Date of Award


Document Type


Degree Name

Master of Science


Department of Engineering Physics

First Advisor

Ronald P. Lowther, PhD


Current moisture initialization sources lack the spatial and temporal resolution required for mesoscale moisture forecast accuracy critical for military operations. The Global Positioning System (GPS) satellite constellation provides an opportunity to extract accurate moisture observations based on the refraction of the GPS signal through the troposphere. GPS-derived precipitable water (PW) from two different research areas was independently compared with the Air Force Weather Agency s (AFWA's) MM5 PW model output. Results were concurrent with similar studies comparing GPS-derived PW with numerical weather models. The mean correlation in CONUS was 92.5%, while in Alaska it was 72.8%. Mean model biases were 1.22 mm in CONUS and 0.69 mm in Alaska. Mean RMSEs were 4.36 mm in CONUS and 2.76 mm in Alaska. In addition, comparisons were made between moist and dry locations, showing a 21.5% difference in correlation and a 17.8% difference in RMSE. The GPS network s superior temporal resolution captured the diurnal variations in PW, while the model consistently failed to take such variations into account as its forecast progressed. This seems it could be the largest source of error between the two data sets. A number of non-meteorological error sources exist that could impact use of GPS-derived PW in operational applications, such as terrain differences between the GPS receiver sites and the model interpolated heights. These error sources need to be further addressed prior to operational assimilation of this data into military weather models.

AFIT Designator


DTIC Accession Number


Included in

Meteorology Commons