A new river discharge and river temperature climatology data set for the pan-Arctic region

Most regional ocean models that use discharge as part of the forcing use relatively coarse river discharge data sets (1 degrees, or similar to 110 km) compared to the model resolution (typically 1/4 degrees or less), and do not account for seasonal changes in river water temperature. We introduce a new climatological data set of river discharge and river water temperature with 1/6 degrees grid spacing over the Arctic region (Arctic River Discharge and Temperature; ARDAT), incorporating observations from 30 Arctic rivers. The annual mean discharge for all rivers in ARDAT is 2817 +/- 330 km(3) yr(-1). River water temperatures range between 0 degrees C in winter to 14.0-17.6 degrees C in July, leading to a long-term mean monthly heat flux from all rivers of 3.2 +/- 0.6 TW, of which 31% is supplied by Alaskan rivers and 69% is supplied by Eurasian rivers. This riverine heat flux is equivalent to 44% of the estimated ocean heat flux associated with the Bering Strait throughflow, but during the spring freshet can be similar to 10 times as large, suggesting that heat flux associated with Arctic rivers is an important component of the Arctic heat budget on seasonal time scales. We apply the ARDAT data set to a high-resolution regional ocean-ice model, and compare results to a model integration using a 1 degrees resolution discharge data set. Integrated freshwater content on the Arctic shelves (<200 m) increases by similar to 3600 km(3) in the ARDAT forced model run compared to the coarser forcing, suggesting that river discharge is contained on the Arctic shelves when forced with the ARDAT data set. Modelled summer heat fluxes over the shelves increase by 8 TW when river water temperature is included, which subsequently reduces basin-wide September sea ice extent by similar to 10%. Regional differences are larger, where e.g., sea ice extent on the Beaufort shelf is reduced by similar to 36%. Using a non-linear free surface parameterization along with the ARDAT data set, we find an increase in the sea surface height gradient around river mouths. Geostrophic velocities increase to form quasi-continuous, fast-moving near-shore boundary currents not reproduced using the 1 degrees resolution data set. Omitting river water temperature, or using a lower resolution data set, can therefore lead to incorrect model estimates of coastal transport, sea ice formation/melt rates, and other regional and basin scale processes. Using a high-resolution discharge data set, and accounting for the considerable heat carried by the Arctic rivers is recommended for future modelling efforts. (C) 2015 Elsevier Ltd. All rights reserved.