When examining the inter-annual variation of cumulative evaporation, we noted that the seasonal onset and the break-up of ice cover did not have a large impact on total yearly evaporative losses. The longer ice cover period in 2019 (145 days) than that of 2020 (133 days) resulted in lower cumulative evaporation. This trend is not repeated in 2021, which featured the shortest ice cover period (119 days) and lower cumulative evaporation compared to 2020 (Fig. 14). We also noted that the timing of ice onset was very consistent, as it began around January 1 of each year, while ice break-up was more variable between years. As for duration, the onset of ice lasted between two and three weeks while ice break-up lasted five to seven weeks prior to the reservoir becoming completely ice-free.
3.3 Uncertainties
There are uncertainties that affected the EC measurements that were used in this study. Random sampling uncertainties were calculated following Finkelstein and Sims (2001) and amounted to ≈ 2% for both sensible and latent heat fluxes over the whole study period. In addition, the error associated with gap-filling was noteworthy for H , LE and Rn because of the difficulty of measuring fluxes over an inland water body. The use of a raft results in oscillations that must be considered in the calculation process, while the use of a flux tower on the shore limits the measurement period over the reservoir.
EC measurements were also subject to underestimations linked to the lack of energy balance closure (Foken, 2008), which is best described in terms of the energy balance ratio (EBR, equation 3). Figure 15 illustrates the monthly EBR values in the reservoir for 2019 to 2021. This confirms that the energy budget layer was subject to non-closure.