It is realized that this investigation only focused on the daily timescale at a relatively coarse 0.5∘ × 0.5∘ resolution suitable for large-scale hydro-climatic studies. Further research is thus required towards performance assessment of various products with respect to precipitation extremes, which often have the greatest hydro-climatic impacts. As new products become available, similar comparisons should be conducted to assess their reliability.

The data used in this study are available from different sources. AHCCD and ANUSPLIN are currently available online through Environment and Climate Change Canada (http://open.canada.ca/data/en/dataset/d6813de6-b20a-46cc-8990-01862ae15c5f; AHCCD, 2016, and http://open.canada.ca/data/en/dataset/d432cb3d-8266-4487-b894-06224a4dfd5b; ANUSPLIN, 2016). Other precipitation products are also available from their respective websites and can be accessed publicly online: Princeton (http://hydrology.princeton.edu/data.pgf.php; Princeton, 2016), WFDEI (http://www.eu-watch.org/data_availability; WFDEI, 2016), NARR (https://www.ncdc.noaa.gov/data-access/model-data/model-datasets/north-american-regional-reanalysis-narr; NARR, 2016), PCIC (https://www.pacificclimate.org/data/statistically-downscaled-climate-scenarios; PCIC, 2016). CaPA are not publicly available, but an information leaflet can be found online (http://collaboration.cmc.ec.gc.ca/cmc/cmoi/product_guide/docs/lib/capa_information_leaflet_20141118_en.pdf; CaPA, 2016). NA-CORDEX datasets are available by contacting the corresponding authors on the NA-CORDEX website (https://na-cordex.org/simulations-modeling-group; NA-CORDEX, 2016).

The Supplement related to this article is available online at doi:10.5194/hess-21-2163-2017-supplement.

The authors declare that they have no conflict of interest.

Acknowledgements

The financial support from the Canada Excellence Research Chair in Water Security is gratefully acknowledged. Thanks are due to Melissa Bukovsky and Katja Winger from the NA-CORDEX modelling group for providing access to RegCM4 and CRCM5 data used in this study. The authors are also grateful to the various organizations including the Environment and Climate Change Canada that made the datasets freely available to the scientific community. We furthermore would like to thank the anonymous reviewers and Jan Seibert, the editor, for their contributions leading to a significantly improved quality of the paper. Edited by: J. Seibert Reviewed by: three anonymous referees

References 1

Adam, J. C. and Lettenmaier, D. P.: Adjustment of global gridded precipitation for systematic bias, J. Geophys. Res.-Atmos., 108, 4257, 10.1029/2002jd002499, 2003.

AHCCD: Adjusted and Homogenized Canadian Climate Data, available at: http://open.canada.ca/data/en/dataset/d6813de6-b20a-46cc-8990-01862ae15c5f, last access: January 2016.

Adler, R. F., Kidd, C., Petty, G., Morissey, M., and Goodman, H. M.: Intercomparison of global precipitation products: The third Precipitation Intercomparison Project (PIP-3), B. Am. Meteorol. Soc., 82, 1377–1396, 10.1175/1520-0477(2001)082<1377:Iogppt>2.3.Co;2, 2001.

ANUSPLIN for Canada: Australian National University Spline model for Canada, available at: http://open.canada.ca/data/en/dataset/d432cb3d-8266-4487-b894-06224a4dfd5b, last access: February 2016.

Asadullah, A., McIntyre, N., and Kigobe, M.: Evaluation of five satellite products for estimation of rainfall over Uganda, Hydrolog. Sci. J., 53, 1137–1150, 10.1623/hysj.53.6.1137, 2008.

Asong, Z. E., Khaliq, M. N., and Wheater, H. S.: Regionalization of precipitation characteristics in the Canadian Prairie Provinces using large-scale atmospheric covariates and geophysical attributes, Stoch. Env. Res. Risk A, 29, 875–892, 10.1007/s00477-014-0918-z, 2015.

Behrangi, A., Christensen, M., Lebsock, M. R., Stephens, G., Huffman, G. J., Bolvin, D., Adler, R. F., Gardner, A., Lambrigtsen, B., and Fetzer, E.: Status of High latitude precipitation estimates from observations and reanalyses, J. Geophys. Res.-Atmos., 121, 4468–4486, 2016.

Betts, A. K., Zhao, M., Dirmeyer, P. A., and Beljaars, A. C. M.: Comparison of ERA40 and NCEP/DOE near-surface data sets with other ISLSCP-II data sets, J. Geophys. Res.-Atmos., 111, D22s04, 10.1029/2006jd007174, 2006.

Bhargava, M. and Danard, M.: Application of Optimum Interpolation to the Analysis of Precipitation in Complex Terrain, J. Appl. Meteorol., 33, 508–518, 10.1175/1520-0450(1994)033<0508:Aooitt>2.0.Co;2, 1994.

Black, T. L.: The step-mountain, eta coordinate regional model: A documentation, National Meteorological Center, Development Division, NOAA Environmental Modeling Center, Room 207, 5200 Auth Road, Camp Springs, MD 20746, 1988.

Blenkinsop, S. and Fowler, H. J.: Changes in European drought characteristics projected by the PRUDENCE regional climate models, Int. J. Climatol., 27, 1595–1610, 10.1002/joc.1538, 2007.

Bonsal, B. R., Aider, R., Gachon, P., and Lapp, S.: An assessment of Canadian prairie drought: past, present, and future, Clim. Dynam., 41, 501–516, 10.1007/s00382-012-1422-0, 2013.

Bosilovich, M. G., Chen, J. Y., Robertson, F. R., and Adler, R. F.: Evaluation of global precipitation in reanalyses, J. Appl. Meteorol. Clim., 47, 2279–2299, 10.1175/2008jamc1921.1, 2008.

Bowman, K. P.: Comparison of TRMM precipitation retrievals with rain gauge data from ocean buoys, J. Climate, 18, 178–190, 10.1175/Jcli3259.1, 2005.

Brooks, R., Harvey, K., Kirk, D., Soulard, F., Paul, P., and Murray, A.: Building a Canadian Digital Drainage Area Framework, 55th Annual CWRA Conference, Winnipeg, Manitoba, Canada, 2002.

Bukovsky, M. S. and Karoly, D. J.: A brief evaluation of precipitation from the North American Regional Reanalysis, J. Hydrometeorol., 8, 837–846, 10.1175/Jhm595.1, 2007.

CaPA: the Canadian Precipitation Analysis, Environment and Climate Change Canada, Information leaflet, available online: http://collaboration.cmc.ec.gc.ca/cmc/cmoi/product_guide/docs/lib/capa_information_leaflet_20141118_en.pdf, last access: March 2016.

Carrera, M. L., Belair, S., and Bilodeau, B.: The Canadian Land Data Assimilation System (CaLDAS): Description and Synthetic Evaluation Study, J. Hydrometeorol., 16, 1293–1314, 10.1175/Jhm-D-14-0089.1, 2015.

Chadburn, S. E., Burke, E. J., Essery, R. L. H., Boike, J., Langer, M., Heikenfeld, M., Cox, P. M., and Friedlingstein, P.: Impact of model developments on present and future simulations of permafrost in a global land-surface model, The Cryosphere, 9, 1505–1521, 10.5194/tc-9-1505-2015, 2015.

Chen, D. L., Achberger, C., Raisanen, J., and Hellstrom, C.: Using statistical downscaling to quantify the GCM-related uncertainty in regional climate change scenarios: A case study of Swedish precipitation, Adv. Atmos. Sci., 23, 54–60, 10.1007/s00376-006-0006-5, 2006.

Choi, W., Kim, S. J., Rasmussen, P. F., and Moore, A. R.: Use of the North American Regional Reanalysis for Hydrological Modelling in Manitoba, Can. Water Resour. J., 34, 17–36, 2009.

Christensen, J. H., Carter, T. R., Rummukainen, M., and Amanatidis, G.: Evaluating the performance and utility of regional climate models: the PRUDENCE project, Climatic Change, 81, 1–6, 10.1007/s10584-006-9211-6, 2007.

Ciach, G. J.: Local random errors in tipping-bucket rain gauge measurements, J. Atmos. Ocean Tech., 20, 752–759, 10.1175/1520-0426(2003)20<752:Lreitb>2.0.Co;2, 2003.

Cote, J., Desmarais, J. G., Gravel, S., Methot, A., Patoine, A., Roch, M., and Staniforth, A.: The operational CMC-MRB Global Environmental Multiscale (GEM) model. Part II: Results, Mon. Weather Rev., 126, 1397–1418, 10.1175/1520-0493(1998)126<1397:Tocmge>2.0.Co;2, 1998a.

Cote, J., Gravel, S., Methot, A., Patoine, A., Roch, M., and Staniforth, A.: The operational CMC-MRB Global Environmental Multiscale (GEM) model. Part I: Design considerations and formulation, Mon. Weather Rev., 126, 1373–1395, 10.1175/1520-0493(1998)126<1373:Tocmge>2.0.Co;2, 1998b.

Covey, C., AchutaRao, K. M., Cubasch, U., Jones, P., Lambert, S. J., Mann, M. E., Phillips, T. J., and Taylor, K. E.: An overview of results from the Coupled Model Intercomparison Project, Global Planet. Change, 37, 103–133, 10.1016/S0921-8181(02)00193-5, 2003.

Cressman, G. P.: An operational objective analysis system, Mon. Weather Rev., 87, 367–374, 1959.

Cuo, L., Beyene, T. K., Voisin, N., Su, F. G., Lettenmaier, D. P., Alberti, M., and Richey, J. E.: Effects of mid-twenty-first century climate and land cover change on the hydrology of the Puget Sound basin, Washington, Hydrol. Process., 25, 1729–1753, 10.1002/hyp.7932, 2011.

Daley, R.: Atmospheric data analysis, 2, Cambridge university press, Cambridge, UK, 1993.

Dee, D. P., Uppala, S. M., Simmons, A. J., Berrisford, P., Poli, P., Kobayashi, S., Andrae, U., Balmaseda, M. A., Balsamo, G., Bauer, P., Bechtold, P., Beljaars, A. C. M., van de Berg, L., Bidlot, J., Bormann, N., Delsol, C., Dragani, R., Fuentes, M., Geer, A. J., Haimberger, L., Healy, S. B., Hersbach, H., Holm, E. V., Isaksen, L., Kallberg, P., Kohler, M., Matricardi, M., McNally, A. P., Monge-Sanz, B. M., Morcrette, J. J., Park, B. K., Peubey, C., de Rosnay, P., Tavolato, C., Thepaut, J. N., and Vitart, F.: The ERA-Interim reanalysis: configuration and performance of the data assimilation system, Q. J. Roy. Meteor. Soc., 137, 553–597, 10.1002/qj.828, 2011.

Dinku, T., Anagnostou, E. N., and Borga, M.: Improving radar-based estimation of rainfall over complex terrain, J. Appl. Meteorol., 41, 1163–1178, 10.1175/1520-0450(2002)041<1163:Irbeor>2.0.Co;2, 2002.

Dinku, T., Connor, S. J., Ceccato, P., and Ropelewski, C. F.: Comparison of global gridded precipitation products over a mountainous region of Africa, Int. J. Climatol., 28, 1627–1638, 10.1002/joc.1669, 2008.

Dore, M. H. I.: Climate change and changes in global precipitation patterns: What do we know?, Environ. Int., 31, 1167–1181, 10.1016/j.envint.2005.03.004, 2005.

Ebert, E. E., Janowiak, J. E., and Kidd, C.: Comparison of near-real-time precipitation estimates from satellite observations and numerical models, B. Am. Meteorol. Soc., 88, 47–64, 10.1175/Bams-88-1-47, 2007.

Ecological Stratification Working Group: A national ecological framework for Canada, Centre for Land and Biological Resources Research, Research Branch, Agriculture and Agrifood Canada, 1996.

Ek, M. B., Mitchell, K. E., Lin, Y., Rogers, E., Grunmann, P., Koren, V., Gayno, G., and Tarpley, J. D.: Implementation of Noah land surface model advances in the National Centers for Environmental Prediction operational mesoscale Eta model, J. Geophys. Res.-Atmos., 108, 8851, 10.1029/2002jd003296, 2003.

Eum, H. I., Gachon, P., Laprise, R., and Ouarda, T.: Evaluation of regional climate model simulations versus gridded observed and regional reanalysis products using a combined weighting scheme, Clim. Dynam., 38, 1433–1457, 10.1007/s00382-011-1149-3, 2012.

Eum, H. I., Dibike, Y., Prowse, T., and Bonsal, B.: Inter-comparison of high-resolution gridded climate data sets and their implication on hydrological model simulation over the Athabasca Watershed, Canada, Hydrol. Process., 28, 4250–4271, 10.1002/hyp.10236, 2014.

Forbes, K. A., Kienzle, S. W., Coburn, C. A., Byrne, J. M., and Rasmussen, J.: Simulating the hydrological response to predicted climate change on a watershed in southern Alberta, Canada, Climatic Change, 105, 555–576, 10.1007/s10584-010-9890-x, 2011.

Fortin, V., Jean, M., Brown, R., and Payette, S.: Predicting Snow Depth in a Forest-Tundra Landscape using a Conceptual Model Allowing for Snow Redistribution and Constrained by Observations from a Digital Camera, Atmos. Ocean, 53, 200–211, 10.1080/07055900.2015.1022708, 2015a.

Fortin, V., Roy, G., Donaldson, N., and Mahidjiba, A.: Assimilation of radar quantitative precipitation estimations in the Canadian Precipitation Analysis (CaPA), J. Hydrol., 531, 296–307, 2015b.

Frei, C., Scholl, R., Fukutome, S., Schmidli, J., and Vidale, P. L.: Future change of precipitation extremes in Europe: Intercomparison of scenarios from regional climate models, J. Geophys. Res.-Atmos., 111, D06105, 10.1029/2005jd005965, 2006.

Fuchs, T.: GPCC annual report for year 2008: Development of the GPCC data base and analysis products, DWD Rep., Global Precipitation Climatology Centre, 13 pp., available at: https://www.dwd.de/EN/ourservices/gpcc/reports_publications/GPCC_annual_report_2008.pdf?__blob=publicationFile&v=3 (last access: January 2016), 2009.

Garand, L. and Grassotti, C.: Toward an Objective Analysis of Rainfall Rate Combining Observations and Short-Term Forecast Model Estimates, J. Appl. Meteorol., 34, 1962–1977, 10.1175/1520-0450(1995)034<1962:Taoaor>2.0.Co;2, 1995.

Gebregiorgis, A. S. and Hossain, F.: How well can we estimate error variance of satellite precipitation data around the world?, Atmos. Res., 154, 39–59, 10.1016/j.atmosres.2014.11.005, 2015.

Gebremichael, M., Krajewski, W. F., Morrissey, M. L., Huffman, G. J., and Adler, R. F.: A detailed evaluation of GPCP 1 degrees daily rainfall estimates over the Mississippi river basin, J. Appl. Meteorol., 44, 665–681, 10.1175/Jam2233.1, 2005.

Giorgi, F., Jones, C., and Asrar, G. R.: Addressing climate information needs at the regional level: the CORDEX framework, World Meteorological Organization (WMO) Bulletin, 58, 175, 2009.

Gottschalck, J., Meng, J., Rodell, M., and Houser, P.: Analysis of multiple precipitation products and preliminary assessment of their impact on global land data assimilation system land surface states, J. Hydrometeorol., 6, 573–598, 10.1175/Jhm437.1, 2005.

Gupta, S. K., Ritchey, N. A., Wilber, A. C., Whitlock, C. H., Gibson, G. G., and Stackhouse, P. W.: A climatology of surface radiation budget derived from satellite data, J. Climate, 12, 2691–2710, 10.1175/1520-0442(1999)012<2691:Acosrb>2.0.Co;2, 1999.

Hively, W. D., Gérard-Marchant, P., and Steenhuis, T. S.: Distributed hydrological modeling of total dissolved phosphorus transport in an agricultural landscape, part II: dissolved phosphorus transport, Hydrol. Earth Syst. Sci., 10, 263–276, 10.5194/hess-10-263-2006, 2006.

Hong, Y., Gochis, D., Cheng, J. T., Hsu, K. L., and Sorooshian, S.: Evaluation of PERSIANN-CCS rainfall measurement using the NAME Event Rain Gauge Network, J. Hydrometeorol., 8, 469–482, 10.1175/Jhm574.1, 2007.

Hong, Y., Robert, F., Adler, R. F., George, J., Huffman, J., and Pierce, H.: Applications of TRMM-based multi-satellite precipitation estimation for global runoff prediction: prototyping a global flood modeling system, in: Satellite Rainfall Applications for Surface Hydrology, edited by: Gebremichael, M. and Hossain F., Springer Netherlands, 245–265, 10.1007/978-90-481-2915-7_15, 2010.

Hopkinson, R. F., McKenney, D. W., Milewska, E. J., Hutchinson, M. F., Papadopol, P., and Vincent, L. A.: Impact of Aligning Climatological Day on Gridding Daily Maximum-Minimum Temperature and Precipitation over Canada, J. Appl. Meteorol. Clim., 50, 1654–1665, 10.1175/2011jamc2684.1, 2011.

Hou, A. Y., Kakar, R. K., Neeck, S., Azarbarzin, A. A., Kummerow, C. D., Kojima, M., Oki, R., Nakamura, K., and Iguchi, T.: The Global Precipitation Measurement Mission, B. Am. Meteorol. Soc., 95, 701–722, 10.1175/Bams-D-13-00164.1, 2014.

Huffman, G. J., Adler, R. F., Morrissey, M. M., Bolvin, D. T., Curtis, S., Joyce, R., McGavock, B., and Susskind, J.: Global precipitation at one-degree daily resolution from multisatellite observations, J. Hydrometeorol., 2, 36–50, 10.1175/1525-7541(2001)002<0036:Gpaodd>2.0.Co;2, 2001.

Huffman, G. J., Adler, R. F., Stocker, E., Bolvin, D. T., and Nelkin, E. J.: Analysis of TRMM 3-hourly multi-satellite precipitation estimates computed in both real and post-real time, Analysis of TRMM 3-Hourly Multi-Satellite Precipitation Estimates Computed in Both Real and Post-Real Time, Combined Preprints CD-ROM, 83rd AMS Annual Meeting, Paper P4.11, 12th Conf. on Sat. Meteor. and Oceanog., 9–13 February 2003, Long Beach, CA, 6 pp., 2002.

Huisman, J. A., Breuer, L., Bormann, H., Bronstert, A., Croke, B. F. W., Frede, H. G., Graff, T., Hubrechts, L., Jakeman, A. J., Kite, G., Lanini, J., Leavesley, G., Lettenmaier, D. P., Lindstrom, G., Seibert, J., Sivapalan, M., Viney, N. R., and Willems, P.: Assessing the impact of land use change on hydrology by ensemble modeling (LUCHEM) III: Scenario analysis, Adv. Water Resour., 32, 159–170, 10.1016/j.advwatres.2008.06.009, 2009.

Hutchinson, M. F., Mckenney, D. W., Lawrence, K., Pedlar, J. H., Hopkinson, R. F., Milewska, E., and Papadopol, P.: Development and Testing of Canada-Wide Interpolated Spatial Models of Daily Minimum-Maximum Temperature and Precipitation for 1961–2003, J. Appl. Meteorol. Clim., 48, 725–741, 10.1175/2008jamc1979.1, 2009.

Jameson, A. R. and Kostinski, A. B.: Spurious power-law relations among rainfall and radar parameters, Q. J. Roy. Meteor. Soc., 128, 2045–2058, 10.1256/003590002320603520, 2002.

Janowiak, J. E., Gruber, A., Kondragunta, C. R., Livezey, R. E., and Huffman, G. J.: A comparison of the NCEP-NCAR reanalysis precipitation and the GPCP rain gauge-satellite combined dataset with observational error considerations, J. Climate, 11, 2960–2979, 10.1175/1520-0442(1998)011<2960:Acotnn>2.0.Co;2, 1998.

Kalnay, E., Kanamitsu, M., Kistler, R., Collins, W., Deaven, D., Gandin, L., Iredell, M., Saha, S., White, G., Woollen, J., Zhu, Y., Chelliah, M., Ebisuzaki, W., Higgins, W., Janowiak, J., Mo, K. C., Ropelewski, C., Wang, J., Leetmaa, A., Reynolds, R., Jenne, R., and Joseph, D.: The NCEP/NCAR 40-year reanalysis project, B. Am. Meteorol. Soc., 77, 437–471, 10.1175/1520-0477(1996)077<0437:Tnyrp>2.0.Co;2, 1996.

Kanamitsu, M., Ebisuzaki, W., Woollen, J., Yang, S. K., Hnilo, J. J., Fiorino, M., and Potter, G. L.: NCEP-DOE AMIP-II reanalysis (R-2), B. Am. Meteorol. Soc., 83, 1631–1643, 10.1175/Bams-83-11-1631, 2002.

Kang, D. H., Shi, X. G., Gao, H. L., and Dery, S. J.: On the Changing Contribution of Snow to the Hydrology of the Fraser River Basin, Canada, J. Hydrometeorol., 15, 1344–1365, 10.1175/Jhm-D-13-0120.1, 2014.

Kay, A. L., Davies, H. N., Bell, V. A., and Jones, R. G.: Comparison of uncertainty sources for climate change impacts: flood frequency in England, Climatic Change, 92, 41–63, 10.1007/s10584-008-9471-4, 2009.

Kidd, C., Bauer, P., Turk, J., Huffman, G. J., Joyce, R., Hsu, K. L., and Braithwaite, D.: Intercomparison of High-Resolution Precipitation Products over Northwest Europe, J. Hydrometeorol., 13, 67–83, 10.1175/Jhm-D-11-042.1, 2012.

Kienzle, S. W., Nemeth, M. W., Byrne, J. M., and MacDonald, R. J.: Simulating the hydrological impacts of climate change in the upper North Saskatchewan River basin, Alberta, Canada, J. Hydrol., 412, 76–89, 10.1016/j.jhydrol.2011.01.058, 2012.

Kimoto, M., Yasutomi, N., Yokoyama, C., and Emori, S.: Projected Changes in Precipitation Characteristics around Japan under the Global Warming, Sola, 1, 85–88, 10.2151/sola.2005-023, 2005.

Kistler, R., Kalnay, E., Collins, W., Saha, S., White, G., Woollen, J., Chelliah, M., Ebisuzaki, W., Kanamitsu, M., Kousky, V., van den Dool, H., Jenne, R., and Fiorino, M.: The NCEP-NCAR 50-year reanalysis: Monthly means CD-ROM and documentation, B. Am. Meteorol. Soc., 82, 247–267, 10.1175/1520-0477(2001)082<0247:Tnnyrm>2.3.Co;2, 2001.

Lespinas, F., Fortin, V., Roy, G., Rasmussen, P., and Stadnyk, T.: Performance Evaluation of the Canadian Precipitation Analysis (CaPA), J. Hydrometeorol., 16, 2045–2064, 10.1175/Jhm-D-14-0191.1, 2015.

Lucas-Picher, P., Somot, S., Deque, M., Decharme, B., and Alias, A.: Evaluation of the regional climate model ALADIN to simulate the climate over North America in the CORDEX framework, Clim. Dynam., 41, 1117–1137, 10.1007/s00382-012-1613-8, 2013.

Maggioni, V., Sapiano, M. R. P., Adler, R. F., Tian, Y. D., and Huffman, G. J.: An Error Model for Uncertainty Quantification in High-Time-Resolution Precipitation Products, J. Hydrometeorol., 15, 1274–1292, 10.1175/Jhm-D-13-0112.1, 2014.

Mahfouf, J. F., Brasnett, B., and Gagnon, S.: A Canadian precipitation analysis (CaPA) project: Description and preliminary results, Atmos. Ocean, 45, 1–17, 2007.

Marshall, I., Schut, P., and Ballard, M.: A national ecological framework for Canada: Attribute data. Ottawa, Ontario: Environmental Quality Branch, Ecosystems Science Directorate, Environment Canada and Research Branch, Agriculture and Agri-Food Canada, 1999.

Martynov, A., Laprise, R., Sushama, L., Winger, K., Separovic, L., and Dugas, B.: Reanalysis-driven climate simulation over CORDEX North America domain using the Canadian Regional Climate Model, version 5: model performance evaluation, Clim. Dynam., 41, 2973–3005, 10.1007/s00382-013-1778-9, 2013.

Maurer, E. P. and Hidalgo, H. G.: Utility of daily vs. monthly large-scale climate data: an intercomparison of two statistical downscaling methods, Hydrol. Earth Syst. Sci., 12, 551–563, 10.5194/hess-12-551-2008, 2008.

Maurer, E. P., Hidalgo, H. G., Das, T., Dettinger, M. D., and Cayan, D. R.: The utility of daily large-scale climate data in the assessment of climate change impacts on daily streamflow in California, Hydrol. Earth Syst. Sci., 14, 1125–1138, 10.5194/hess-14-1125-2010, 2010.

Mearns, L., Arritt, R., Biner, S., Bukovsky, M. S., McGinnis, S., Sain, S., Caya, D., Correia, J., Flory, D., Gutowski, W., Takle, E. S., Jones, R., Leung, R., Moufouma-Okia, W., McDaniel, L., Nunes, A. M. B., Qian, Y., Roads, J., Sloan, L., and Snyder, M.: Overview of the North American regional climate change assessment program, NOAA RISA-NCAR Meeting, Tuscon, AZ, March, 2006.

Mearns, L. O., Arritt, R., Biner, S., Bukovsky, M. S., McGinnis, S., Sain, S., Caya, D., Correia, J., Flory, D., Gutowski, W., Takle, E. S., Jones, R., Leung, R., Moufouma-Okia, W., McDaniel, L., Nunes, A. M. B., Qian, Y., Roads, J., Sloan, L., and Snyder, M.: The North American Regional Climate Change Assessment Program Overview of Phase I Results, B. Am. Meteorol. Soc., 93, 1337–1362, 2012.

Meinshausen, M., Smith, S. J., Calvin, K., Daniel, J. S., Kainuma, M. L. T., Lamarque, J. F., Matsumoto, K., Montzka, S. A., Raper, S. C. B., Riahi, K., Thomson, A., Velders, G. J. M., and van Vuuren, D. P. P.: The RCP greenhouse gas concentrations and their extensions from 1765 to 2300, Climatic Change, 109, 213–241, 10.1007/s10584-011-0156-z, 2011.

Mekis, E. and Hogg, W. D.: Rehabilitation and analysis of Canadian daily precipitation time series, Atmos. Ocean, 37, 53–85, 1999.

Mekis, E. and Vincent, L. A.: An Overview of the Second Generation Adjusted Daily Precipitation Dataset for Trend Analysis in Canada, Atmos. Ocean, 49, 163–177, 10.1080/07055900.2011.583910, 2011.

Mesinger, F., Janjic, Z. I., Nickovic, S., Gavrilov, D., and Deaven, D. G.: The Step-Mountain Coordinate – Model Description and Performance for Cases of Alpine Lee Cyclogenesis and for a Case of an Appalachian Redevelopment, Mon. Weather Rev., 116, 1493–1518, 10.1175/1520-0493(1988)116<1493:Tsmcmd>2.0.Co;2, 1988.

Mesinger, F., DiMego, G., Kalnay, E., Mitchell, K., Shafran, P. C., Ebisuzaki, W., Jovic, D., Woollen, J., Rogers, E., Berbery, E. H., Ek, M. B., Fan, Y., Grumbine, R., Higgins, W., Li, H., Lin, Y., Manikin, G., Parrish, D., and Shi, W.: North American regional reanalysis, B. Am. Meteorol. Soc., 87, 343–360, 10.1175/Bams-87-3-343, 2006.

Metcalfe, J. R., Routledge, B., and Devine, K.: Rainfall measurement in Canada: Changing observational methods and archive adjustment procedures, J. Climate, 10, 92–101, 10.1175/1520-0442(1997)010<0092:Rmicco>2.0.Co;2, 1997.

Middelkoop, H., Daamen, K., Gellens, D., Grabs, W., Kwadijk, J. C. J., Lang, H., Parmet, B. W. A. H., Schadler, B., Schulla, J., and Wilke, K.: Impact of climate change on hydrological regimes and water resources management in the rhine basin, Climatic Change, 49, 105–128, 10.1023/A:1010784727448, 2001.

Mitchell, K. E., Lohmann, D., Houser, P. R., Wood, E. F., Schaake, J. C., Robock, A., Cosgrove, B. A., Sheffield, J., Duan, Q. Y., Luo, L. F., Higgins, R. W., Pinker, R. T., Tarpley, J. D., Lettenmaier, D. P., Marshall, C. H., Entin, J. K., Pan, M., Shi, W., Koren, V., Meng, J., Ramsay, B. H., and Bailey, A. A.: The multi-institution North American Land Data Assimilation System (NLDAS): Utilizing multiple GCIP products and partners in a continental distributed hydrological modeling system, J. Geophys. Res.-Atmos., 109, D07s90, 10.1029/2003jd003823, 2004.

Nalley, D., Adamowski, J., and Khalil, B.: Using discrete wavelet transforms to analyze trends in streamflow and precipitation in Quebec and Ontario (1954–2008), J. Hydrol., 475, 204–228, 10.1016/j.jhydrol.2012.09.049, 2012.

NA-CORDEX: COordinated Regional climate Downscaling EXperiment over the North American domain, available at: https://na-cordex.org/simulations-modeling-group, last access: May 2016.

NARR: the North American Regional Reanalysis, available at: https://www.ncdc.noaa.gov/data-access/model-data/model-datasets/north-american-regional-reanalysis-narr, last access: May 2016.

Nešpor, V., Krajewski, W. F., and Kruger, A.: Wind-induced error of raindrop size distribution measurement using a two-dimensional video disdrometer, J. Atmos. Ocean Tech., 17, 1483–1492, 2000.

New, M., Hulme, M., and Jones, P.: Representing twentieth-century space-time climate variability. Part I: Development of a 1961–90 mean monthly terrestrial climatology, J. Climate, 12, 829–856, 10.1175/1520-0442(1999)012<0829:Rtcstc>2.0.Co;2, 1999.

New, M., Hulme, M., and Jones, P.: Representing twentieth-century space-time climate variability. Part II: Development of 1901–96 monthly grids of terrestrial surface climate, J. Climate, 13, 2217–2238, 10.1175/1520-0442(2000)013<2217:Rtcstc>2.0.Co;2, 2000.

Nijssen, B. and Lettenmaier, D. P.: Effect of precipitation sampling error on simulated hydrological fluxes and states: Anticipating the Global Precipitation Measurement satellites, J. Geophys. Res.-Atmos., 109, D02103, 10.1029/2003jd003497, 2004.

Pacific Climate Impacts Consortium, University of Victoria: Statistically Downscaled Climate Scenarios, available at: https://www.pacificclimate.org/data/statistically-downscaled-climate-scenarios (last access: 20 April 2016), 2014.

Park, H., Fedorov, A. N., Zheleznyak, M. N., Konstantinov, P. Y., and Walsh, J. E.: Effect of snow cover on pan-Arctic permafrost thermal regimes, Clim. Dynam., 44, 2873–2895, 10.1007/s00382-014-2356-5, 2015.

Park, H., Yoshikawa, Y., Oshima, K., Kim, Y., Thanh, N. D., Kimball, J. S., and Yang, D. Q.: Quantification of Warming Climate-Induced Changes in Terrestrial Arctic River Ice Thickness and Phenology, J. Climate, 29, 1733–1754, 10.1175/Jcli-D-15-0569.1, 2016.

PCIC: the Pacific Climate Impacts Consortium, available at: https://www.pacificclimate.org/data/statistically-downscaled-climate-scenarios, last access: April 2016.

Pearse, P. H., Bertrand, F., and MacLaren, J. W.: Currents of change; Final Report: inquiry on Federal water policy, Inquiry on Federal Water Policy, 1985.

Pietroniro, A., Fortin, V., Kouwen, N., Neal, C., Turcotte, R., Davison, B., Verseghy, D., Soulis, E. D., Caldwell, R., Evora, N., and Pellerin, P.: Development of the MESH modelling system for hydrological ensemble forecasting of the Laurentian Great Lakes at the regional scale, Hydrol. Earth Syst. Sci., 11, 1279–1294, 10.5194/hess-11-1279-2007, 2007.

100

Princeton: Global dataset at the Princeton University, available at: http://hydrology.princeton.edu/data.pgf.php, last access: March 2016.

101

Rapaic, M., Brown, R., Markovic, M., and Chaumont, D.: An Evaluation of Temperature and Precipitation Surface-Based and Reanalysis Datasets for the Canadian Arctic, 1950–2010, Atmos. Ocean, 53, 283–303, 10.1080/07055900.2015.1045825, 2015.

102

Rasmussen, R., Baker, B., Kochendorfer, J., Meyers, T., Landolt, S., Fischer, A. P., Black, J., Theriault, J. M., Kucera, P., Gochis, D., Smith, C., Nitu, R., Hall, M., Ikeda, K., and Gutmann, E.: How well are we measuring snow? The NOAA/FAA/NCAR Winter Precipitation Test Bed, B. Am. Meteorol. Soc., 93, 811–829, 10.1175/Bams-D-11-00052.1, 2012.

103

Rauscher, S. A., Coppola, E., Piani, C., and Giorgi, F.: Resolution effects on regional climate model simulations of seasonal precipitation over Europe, Clim. Dynam., 35, 685–711, 10.1007/s00382-009-0607-7, 2010.

104

Rudolf, B. and Schneider, U.: Calculation of gridded precipitation data for the global land-surface using in-situ gauge observations, Proc. Second Workshop of the Int. Precipitation Working Group, 231–247, 2005.

105

Sapiano, M. R. P. and Arkin, P. A.: An Intercomparison and Validation of High-Resolution Satellite Precipitation Estimates with 3-Hourly Gauge Data, J. Hydrometeorol., 10, 149–166, 10.1175/2008jhm1052.1, 2009.

106

Schneider, U., Fuchs, T., Meyer-Christoffer, A., and Rudolf, B.: Global precipitation analysis products of the GPCC, Global Precipitation Climatology Centre (GPCC), DWD, Internet Publikation, 112, 2008.

107

Schnorbus, M., Werner, A., and Bennett, K.: Impacts of climate change in three hydrologic regimes in British Columbia, Canada, Hydrol. Process., 28, 1170–1189, 10.1002/hyp.9661, 2014.

108

Separovic, L., Alexandru, A., Laprise, R., Martynov, A., Sushama, L., Winger, K., Tete, K., and Valin, M.: Present climate and climate change over North America as simulated by the fifth-generation Canadian regional climate model, Clim. Dynam., 41, 3167–3201, 10.1007/s00382-013-1737-5, 2013.

109

Sheffield, J., Goteti, G., and Wood, E. F.: Development of a 50-year high-resolution global dataset of meteorological forcings for land surface modeling, J. Climate, 19, 3088–3111, 10.1175/Jcli3790.1, 2006.

110

Shen, S. S. P., Dzikowski, P., Li, G. L., and Griffith, D.: Interpolation of 1961–97 daily temperature and precipitation data onto Alberta polygons of ecodistrict and soil landscapes of Canada, J. Appl. Meteorol., 40, 2162–2177, 10.1175/1520-0450(2001)040<2162:Iodtap>2.0.Co;2, 2001.

111

Shen, Y., Xiong, A. Y., Wang, Y., and Xie, P. P.: Performance of high-resolution satellite precipitation products over China, J. Geophys. Res.-Atmos., 115, D02114, 10.1029/2009jd012097, 2010.

112

Shook, K. and Pomeroy, J.: Changes in the hydrological character of rainfall on the Canadian prairies, Hydrol. Process., 26, 1752–1766, 10.1002/hyp.9383, 2012.

113

Shrestha, R., Berland, A., Schnorbus, M., and Werner, A.: Climate change impacts on hydro-climatic regimes in the Peace and Columbia watersheds, British Columbia, Canada, Pacific climate impacts consortium, University of Victoria, Victoria, 37, 2011.

114

Shrestha, R. R., Dibike, Y. B., and Prowse, T. D.: Modelling of climate-induced hydrologic changes in the Lake Winnipeg watershed, J. Great Lakes Res., 38, 83–94, 10.1016/j.jglr.2011.02.004, 2012a.

115

Shrestha, R. R., Schnorbus, M. A., Werner, A. T., and Berland, A. J.: Modelling spatial and temporal variability of hydrologic impacts of climate change in the Fraser River basin, British Columbia, Canada, Hydrol. Process., 26, 1841–1861, 10.1002/hyp.9283, 2012b.

116

Strangeways, I.: Improving precipitation measurement, Int. J. Climatol., 24, 1443–1460, 10.1002/joc.1075, 2004.

117

Tapiador, F. J., Turk, F. J., Petersen, W., Hou, A. Y., Garcia-Ortega, E., Machado, L. A. T., Angelis, C. F., Salio, P., Kidd, C., Huffman, G. J., and de Castro, M.: Global precipitation measurement: Methods, datasets and applications, Atmos. Res., 104, 70–97, 10.1016/j.atmosres.2011.10.021, 2012.

118

Taubenböck, H., Wurm, M., Netzband, M., Zwenzner, H., Roth, A., Rahman, A., and Dech, S.: Flood risks in urbanized areas – multi-sensoral approaches using remotely sensed data for risk assessment, Nat. Hazards Earth Syst. Sci., 11, 431–444, 10.5194/nhess-11-431-2011, 2011.

119

Taylor, K. E., Stouffer, R. J., and Meehl, G. A.: An Overview of Cmip5 and the Experiment Design, B. Am. Meteorol. Soc., 93, 485–498, 10.1175/Bams-D-11-00094.1, 2012.

120

Teutschbein, C. and Seibert, J.: Regional climate models for hydrological impact studies at the catchment scale: a review of recent modeling strategies, Geography Compass, 4, 834–860, 2010.

121

Tian, Y. D. and Peters-Lidard, C. D.: A global map of uncertainties in satellite-based precipitation measurements, Geophys. Res. Lett., 37, L24407, 10.1029/2010gl046008, 2010.

122

Tian, Y. D., Peters-Lidard, C. D., Eylander, J. B., Joyce, R. J., Huffman, G. J., Adler, R. F., Hsu, K. L., Turk, F. J., Garcia, M., and Zeng, J.: Component analysis of errors in satellite-based precipitation estimates, J. Geophys. Res.-Atmos., 114, D24101, 10.1029/2009jd011949, 2009.

123

Turk, F. J., Arkin, P., Ebert, E. E., and Sapiano, M. R. P.: Evaluating High-Resolution Precipitation Products, B. Am. Meteorol. Soc., 89, 1911–1916, 10.1175/2008bams2652.1, 2008.

124

Uppala, S. M., Kållberg, P., Simmons, A., Andrae, U., Bechtold, V. d., Fiorino, M., Gibson, J., Haseler, J., Hernandez, A., and Kelly, G.: The ERA-40 re-analysis, Q. J. Roy. Meteor. Soc., 131, 2961–3012, 2005.

125

Vila, D. A., de Goncalves, L. G. G., Toll, D. L., and Rozante, J. R.: Statistical Evaluation of Combined Daily Gauge Observations and Rainfall Satellite Estimates over Continental South America, J. Hydrometeorol., 10, 533–543, 10.1175/2008jhm1048.1, 2009.

126

Villarini, G. and Krajewski, W. F.: Review of the Different Sources of Uncertainty in Single Polarization Radar-Based Estimates of Rainfall, Surv. Geophys., 31, 107–129, 10.1007/s10712-009-9079-x, 2010.

127

Wan, H., Zhang, X. B., Zwiers, F. W., and Shiogama, H.: Effect of data coverage on the estimation of mean and variability of precipitation at global and regional scales, J. Geophys. Res.-Atmos., 118, 534–546, 10.1002/jgrd.50118, 2013.

128

Wang, S., Yang, Y., Luo, Y., and Rivera, A.: Spatial and seasonal variations in evapotranspiration over Canada's landmass, Hydrol. Earth Syst. Sci., 17, 3561–3575, 10.5194/hess-17-3561-2013, 2013.

129

Wang, S. S., Huang, J. L., Li, J. H., Rivera, A., McKenney, D. W., and Sheffield, J.: Assessment of water budget for sixteen large drainage basins in Canada, J. Hydrol., 512, 1–15, 10.1016/j.jhydrol.2014.02.058, 2014.

130

Wang, X. L. L. and Lin, A.: An algorithm for integrating satellite precipitation estimates with in situ precipitation data on a pentad time scale, J. Geophys. Res.-Atmos., 120, 3728–3744, 10.1002/2014jd022788, 2015.

131

Weedon, G. P., Gomes, S., Viterbo, P., Shuttleworth, W. J., Blyth, E., Osterle, H., Adam, J. C., Bellouin, N., Boucher, O., and Best, M.: Creation of the WATCH Forcing Data and Its Use to Assess Global and Regional Reference Crop Evaporation over Land during the Twentieth Century, J. Hydrometeorol., 12, 823–848, 10.1175/2011jhm1369.1, 2011.

132

Weedon, G. P., Balsamo, G., Bellouin, N., Gomes, S., Best, M. J., and Viterbo, P.: The WFDEI meteorological forcing data set: WATCH Forcing Data methodology applied to ERA-Interim reanalysis data, Water. Resour. Res., 50, 7505–7514, 10.1002/2014wr015638, 2014.

133

Werner, A. T. and Cannon, A. J.: Hydrologic extremes – an intercomparison of multiple gridded statistical downscaling methods, Hydrol. Earth Syst. Sci., 20, 1483–1508, 10.5194/hess-20-1483-2016, 2016.

134

West, G. L., Steenburgh, W. J., and Cheng, W. Y. Y.: Spurious grid-scale precipitation in the North American regional reanalysis, Mon. Weather Rev., 135, 2168–2184, 10.1175/Mwr3375.1, 2007.

135

Wetterhall, F., Bardossy, A., Chen, D. L., Halldin, S., and Xu, C. Y.: Daily precipitation-downscaling techniques in three Chinese regions, Water Resour. Res., 42, W11423, 10.1029/2005wr004573, 2006.

136

WFDEI: the WATCH Forcing Data methodology applied to the ERA-Interim, available at: http://www.eu-watch.org/data_availability, last access: March 2016.

137

Willmott, C. J., Matsuura, K., and Legates, D.: Terrestrial air temperature and precipitation: monthly and annual time series (1950–1999), Center for climate research version, 1, 2001.

138

Woo, M. K. and Thorne, R.: Snowmelt contribution to discharge from a large mountainous catchment in subarctic Canada, Hydrol. Process., 20, 2129–2139, 10.1002/hyp.6205, 2006.

139

Wood, A. W., Leung, L. R., Sridhar, V., and Lettenmaier, D. P.: Hydrologic implications of dynamical and statistical approaches to downscaling climate model outputs, Climatic Change, 62, 189–216, 10.1023/B:CLIM.0000013685.99609.9e, 2004.

140

Xie, P. P. and Arkin, P. A.: An Intercomparison of Gauge Observations and Satellite Estimates of Monthly Precipitation, J. Appl. Meteorol., 34, 1143–1160, 10.1175/1520-0450(1995)034<1143:Aiogoa>2.0.Co;2, 1995.

141

Xie, P. P. and Arkin, P. A.: Global monthly precipitation: An intercomparison of several datasets based on gauge observations, satellite estimates and model predictions, Eighth Conference on Satellite Meteorology and Oceanography, 225–229, 1996.

142

Xu, C. Y., Widen, E., and Halldin, S.: Modelling hydrological consequences of climate change – Progress and challenges, Adv. Atmos. Sci., 22, 789–797, 10.1007/Bf02918679, 2005.

143

Yang, D. Q., Goodison, B. E., Metcalfe, J. R., Golubev, V. S., Bates, R., Pangburn, T., and Hanson, C. L.: Accuracy of NWS 8” standard nonrecording precipitation gauge: Results and application of WMO intercomparison, J. Atmos. Ocean. Tech., 15, 54–68, 10.1175/1520-0426(1998)015<0054:Aonsnp>2.0.Co;2, 1998.

144

Young, C. B., Nelson, B. R., Bradley, A. A., Smith, J. A., Peters-Lidard, C. D., Kruger, A., and Baeck, M. L.: An evaluation of NEXRAD precipitation estimates in complex terrain, J. Geophys. Res.-Atmos., 104, 19691–19703, 10.1029/1999jd900123, 1999.

145

Zhang, Q., Sun, P., Singh, V. P., and Chen, X. H.: Spatial-temporal precipitation changes (1956–2000) and their implications for agriculture in China, Global Planet. Change, 82–83, 86–95, 10.1016/j.gloplacha.2011.12.001, 2012.

</app></app-group></back> </article>