Statistical forecast modeling for San Francisco International Airport.

摘要

During the summer at San Francisco International Airport (SFO), the presence of low stratus in the final approach zone reduces visibility and prevents the airport from operating at full capacity. Data from 1996 through 2000 were collected at San Jose State University (SJSU) and used to build nonlinear models that forecasted for the 2001 stratus season. Their performances are compared to five operational forecasting models developed by MIT Lincoln Laboratory (MIT/LL). The SJSU nonlinear models were more reliable and had smaller overall error than the MIT/LL models. For each forecast hour, the SJSU nonlinear models matched or exceeded the performance of every MIT/LL model except at 1500 UTC and 1800 UTC. This indicates that the SJSU models, which were developed using only local SFO area data, may not represent the major factors that affect stratus dissipation at those times.