Time series of annual water discharge
Discharge Regime of Arctic Rivers
The rivers that discharge to the Arctic Ocean freeze during the winter months and during that time the discharge of the large rivers is low. Some small rivers freeze completely. In April the ice generally starts to melt and in May or June the peak discharge associated with the melting of the snow and ice in the catchments occurs. This spring discharge peak is called ‘freshet’. After this peak the discharge gradually decreases. This discharge regime has important implications for the discharge of sediment and nutrients. The bulk of these discharges occur in approximately 5 months. The typical shape of the hydrograph is illustrated by the average yearly discharge hydrograph for the 6 largest Arctic Rivers plus the Kuskokwim in the figure below (Whitefield et al. 2015).
Fig. 2. Climatologies for (a) integrated discharge (+- 1 standard deviation) for all rivers in ARDAT, (b) mean river temperature (+-1 standard deviation) for the "Big 6" Arctic rivers and one sub-Arctic river (Kuskokwim), (c) calculated heat flux (+-1 standard deviation) using integrated discharge from all rivers in ARDAT and mean river water temperatures. Numbers on top axes of (a) and (b) denote total number of monthly-mean observations for each month.
Time series of yearly total discharge into the Arctic Ocean
The challenge set by DG-mare is to provide time series of the annual mass of water discharged to the Arctic ocean. These time series have been created by integrating the monthly averaged discharge over a year. The stations marked as ‘downstream station’ for all rivers in the ArcticHycos dataset have been processed. The results can be found on the map viewer. The usual unit to report river discharge is a volume per unit time, this has been reported instead of the annual mass of water. The mapviewer provides access to the time series by clicking on the ‘i’ button and the measurement station in the map. The time series are presented as an average discharge in m3/s and km3/y.
The ArcticHycos dataset is the most complete dataset in terms of spatial and temporal coverage and other datasets only contain copies of a subset of this data. Therefore no additional data from other datasets has been used. The datasets for the large Russian rivers typically start early 20th century or even late 19th century. This makes these rivers interesting for long-term trend detection in climate studies. However, the recent years discharges are missing, observations stop approximately after 2010. This data will probably be delivered by the SHI in the future (Looser BfG, pers. comm. 2016). The record typically starts in the 60’s and 70’s for the North American rivers and recent observations are available.
The time series of the monthly data contain data gaps. Averaging over an incomplete year will result in a deviation of the average temperature from the true average temperature. These defects have not been repaired in this project. For the most significant rivers the time series are also presented below. The years with incomplete time series have been indicated with a lighter blue color in the graphs below. The grey lines show the monthly time series.
Variability and change in discharge
The time series show a considerable variability in discharge from year to year. An obvious question in the context of climate change is whether trends can be observed in the discharge. These trends are obscured by the variability therefore a statistical analysis of the data is required. Déry et al. (2016) conducted such an analysis for rivers in Northern Canada. They find that for many rivers no significant trends can be detected, and when a trend is detected it sometimes is anthropogenic (e.g. a river diversion). They do find a significant increase for the total discharge from the Canadian rivers, an 18% increase between 1989 and 2013. This is in line with the work of Zhang et al. (2012) who analyzed the discharge from the 3 largest Eurasian Rivers (Ob, Yenisei and Lena) see figure below. These papers show that trends can be detected using the current data sets.
Figure 2 Year-by-year annual net AMT and river discharg. Annual net AMT converged into the Ob, Yenisei and Lena river basins (red solid line) and annual discharge from these three rivers (blue solid line) from 1948 to 2008. The five-year running means were applied to detect QDV of net AMT (red dashed line) and discharge (blue dashed line). The linear trends are derived from the linear regression.
The Hudson bay receives considerable freshwater input from rivers. Time series for these discharges are available from the Wateroffice Canada. These discharges are not considered to be freshwater input for the Arctic Ocean. Whitefield at al. (2015) argue as follows: While discharge data are available for many Canadian rivers in the Hudson Bay region, these data were not used to develop ARDAT because rivers in this region fall outside of the pan-Arctic watershed boundary that was selected for data set development. Inclusion of these rivers would add about 10% more water to the total budget (Déry et al., 2005), but mean ocean circulation in the Hudson Bay region tends to advect river inputs east and southward, away from the Arctic Basin (Prinsenberg, 1986).
- Déry, Stephen & A. Stadnyk, Tricia & K. MacDonald, Matthew & Gauli-Sharma, Bunu. (2016). Recent trends and variability in river discharge across northern Canada. Hydrology and Earth System Sciences. 20. 4801-4818. Doi:10.5194/hess-20-4801-2016.
- Prinsenberg, S.J. 1986. Salinity and temperature distributions of Hudson Bay and James Bay. In: Canadian inland seas, I.P Martini, ed. Elsevier Publishers Ltd., Amsterdam.
- Whitefield, J., Winsor P., McClelland J., Menemenlis D., A new river discharge and river temperature climatology data set for the pan-Arctic region, Ocean Modelling, Volume 88, 2015, Pages 1-15, ISSN 1463-5003, http://dx.doi.org/10.1016/j.ocemod.2014.12.012.
- Zhang, X. He, J. Zhang, J. and Wu, P. (2013) Enhanced poleward moisture transport and amplified northern high-latitude wetting trend Nature Climate Change 3(1):47-51 January 2013 DOI: 10.1038/nclimate1631