Using Geographical Surface Characteristics to Detect Near Real-Time Valley Fog

Valley fog creates difficult navigation conditions across multiple types of transportation networks. For example, pilots can experience poor visualization during the development of fog in valleys in addition to flight delays or cancellations. Detection of valley fog is a specific challenge due to its rapid development, small spatial extent, and the time of day it forms. Polar-orbiting Operational Environmental Satellites (POES) such as MODIS and VIIRS can capture valley fog events because of the higher spatial resolution. However, for the mid-latitudes, POESs only offer one image during the typical time intervals when valley fog is present (generally overnight and early morning hours), increasing the potential to miss valley fog events and not providing information on its development or dissipation. While Geostationary Operational Environmental Satellites (GOES) provide higher temporal resolution, they fail to capture many valley fog events due to the lower spatial resolution over POESs. This research presents an approach for capturing valley fog events that leverages information about the geographical location and physical attributes of valleys in the US to increase valley fog detection. The System for Automated Geoscientific Analysis (SAGA) is used to calculate the locations of valleys using an upscaled version of the 1 arc-second U.S. Geological Survey's National Elevation Dataset. SAGA is executed using Python and a 3°x3° moving window is utilized to calculate the local valleyness of an area to differentiate changes in regional topography. The local valleyness scenes are smoothed and merged together to create a single layer of valleyness for the continental US and Alaska. The valleyness layer is then included in the current GOES, MODIS, and VIIRS fog and low stratus detection algorithm to enhance the detection of valley fog events.