Canadian Seasonal To Interannual Prediction System Research Articles

Improved Seasonal Forecast Skill of Pan-Arctic and Regional Sea Ice Extent in CanSIPS Version 2

Abstract This study assesses the forecast skill of the Canadian Seasonal to Interannual Prediction System (CanSIPS), version 2, in predicting Arctic sea ice extent on both the pan-Arctic and regional scales. In addition, the forecast skill is compared to that of CanSIPS, version 1. Overall, there is a net increase of forecast skill when considering detrended data due to the changes made in the development of CanSIPSv2. The most notable improvements are for forecasts of late summer and autumn target months that have been initialized in the months of April and May that, in previous studies, have been associated with the spring predictability barrier. By comparison of the skills of CanSIPSv1 and CanSIPSv2 to that of an intermediate version of CanSIPS, CanSIPSv1b, we can attribute skill differences between CanSIPSv1 and CanSIPSv2 to two main sources. First, an improved initialization procedure for sea ice initial conditions markedly improves forecast skill on the pan-Arctic scale as well as regionally in the central Arctic, Laptev Sea, Sea of Okhotsk, and Barents Sea. This conclusion is further supported by analysis of the predictive skill of the sea ice volume initialization field. Second, the change in model combination from CanSIPSv1 to CanSIPSv2 (exchanging the constituent CanCM3 model for GEM-NEMO) improves forecast skill in the Bering, Kara, Chukchi, Beaufort, East Siberian, Barents, and the Greenland–Iceland–Norwegian (GIN) Seas. In Hudson and Baffin Bay, as well as the Labrador Sea, there is limited and unsystematic improvement in forecasts of CanSIPSv2 as compared to CanSIPSv1.

Development and Calibration of Seasonal Probabilistic Forecasts of Ice-Free Dates and Freeze-Up Dates

AbstractDynamical forecasting systems are being used to skillfully predict deterministic ice-free and freeze-up date events in the Arctic. This paper extends such forecasts to a probabilistic framework and tests two calibration models to correct systematic biases and improve the statistical reliability of the event dates: trend-adjusted quantile mapping (TAQM) and nonhomogeneous censored Gaussian regression (NCGR). TAQM is a probability distribution mapping method that corrects the forecast for climatological biases, whereas NCGR relates the calibrated parametric forecast distribution to the raw ensemble forecast through a regression model framework. For NCGR, the observed event trend and ensemble-mean event date are used to predict the central tendency of the predictive distribution. For modeling forecast uncertainty, we find that the ensemble-mean event date, which is related to forecast lead time, performs better than the ensemble variance itself. Using a multidecadal hindcast record from the Canadian Seasonal to Interannual Prediction System (CanSIPS), TAQM and NCGR are applied to produce categorical forecasts quantifying the probabilities for early, normal, and late ice retreat and advance. While TAQM performs better than adjusting the raw forecast for mean and linear trend bias, NCGR is shown to outperform TAQM in terms of reliability, skill, and an improved tendency for forecast probabilities to be no worse than climatology. Testing various cross-validation setups, we find that NCGR remains useful when shorter hindcast records (~20 years) are available. By applying NCGR to operational forecasts, stakeholders can be more confident in using seasonal forecasts of sea ice event timing for planning purposes.

Initialization and Potential Predictability of Soil Moisture in the Canadian Seasonal to Interannual Prediction System

The initialization and potential predictability of soil moisture in CanCM4 hindcasts during 1981–2010 is assessed. CanCM4 is one of the two global climate models employed by the Canadian Seasonal to Interannual Prediction System (CanSIPS) providing operational multiseasonal forecasts for Environment and Climate Change Canada (ECCC). Soil moisture forecast initialization in CanSIPS is determined by the response of the land component to forcing from data-constrained model atmospheric fields. We evaluate hindcast initial conditions for soil moisture and its atmospheric forcings against observation-based datasets. Although model values of soil moisture variability compare relatively well with a blend of two reanalysis products, there is significant disagreement in the tropics and arid regions linked to biases in precipitation, as well as in snow-covered regions, likely the result of biases in the timing of snow onset and melt. The temporal variance of initial soil moisture anomalies is typically larger in regions of considerable precipitation variability and in cold continental areas of shallow soil depth. Appreciable variance of initial conditions, combined with persistence of the initial anomalies and the model’s ability to represent future climate variations, lead to potentially predictable soil moisture variance exceeding 60% of the total variance for up to 3–4 months in the tropics and 6–7 months in the mid- to high latitudes during hemispheric winter. Potential predictability at longer leads is primarily found in the tropics and extratropical areas of ENSO-teleconnected influences. We use lagged partial correlations to show that ENSO-teleconnected precipitation in CanCM4 is a likely source of potential predictability of soil moisture up to 1-yr lead in CanSIPS hindcasts.

Canadian snow and sea ice: assessment of snow, sea ice, and related climate processes in Canada's Earth system model and climate-prediction system

Abstract. The Canadian Sea Ice and Snow Evolution (CanSISE) Network is a climate research network focused on developing and applying state-of-the-art observational data to advance dynamical prediction, projections, and understanding of seasonal snow cover and sea ice in Canada and the circumpolar Arctic. This study presents an assessment from the CanSISE Network of the ability of the second-generation Canadian Earth System Model (CanESM2) and the Canadian Seasonal to Interannual Prediction System (CanSIPS) to simulate and predict snow and sea ice from seasonal to multi-decadal timescales, with a focus on the Canadian sector. To account for observational uncertainty, model structural uncertainty, and internal climate variability, the analysis uses multi-source observations, multiple Earth system models (ESMs) in Phase 5 of the Coupled Model Intercomparison Project (CMIP5), and large initial-condition ensembles of CanESM2 and other models. It is found that the ability of the CanESM2 simulation to capture snow-related climate parameters, such as cold-region surface temperature and precipitation, lies within the range of currently available international models. Accounting for the considerable disagreement among satellite-era observational datasets on the distribution of snow water equivalent, CanESM2 has too much springtime snow mass over Canada, reflecting a broader northern hemispheric positive bias. Biases in seasonal snow cover extent are generally less pronounced. CanESM2 also exhibits retreat of springtime snow generally greater than observational estimates, after accounting for observational uncertainty and internal variability. Sea ice is biased low in the Canadian Arctic, which makes it difficult to assess the realism of long-term sea ice trends there. The strengths and weaknesses of the modelling system need to be understood as a practical tradeoff: the Canadian models are relatively inexpensive computationally because of their moderate resolution, thus enabling their use in operational seasonal prediction and for generating large ensembles of multidecadal simulations. Improvements in climate-prediction systems like CanSIPS rely not just on simulation quality but also on using novel observational constraints and the ready transfer of research to an operational setting. Improvements in seasonal forecasting practice arising from recent research include accurate initialization of snow and frozen soil, accounting for observational uncertainty in forecast verification, and sea ice thickness initialization using statistical predictors available in real time.

Representation of Snow in the Canadian Seasonal to Interannual Prediction System. Part II: Potential Predictability and Hindcast Skill

Abstract This paper examines potential predictability (PP) and actual skill for snow water equivalent (SWE) in the Canadian Seasonal to Interannual Prediction System (CanSIPS). A significant PP is found for SWE, with potentially predictable variance over 50% of the total variance at up to a 5-month lead in mid- to high latitudes in forecasts initialized after snow onset. Much, though not all, of this PP stems from a tendency for SWE anomalies to persist through the snow season. Although the spring melt acts as a PP barrier regardless of initialization date, in some regions significant PP reemerges in the following snow season. This is due primarily to ENSO teleconnections that are modeled realistically by CanSIPS, particularly in northwestern North America. Actual skill of CanSIPS in forecasting SWE is assessed using several verification datasets. Highest skills are obtained using a blend of five such datasets, consistent with the hypothesis that skill scores are degraded by errors in the verification data as well as by forecast errors, and that observational errors can be reduced by blending multiple datasets, much as forecast errors can be reduced by averaging across different models. Actual skill for SWE is comparable to, though generally lower than, that implied by PP. This is due in part to the similar autocorrelation properties of the forecast and observed SWE anomalies, which provide skill through anomaly persistence, combined with reasonably accurate initialization of SWE by CanSIPS. Long-lead skill across snow seasons is found to be linked to ENSO, particularly in western North America, much as for PP.

Representation of Snow in the Canadian Seasonal to Interannual Prediction System. Part I: Initialization

Abstract The ability of the Canadian Seasonal to Interannual Prediction System (CanSIPS) to provide realistic forecast initial conditions for snow cover is assessed using in situ measurements and gridded snow analyses. Forecast initial conditions for snow in CanCM3 and CanCM4 employed by CanSIPS are determined by the response of the Canadian Land Surface Scheme (CLASS) used in both models to forcing from model atmospheric fields constrained by assimilation of 6-hourly reanalysis data. These snow initial conditions are found to be representative of the daily climatology of snow water equivalent (SWE) as well as interannual variations in maximum SWE and the timing of snow onset and snowmelt observed at eight in situ measurement sites located across Canada. The level of this agreement is similar to that of three independent gridded snow analyses (MERRA, the European Space Agency’s GlobSnow, and an offline forced version of CLASS). Total Northern Hemisphere snow mass generated by the CanSIPS initialization procedure is larger for both models (especially CanCM3) than in MERRA, mostly because of higher SWE in regions of common snow cover. Globally, the interannual variability of initial SWE is found to correlate highly with that of MERRA in locations with appreciable snow. These initial values are compared to SWE in freely running CanCM3 and CanCM4 simulations produced without data assimilation of atmospheric fields. Differences in climatological SWE relative to MERRA are similar in the freely running and assimilating CanCM3 and CanCM4 simulations, suggesting that inherent model biases are a major contributor to biases in CanSIPS snow initial conditions.

Real‐time estimation of Arctic sea ice thickness through maximum covariance analysis

AbstractA challenge for model‐based seasonal predictions of sea ice is an accurate representation of sea ice initial conditions, particularly sparsely observed sea ice thickness (SIT). The Canadian Seasonal to Interannual Prediction System (CanSIPS) currently initializes SIT by nudging simulated values toward a model‐based climatology. To improve on this, we use sea ice data from Pan‐Arctic Ice Ocean Modeling and Assimilation System to investigate how accurately SIT can be estimated in real time using better observed and physically relevant predictors. We (1) test the skill of several predictors using maximum covariance analysis (MCA), (2) apply an approach which blends sea ice concentration and lagged (4 month averaged) sea level pressure, and (3) compare this method against the current CanSIPS initialization scheme over 1981–2012. The MCA‐based statistical model reduces SIT areal mean and temporal mean absolute errors by 48% relative to the current CanSIPS initialization and shows consistent skill estimating ice volume in all months (r = 0.95).

A Dynamical Climate Model–Driven Hydrologic Prediction System for the Fraser River, Canada

Abstract Recent improvements in forecast skill of the climate system by dynamical climate models could lead to improvements in seasonal streamflow predictions. This study evaluates the hydrologic prediction skill of a dynamical climate model–driven hydrologic prediction system (CM-HPS), based on an ensemble of statistically downscaled outputs from the Canadian Seasonal to Interannual Prediction System (CanSIPS). For comparison, historical and future climate traces–driven ensemble streamflow prediction (ESP) was employed. The Variable Infiltration Capacity model (VIC) hydrologic model setup for the Fraser River basin, British Columbia, Canada, was used as a test bed for the two systems. In both cases, results revealed limited precipitation prediction skill. For streamflow prediction, the ESP approach has very limited or no correlation skill beyond the months influenced by initial hydrologic conditions, while the CM-HPS has moderately better correlation skill, attributable to the enhanced temperature prediction skill that results from CanSIPS’s ability to predict El Niño–Southern Oscillation (ENSO) and its teleconnections. The root-mean-square error, bias, and categorical skills for the two methods are mostly similar. Hydrologic modeling uncertainty also affects the prediction skill, and in some cases prediction skill is constrained by hydrologic model skill. Overall, the CM-HPS shows potential for seasonal streamflow prediction, and further enhancements in climate models could potentially to lead to more skillful hydrologic predictions.

Skill evaluation of the CanCM4 and its MOS for seasonal rainfall forecast in Malaysia during the early and late winter monsoon periods

This study evaluates the forecast skill of the fourth version of the Canadian coupled ocean–atmosphere general circulation model (CanCM4) and its model output statistics (MOS) to forecast the seasonal rainfall in Malaysia, particularly during early (October–November–December) and late (January–February–March) winter monsoon periods. CanCM4 is the latest component of the Canadian Seasonal to Inter-annual Prediction System (CanSIPS), which is a multi-seasonal climate prediction system developed particularly for Canada but applicable globally. Generally, CanCM4's skill in reproducing the climatology during winter is not as good as in other seasons because of the model's inability to simulate the regional synoptic circulations over the western Maritime Continent. In particular, the model fails to forecast the cold surges and Borneo vortex circulations that play critical roles in moisture horizontal advection. Moreover, its forecast skill during the early winter monsoon period is poorer than during the late period. Interestingly, forecast skill is enhanced when MOS models are applied as the MOS utilizes the predictive signals in the quasi-global predictors from the CanCM4 forecast system. The predictability can be traced to the conventional El Nino–Southern Oscillation (ENSO) and ENSO Modoki signals that are present in the CanCM4 forecast MOS predictor fields. The quasi-global sea-surface temperature and quasi-global sea-level pressure fields are found to be the most useful predictors. Interestingly, CanCM4 forecast signals associated with the Indian Ocean Dipole also contribute to the skill. Skill enhancement is particularly significant for northern Borneo during early monsoon periods in medium- and long-lead forecasts when the CanCM4 has minimal direct skill in the region.

The Canadian Seasonal to Interannual Prediction System. Part I: Models and Initialization

Abstract The Canadian Seasonal to Interannual Prediction System (CanSIPS) became operational at Environment Canada's Canadian Meteorological Centre (CMC) in December 2011, replacing CMC's previous two-tier system. CanSIPS is a two-model forecasting system that combines ensemble forecasts from the Canadian Centre for Climate Modeling and Analysis (CCCma) Coupled Climate Model, versions 3 and 4 (CanCM3 and CanCM4, respectively). Mean climate as well as climate trends and variability in these models are evaluated in freely running historical simulations. Initial conditions for CanSIPS forecasts are obtained from an ensemble of coupled assimilation runs. These runs assimilate gridded atmospheric analyses by means of a procedure that resembles the incremental analysis update technique, but introduces only a fraction of the analysis increment in order that differences between ensemble members reflect the magnitude of observational uncertainties. The land surface is initialized through its response to the assimilative meteorology, whereas sea ice concentration and sea surface temperature are relaxed toward gridded observational values. The subsurface ocean is initialized through surface forcing provided by the assimilation run, together with an offline variational assimilation of gridded observational temperatures followed by an adjustment of the salinity field to preserve static stability. The performance of CanSIPS historical forecasts initialized every month over the period 1981–2010 is documented in a companion paper. The CanCM4 model and the initialization procedures developed for CanSIPS have been employed as well for decadal forecasts, including those contributing to phase 5 of the Coupled Model Intercomparison Project.

Multi‐system seasonal predictions of Arctic sea ice

The utility of multi‐system, coupled model‐based seasonal predictions of Arctic sea ice area and extent is investigated for combined predictions from the Climate Forecast System version 2 (CFSv2) and Canadian Seasonal to Interannual Prediction System (CanSIPS) operational seasonal forecasting systems, which are among the first to have sea ice as a prognostic variable. Forecast skills for predictions of total anomalies and departures from long‐term linear trends are examined both for the individual systems and the combined forecasts, and are compared against simple predictions such as damped anomaly persistence. Results indicate that the tendency for climate forecasts based on combined output from multiple prediction systems to outperform any one system, demonstrated previously for global variables such as temperature and precipitation, is realized for predictions of Arctic sea ice as well.

Seasonal Prediction of Killing-Frost Frequency in South-Central Canada during the Cool/Overwintering-Crop Growing Season

AbstractSeasonal killing-frost frequency (KFF) during the cool/overwintering-crop growing season is important for the Canadian agricultural sector to prepare and respond to such extreme agrometeorological events. On the basis of observed daily surface air temperature across Canada for 1957–2007, this study found that more than 86% of the total killing-frost events occur in April–May and exhibit consistent variability over south-central Canada, the country’s major agricultural region. To quantify the KFF year-to-year variations, a simple index is defined as the mean KFF of the 187 temperature stations in south-central Canada. The KFF variability is basically dominated by two components: the decadal component with a peak periodicity around 11 yr and the interannual component of 2.5–3.8 yr. A statistical method called partial least squares (PLS) regression is utilized to uncover principal sea surface temperature (SST) modes in the winter preceding the KFF anomalies. It is found that most of the leading SST modes resemble patterns of El Niño–Southern Oscillation (ENSO) and/or the Pacific decadal oscillation (PDO). This indicates that ENSO and the PDO might be two dominant factors for the KFF variability. From a 41-yr training period (1957–97), a PLS seasonal prediction model is established, and 1-month-lead real-time forecasts are performed for the validation period of 1998–2007. A promising skill level is obtained. For the KFF variability, the prediction skill of the PLS model is comparable to or even better than the newly developed Canadian Seasonal to Interannual Prediction System (CanSIPS), which is a state-of-the-art global coupled dynamical system.