Forecasting a number of environmental hazards relies upon accurate knowledge of meteorological variables (e.g., air quality, heat stress, ice and snow, floods, high winds, fog). The skill of numerical weather prediction (NWP) is strongly constrained by the accuracy of the initial data, estimated by assimilating expensive observations. However, there are burgeoning sources of inexpensive datasets of opportunity (citizen science, sensor networks etc.) that could be utilized. Lack of knowledge about natural variability in urban areas, data inhomogeneity and data quality hinders uptake of these data.