4.1 Drivers of inter-annual variability
Inter-annual variability in the timing and magnitude of the melt pulse
appears somewhat, though imperfectly correlated with metrics of
precipitation and snowpack. This is likely due to the complex nature of
how precipitation translates to runoff in this catchment, and the
difficulty in characterizing something as nuanced as snowpack and a melt
pulse with singular metrics. Annual WY was only weakly correlated with
annual precipitation (Figure 6a), and constituted only around 30-40% of
the latter. This difference is the result of sublimation and
evapotranspiration (ET), and is consistent with findings in adjacent
catchments (Sueker et al., 2001). Maximum snowpack depth was not
correlated with annual precipitation (not shown, p = 0.71). This can
partially be explained by remembering that around a third of the annual
precipitation falls as rain, and does not contribute to the snowpack
(and may even accelerate melting of the snowpack). This fraction also
changes from year to year. For example, 2020 experienced relatively
heavy snowfall followed by a relatively dry summer, with 76% of the
annual precipitation falling as snow. In contrast, the reverse was true
in 2021 and this fraction was only 58%. Even then, the maximum snowpack
is not a perfect reflection of cumulative snowfall; the region is famous
for its Chinook Winds, and some fraction of the snowpack may sublimate.
Timing also appears critically important. For example, 2020 saw rapid
early snowpack accumulation, and late season storms pushed the maximum
depth to near-record levels in late April (Figure 7b). However, it began
to rapidly melt and was gone by mid-May. In contrast, 2021 saw near
record low snowpack early in the season and a lower maximum depth, but
it ultimately persisted into June. Persistence of the snowpack later
into the spring appears to be an important factor in generating a larger
melt pulse and a higher annual water yield (Figure 7c). Thus, while
annual WY shows some correlation with maximum snowpack depth (Figure
6b), the correlation with the date the snowpack remains above 10 cm is
stronger (Figure 6c). Persistence of the snowpack later into the spring
appears a critical factor in generating larger melt pulses. This may be
because a greater volume is melting and becoming runoff, versus losses
earlier in the season which are more to sublimation.
These larger melt pulses in turn seem to transport a larger mass of DOC
(Figure 6e,f). This correlation was less apparent for
NO3-N mass flux in the years with high frequency data.
This potentially has to do with differences in the seasonality of
NO3-N vs DOC concentrations, with the latter peaking
several weeks earlier than the former, and thus less temporally aligned
with the Q peak. It is also worth noting that NO3-N is
only one component of dissolved nitrogen. Grab samples revealed
concentrations of NH4-N often comparable to
NO3-N, and by difference suggest the largest component
of TDN may be dissolved organic nitrogen (DON). TOC and TN values were
almost identical to DOC and TDN, suggesting minimal particulates.
Interestingly, TDN concentrations appeared to peak later in the spring
than NO3-N, closer to the annual peak Q. The NEON grab
sample dataset is admittedly short (and COVID-19 mitigation protocols
further limited grab sampling in much of 2020), but historic samples
collected by the LTER seem to confirm this (Supplemental Figure 1c).
Flux of TDN is likely to show greater alignment with Q, and respond more
strongly to inter-annual variability in the melt pulse (Figure 6f).
Looking at extremes from the historical record (Supplemental Table 1),
2011 had the highest annual precipitation, the deepest maximum snowpack,
the latest centroid of the melt pulse, and nearly the highest annual
water yield. Unsurprisingly, this resulted in the greatest annual export
of DOC and TDN. Conversely, 2012 had the lowest annual precipitation,
the shallowest maximum snowpack, the earliest centroid of the melt
pulse, and the lowest annual water yield. Again unsurprisingly, this
resulted in the lowest annual export of DOC, and nearly the lowest of
TDN. Similar to what we observed in the high frequency data, years with
below average maximum snowpack (e.g., 2010) can still produce above
average annual water yield and in turn above average solute fluxes, but
only when the snowpack persists later into the spring.