Robustness of oxygen metrics to sampling frequency in a stratified lake

Long-term lake monitoring often relies on monthly dissolved oxygen profiles, but the robustness of derived oxygen metrics to this temporal resolution is uncertain. We compared three years (2022-2024) of hourly sensor observations (high-frequency (HF)) with approximately monthly discrete profiles (low-frequency (LF)) from a small, dimictic subalpine lake. We evaluated volume-weighted oxygen concentrations, hypolimnetic oxygen depletion rates (ODR) and depth-specific counts of days with oxygen < 2 mg L⁻¹. To isolate temporal sampling effects on ODR, we subsampled the HF daily record at 7-, 14-, 30- and 60-day intervals across all starting-day offsets within fixed annual fitting periods. LF profiles interpolated to daily resolution reproduced broad seasonal oxygen trajectories, but concentration differences persisted on matched sampling dates, particularly at 9 and 12 m. The mean within-year range width of signed ODR deviations from the full-HF reference was 8.0 percentage points at 7-day sampling, 20.5 at 14 days, 47.5 at 30 days and 48.6 at 60 days. Depth-specific low-oxygen-day counts were strongly correlated between datasets (Pearson's r = 0.933-0.977), but unequal observation coverage precluded unbiased estimates of annual duration. The results indicate that approximately monthly profiles can describe broad seasonal oxygen dynamics in this lake, whereas more frequent sampling reduces the sensitivity of ODR estimates to sampling timing.