Frequent Near-Neutral Atmospheric Conditions over Boreal Wetlands Improve the Estimation of Daily Evapotranspiration in Remote Areas

A significant fraction of the energy supply to eastern Canada and northeastern USA comes from large hydropower plants located in the Canadian boreal shield. For instance, the La Grande River watershed near James Bay (Canada) provides inflows to a hydropower complex that produce nearly 40% of the overall peak load of Quebec. In this northern, remote and vast (≈100,000 km2) watershed, boreal forest is predominant landscape, but wetlands (25% of the land cover) represent another key landscape of the river's water budget, inducing significant water transfer by evapotranspiration, among others. Needless to say, these water exchanges greatly affect hydropower production and, therefore, have to be measured, estimated or modelled. The general objectives of this study were to: (i) gain a better understanding of the evapotranspiration process over boreal wetlands using field observations, and (ii) foresee the design of a future network of weather stations to monitor daily rates using basic weather instruments, considering the operating costs in such a large and remote watershed. The study site was a 60-ha bog (53.7°N, 78.2°W) located next to the Necopastic River, a tributary of the La Grande River. The peatland is of ombrotrophic type, meaning that it receives most of its water and nutrients from precipitation. The analysis relied on data collected by a flux tower during a field campaign throughout summer 2012, as well as detailed measurements of the water budget. The eddy covariance data revealed that the atmosphere was neutrally-stratified for more than 60% of the summer. This unusual feature greatly simplifies the Monin-Obukhov Similarity equations for wind speed, temperature and humidity profiles, with the lowest level taken as the surface. Precisely, when assuming a wet surface, the equations can be solved with only air and surface temperature, humidity and pressure, wind speed and average vegetation height. Using kB-1 = ln(z0/z0v) = 9.5 where z0 and z0v are the roughness lengths for momentum and humidity respectively, the method leads to an excellent approximation of daily evapotranspiration fluxes, with a normalized mean error (NME) of about 13%, R2 of 0.75, and a root-mean-squared error (RMSE) of 0.58 mm/day. As a comparison, using direct measurements of net radiation, the Priestley-Taylor formulation yields a NME of 11%, R2 = 0.91 and a RMSE of 0.44 mm/day, while Penman-Monteith has a NME of 11%, R2 = 0.87 and a RMSE of 0.43 mm/day. However, these classic formulations command measurements of net radiation, and the cost of net radiometers becomes prohibitive in this context. The method was also successfully applied at two other Canadian wetlands with significantly different environmental conditions. Ultimately, this study has provided us with a better understanding of water and energy exchanges between subarctic bogs and the atmosphere, with insight on the appropriate evapotranspiration formulations to use in hydrological modelling.