To facilitate decision making for weather-sensitive operations during severe events, the availability of highly localized predictions focused on business operations is critical. The timely optimization of these operations for economic or societal benefit is primarily driven by efficient allocation of resources, and the routing and scheduling of their deployment. As the lead time for weather predictions is extended, to meet the needs of decision makers, both the temporal and spatial uncertainty in the information, especially for precipitation, becomes critical. To explore the relevance of this idea, we build upon earlier efforts at the IBM Thomas J. Watson Research Center to implement and apply an operational mesoscale prediction system to business problems, dubbed “Deep Thunder”. Initially, this work has focused at the meso-γ-scale up to 24 hours for a number of extended metropolitan areas in the US. We are currently expanding the capabilities on a broader scale to provide lead times for near-surface weather information with lead time up to 72 hours, when the impact of uncertainty on business decisions is potentially greater. We are estimating the uncertainty of high-resolution near-surface forecasts associated with surface characteristics initialization and employing an ensemble-based technique using surface observations from more than 100 sites in the greater New York City metropolitan area. Commonly used surface datasets are supplemented by the local area data provided by organizations that operate in this region. We discuss several summer (convective) rain events to analyze the impact on the storm intensity and precipitation patterns. Additionally, we will discuss how this technique can be used to calibrate the mesoscale system, specifically for the high-resolution local near-surface forecasting in the New York City metropolitan area.