Recently, work has begun on the development of winter weather forecasts that are both highly automated and designed to address road maintenance issues. These forecasts are intended to be tailored to the specific needs of state Departments of Transportation (DOTs), with the eventual goal of assisting with maintenance decisions, such as chemical choices and crew scheduling. Such forecasts are required to be made on scales on the order of a few kilometers. This presents several large challenges, meteorologically. Both surface observations (e.g. METARs) and readily available forecast models (e.g. NCEP's Eta and Aviation) typically provide information on scales of tens to hundreds of kilometers. This mismatch in scale is currently handled using simple interpolation. While this is probably adequate for general features, such as air temperature a few meters above the ground, it is unlikely to do a good job at handling nuances that are critical to the occurrence of hazardous conditions along a roadway. These include 1) fine-scale topography (inclduing which way the road faces, e.g. north versus south side of a hill), 2) location relative to trees, buildings and water bodies, 3) wind exposure, and 4) if the road surface is on a bridge, to name a few. Fine-scale climatology information may provide some clues how to handle these issues in an intelligent fashion, but such data typically does not exist. This paper will discuss some of the challenges of using interpolation techniques, and developing/using fine-scale climatological data. The incorporation of special observations made by DOT weather and road-condition-sensing stations will also be discussed, including its use for verification of model-based forecasts.