The use of a dense network of low-cost CO₂ sensors is an attractive option for measuring CO₂ emissions in cities. However, these low-cost sensors are also subject to uncertainties. Here, we describe a novel method of field calibration for correcting temperature-related errors in the CO₂ sensors deployed in the BEACO₂N network. We show that with this temperature correction, we can achieve a sufficiently low network error to allow for the evaluation of CO₂ emissions at a neighborhood scale.