Uncertainty of eddy covariance flux measurements over an urban area based on two towers

Abstract

<p><strong>Abstract.</strong> The eddy-covariance (EC) technique is the most direct method to measure the exchange between the surface and the atmosphere in different ecosystems. Thus, it is commonly used to get information on air pollutant and greenhouse gas emissions, and on turbulent heat transfer. Often one ecosystem is monitored using only a single EC measurement station bringing uncertainties to the ecosystem-level flux values. Furthermore, in urban ecosystems we are often compromised to conduct the single-point measurements in non-ideal locations such as close to buildings and/or in the roughness sublayer bringing further complications to data analysis and flux estimations. In order to tackle the question of how representative a single EC measurement point in an urban area can be, two identical EC systems &amp;ndash; measuring momentum, sensible and latent heat, and carbon dioxide fluxes &amp;ndash; were installed on each side of the same building structure in central Helsinki, Finland, for July 2013&amp;ndash;September 2015. The main interests were to understand what is the sensitivity of the vertical fluxes on the single measurement point and to estimate the systematic uncertainty on annual cumulative values due to missing data if certain, relatively wide, flow-distorted wind sectors are disregarded. <br><br> The momentum and measured scalar fluxes respond very differently to the distortion caused by the building structure. The momentum flux is the most sensitive to the measurement location whereas scalar fluxes are less impacted by the measurement structures. The flow distortion areas of the two EC systems (40&amp;ndash;150&amp;deg; and 230&amp;ndash;340&amp;deg;) are best detected from the mean-wind-normalised turbulent kinetic energy and outside these areas, the random uncertainties of the two systems are between 10 and 40<span class="thinspace"></span>%. Different gap-filling methods to yield annual cumulative fluxes show how using data from a single EC measurement point can cause up to 12<span class="thinspace"></span>% underestimation in the cumulative carbon fluxes when compared to combined data from the two systems. Combining data from two EC systems increases also data coverage from 45<span class="thinspace"></span>% to 69<span class="thinspace"></span>%. For sensible and latent heat, the respective underestimations are up to 5&amp;ndash;8<span class="thinspace"></span>%. We also show how the commonly used data flagging criteria in natural ecosystems, kurtosis and skewness, are not necessarily suitable to filter out data in a densely built urban environment. The results show how the single measurement system can be used to derive representative flux values for central Helsinki but the addition of second system to other side of the building structure decreases the uncertainties. The same results can be assumed to apply in similar dense city locations where no large directional deviations in the source area are seen. In general, the obtained results will aid the scientific community by providing information about the sensitivity of EC measurements and their quality flagging in urban areas.</p>