Biogeochemical and biophysical responses to episodes of wildfire smoke from natural ecosystems in southwestern British Columbia, Canada

Area burned, number of fires, seasonal fire severity, and fire season length are all expected to increase in Canada, with largely unquantified ecosystem feedbacks. However, there are few observational studies measuring the ecosystem‐scale biogeochemical and biophysical properties during smoke episodes, and hence accessing productivity effects of changes in incident diffuse photosynthetically active radiation (PAR). In this study, we leverage two long-term eddy covariance measurement sites in forest and wetland to study four smoke episodes, which happened at different times and differed in length, over four different years. We found that the highest decrease of shortwave irradiance due to smoke was about 50 % in July and August but increased to about 90 % when the smoke arrived in September. When the smoke arrived in the later stage of summer, impacts on H and LE were also greatest. Smoke generally increased the diffuse fraction from ~0.30 to ~0.50 and turned both sites into stronger carbon-dioxide (CO2) sinks with increased productivity of ~18 % and ~7 % at the forest and wetland sites, respectively. However, when the diffuse fraction exceeded 0.80 as a result of dense smoke, both ecosystems became CO2 sources as total PAR dropped to low values. The results suggest that this kind of natural experiment is important for validating future predictions of smoke‐productivity feedbacks.

How Does the Choice of the Lower Boundary Conditions in Large-Eddy Simulations Affect the Development of Dispersive Fluxes Near the Surface?

Large-eddy simulations (LES) are an important tool for investigating the longstanding energy-balance-closure problem, as they provide continuous, spatially-distributed information about turbulent flow at a high temporal resolution. Former LES studies reproduced an energy-balance gap similar to the observations in the field typically amounting to 10–30% for heights on the order of 100 m in convective boundary layers even above homogeneous surfaces. The underestimation is caused by dispersive fluxes associated with large-scale turbulent organized structures that are not captured by single-tower measurements. However, the gap typically vanishes near the surface, i.e. at typical eddy-covariance measurement heights below 20 m, contrary to the findings from field measurements. In this study, we aim to find a LES set-up that can represent the correct magnitude of the energy-balance gap close to the surface. Therefore, we use a nested two-way coupled LES, with a fine grid that allows us to resolve fluxes and atmospheric structures at typical eddy-covariance measurement heights of 20 m. Under different stability regimes we compare three different options for lower boundary conditions featuring grassland and forest surfaces, i.e. (1) prescribed surface fluxes, (2) a land-surface model, and (3) a land-surface model in combination with a resolved canopy. We show that the use of prescribed surface fluxes and a land-surface model yields similar dispersive heat fluxes that are very small near the vegetation top for both grassland and forest surfaces. However, with the resolved forest canopy, dispersive heat fluxes are clearly larger, which we explain by a clear impact of the resolved canopy on the relationship between variance and flux–variance similarity functions.