Two years of mesoscale weather data collected from the Federal Aviation Administration's Atlantic City mesometeorological network (mesonet) were used for the development and testing of techniques for 15-, 30- and 60-min predictions of runway visual range, cloud-base height and wind. The primary purpose of this work was to evaluate the utility of the mesonet for improved prediction of aviation terminal weather.
The first year of mesonet data was used for development work. Studies included preliminary development and testing of a prediction model based on advection; and, using detailed data analyses and a numerical-physical boundary-layer model, studies of methods for predicting changes due to non-advective processes.
The second year of mesonet data was used as independent data for testing six forecast techniques. Of the six techniques tested, three did not utilize mesonet data and were used for control purposes. The control forecasts were prepared from persistence, from trend, and by experienced U. S. Weather Bureau terminal weather forecasters. The forecasts based on mesonet data were prepared by a simple advection model which used low-level winds, by The Travelers Research Corporation (TRC) meteorologists and by U. S. Weather Bureau terminal weather forecasters.
The mesonet evaluation was made by comparing results from techniques that used mesonet data with those that did not use mesonet data. The results showed, in general, that none of the techniques that used mesonet data provided forecasts significantly better than those provided by persistence. It is therefore concluded, using the forecast methods evaluated, that the mesoscale data provided from the Atlantic City mesonet are not adequate for providing significantly improved short-range terminal weather forecasts.
There were three weather situations in which forecasts significantly better than persistence were obtained by TRC meteorologists. These forecasts were made by using a combination of mesonet and weather radar data. These special situations were 1) visual range restricted by snow, 2) dense fog accreted by rain, and 3) wind shift lines identified by radar or surface wind reports.
The general failure of the techniques that used mesonet data to provide better forecasts is attributed to the following: low-level winds alone are frequently not adequate for estimating the movement of the fields of cloud-base height and visual range; the number and arrangement of mesonet stations is not adequate for describing these fields or for determining their movement by analysis; and the techniques are not capable, in general, of predicting changes due to formation or dissipation.