Giesbrecht, Ian;
Tank, Suzanne;
Frazer, Gordon;
Hood, Eran;
Gonzalez Arriola, Santiago;
Butman, David;
D'Amore, Dave;
Hutchinson, David;
Bidlack, Allison;
Lertzman, Ken
—
2022-01-25
<p class="Abstract" style="margin-top:8px;">Watershed classification has long been a key tool in the hydrological sciences, but few studies have been extended to biogeochemistry. We developed a combined hydro-biogeochemical classification for watersheds draining to the coastal margin of the Northeast Pacific coastal temperate rainforest (1,443,062<i> </i>km<sup>2</sup>), including 2,695 small coastal rivers (SCR) and 10 large continental watersheds. We used cluster analysis to group SCR watersheds into 12 types, based on watershed properties. The most important variables for distinguishing SCR watershed types were evapotranspiration, slope, snowfall, and total precipitation. We used both streamflow and dissolved organic carbon (DOC) measurements from rivers (<i>n</i> = 104 and 90 watersheds respectively) to validate the classification. Watershed types corresponded with broad differences in streamflow regime, mean annual runoff, DOC seasonality, and mean DOC concentration. These links between watershed type and river conditions enabled the first region-wide empirical characterization of river hydro-biogeochemistry at the land-sea margin, spanning extensive ungauged and unsampled areas. We found very high annual runoff (mean > 3000 mm, <i>n</i> = 10) in three watershed types totaling 59,024 km<sup>2</sup> and ranging from heavily glacierized mountain watersheds with high flow in summer to a rain-fed mountain watershed type with high flow in fall-winter. DOC hotspots (mean > 4 mg L<sup>-1</sup>, <i>n</i> = 14) were found in three other watershed types (48,557 km<sup>2</sup>) with perhumid rainforest climates and less-mountainous topography. We described four patterns of DOC seasonality linked to watershed hydrology, with fall-flushing being widespread. Hydro-biogeochemical watershed classification may be useful for other complex regions with sparse observation networks.</p>