Cities of the future will have to overcome congestion, air pollution and increasing infrastructure cost while moving more people and goods smoothly, efficiently and in an eco-friendly manner. Urban air mobility (UAM) is expected to be an integral component of achieving this new type of city. This is a new environment for sustained aviation operations. The heterogeneity of the urban fabric and the roughness elements within it create a unique environment where flight conditions can change frequently across very short distances. UAM vehicles with their lower mass, more limited thrust and slower speeds are especially sensitive to these conditions. Since traditional aviation weather products for observations and forecasts at an airport on the outskirts of a metropolitan area do not translate well to the urban environment, weather data for low-altitude urban airspace is needed and will be particularly critical for unlocking the full potential of UAM. To help address this need, crowdsourced weather data from sources prevalent in urban areas offer the opportunity to create dense meteorological observation networks in support of UAM. This paper considers a variety of potential observational sources and proposes a cyber-physical system architecture, including an incentive-based crowdsensing application, which empowers UAM weather forecasting and operations.