Diagnostic evaluation of river discharge into the Arctic Ocean and its impact on oceanic volume transports

Abstract

Abstract. This study analyses river discharge into the Arctic Ocean using state-of-the-art reanalyses such as the fifth-generation European Reanalysis (ERA5) and the reanalysis component from the Global Flood Awareness System (GloFAS). GloFAS, in its operational version 2.1, combines the land surface model (Hydrology Tiled European Centre for Medium-Range Weather Forecasts – ECMWF – Scheme for Surface Exchanges over Land, HTESSEL) from ECMWF’s ERA5 with a hydrological and channel routing model (LISFLOOD). Furthermore, we analyse GloFAS' most recent version 3.1, which is not coupled to HTESSEL but uses the full configuration of LISFLOOD. Seasonal cycles as well as annual runoff trends are analysed for the major Arctic watersheds – Yenisei, Ob, Lena, and Mackenzie – where reanalysis-based runoff can be compared to available observed river discharge records. Furthermore, we calculate river discharge over the whole pan-Arctic region and, by combination with atmospheric inputs, storage changes from the Gravity Recovery and Climate Experiment (GRACE) and oceanic volume transports from ocean reanalyses, we assess closure of the non-steric water volume budget. Finally, we provide best estimates for every budget equation term using a variational adjustment scheme. Runoff from ERA5 and GloFAS v2.1 features pronounced declining trends induced by two temporal inhomogeneities in ERA5's data assimilation system, and seasonal river discharge peaks are underestimated by up to 50 % compared to observations. The new GloFAS v3.1 product exhibits distinct improvements and performs best in terms of seasonality and long-term means; however, in contrast to gauge observations, it also features declining runoff trends. Calculating runoff indirectly through the divergence of moisture flux is the only reanalysis-based estimate that is able to reproduce the river discharge increases measured by gauge observations (pan-Arctic increase of 2 % per decade). In addition, we examine Greenlandic discharge, which contributes about 10 % of the total pan-Arctic discharge and features strong increases mainly due to glacial melting. The variational adjustment yields reliable estimates of the volume budget terms on an annual scale, requiring only moderate adjustments of less than 3 % for each individual term. Approximately 6583±84 km3 of freshwater leaves the Arctic Ocean per year through its boundaries. About two-thirds of this is contributed by runoff from the surrounding land areas to the Arctic Ocean (4379±25 km3 yr−1), and about one-third is supplied by the atmosphere. However, on a seasonal scale budget residuals of some calendar months were too large to be eliminated within the a priori spreads of the individual terms. This suggests that systematical errors are present in the reanalyses and ocean reanalysis data sets, which are not considered in our uncertainty estimation.