This study presents a comprehensive evaluation of snow albedo feedback (SAF) in two generations of climate models (Coupled Model Intercomparison Project versions 3 (CMIP3) and 5 (CMIP5)). A comparison of the models is performed against a multiobservation‐based reference data set (mOBS) derived from the seasonal cycle of albedo, snow cover, and temperature. The observed total SAF shows low uncertainty and is generally well simulated by the CMIP3 and CMIP5 ensemble mean, except for a low (high) bias over the Arctic (northern boreal forest). Most CMIP5 models overestimate the snow cover component of SAF (SNC) and underestimate the temperature sensitivity component (TEM). The high bias in SNC is due to simulated snow albedos 4–5% brighter than observed driving unrealistically large albedo contrasts. However, overall representation of surface albedo—and mean climate—has improved, as fewer CMIP5 models exhibit large cold temperature, or high snow, biases. The low bias in TEM is related to overly persistent snow albedo during spring, particularly over southern Eurasia and North America. There is large observational uncertainty in the reference data set mOBS that is traced primarily to the different snow cover products, with a secondary contribution from the albedo products and a small contribution from the temperature products. The conclusion is that the model mean tends to simulate the multiobservation mean very closely; however, this masks considerable spread in both models and observations. There is clear motivation for producing improved submonthly snow cover products for the purpose of model evaluation.

Summer near‐surface seawater sampling in the Canadian Arctic revealed potential for significant errors (nearly 0.1 μmol·(m‐2 s‐1)) in CO2 fluxes calculated from measured air‐sea CO2 gradients. River runoff and sea ice melt strongly stratify these waters, often resulting in surface mixed layers only a few meters thick and isolated from waters sampled by shipboard underway systems. Samples collected with the underway system, rosette, and small boats exposed substantial near‐surface gradients in CO2 partial pressure (pCO2) over the top 7 m at many stations. Distributions of temperature, salinity, and fluorescence indicated that the sources of the CO2 system gradients varied between stations, precluding simple corrections to align subsurface data with shallower conditions. Overall, the strong summertime sink of atmospheric CO2 implied by the underway data was not supported by shallower data.

Based on a 2 year data set, the eastern Canadian Arctic Archipelago and Baffin Bay appear to be a modest summertime sink of atmospheric CO2. We measured surface water CO2 partial pressure (pCO2), salinity, and temperature throughout northern Baffin Bay, Nares Strait, and Lancaster Sound from the CCGS Amundsen during its 2013 and 2014 summer cruises. Surface water pCO2 displayed considerable variability (144–364 μatm) but never exceeded atmospheric concentrations, and average calculated CO2 fluxes in 2013 and 2014 were −12 and −3 mmol C m−2 d−1 (into the ocean), respectively. Ancillary measurements of chlorophyll a reveal low summertime productivity in surface waters. Based on total alkalinity and stable oxygen isotopes (δ18O) data, a strong riverine signal in northern Nares Strait coincided with relatively high surface pCO2, whereas areas of sea‐ice melt occur with low surface pCO2. Further assessments, extending the seasonal observation period, are needed to properly constrain both seasonal and annual CO2 fluxes in this region.

This study provides 6 years of high‐resolution underway measurements of the sea surface partial pressure of CO2 (pCO2sw), sea surface temperature, and salinity across the Canadian Arctic Archipelago (CAA). Observed pCO2sw varied regionally, with the northern and central channels of the CAA undersaturated in pCO2sw (with respect to the atmosphere), while the western regions were typically saturated to supersaturated in pCO2sw. This apparent spatial variability was caused to some extent by the timing of our ship transit through the CAA, as we also found a general seasonal trend of pCO2sw being undersaturated in the early summer, followed by saturation to supersaturation in late summer, and a return to undersaturation during the autumn. Sea surface temperature was significantly correlated with pCO2sw at various locations across the CAA, but we also observed the effects of other regional processes like upwelling, primary production, riverine input, and sea ice melt. These processes are linked to each other, and hence, it is impossible to pinpoint only one dominant factor controlling pCO2sw variability in the CAA. However, we found that sea ice dominates the seasonal cycle of all these processes, thus making the timing of sea ice breakup a useful predictor of pCO2sw variability in the CAA. We calculated an average net oceanic sink of 14 mmol CO2 · m−2 · day−1 for the CAA during the summer and autumn seasons, but caution that a more rigorous budgeting approach is required to fully account for biases in dates and locations of our measurements.

Measurements of net community production (NCP) provide an upper constraint on the strength of the oceanic biological pump, the dominant mechanism for removing CO2 from the ocean surface and sequestering it at depth. In this investigation, our objectives were to describe the spatial and temporal variability of NCP associated with the spring ice-edge bloom in Baffin Bay and to identify the key environmental drivers controlling its variability. Using data collected between June 9 and July 10, 2016, we estimated NCP based on (1) underway measurements of surface water oxygen to argon ratios (O2:Ar), (2) underway measurements of the partial pressure of CO2, and (3) seasonal nitrate drawdown from discrete samples. These multiple approaches displayed high NCP (up to 5.7 mol C m–2) in eastern Baffin Bay, associated with modified Atlantic waters, and low NCP (<1 mol C m–2) in the presence of Arctic outflow waters in western Baffin Bay. Arctic outflow waters were characterized by low surface salinities and nitrate concentrations, suggesting that high freshwater content may have limited the nutrient availability of these waters. Different integration depths and timescales associated with each NCP approach were exploited to understand the temporal progression and succession of the bloom, revealing that the bloom was initiated under ice up to 15 days prior to ice retreat and that a large portion of NCP in eastern Baffin Bay (potentially up to 70%) was driven by primary production occurring below the surface-mixed layer.