Exploration of a Model Relating Route Availability in En Route Airspace to Actual Weather Coverage Parameters

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Thursday, 2 February 2006
Exploration of a Model Relating Route Availability in En Route Airspace to Actual Weather Coverage Parameters
Exhibit Hall A2 (Georgia World Congress Center)
Brian Martin, MIT, Lexington, MA; and J. Evans and R. DeLaura

Poster PDF (831.2 kB)

A major concern in contemporary traffic flow management (TFM) is reducing the adverse impact that convective weather (Wx) has on Air Traffic Control (ATC) en route sectors throughout the National Airspace System (NAS). The FAA is currently seeking to reduce the delays through the use of multi-hour (e.g. 2, 4, and 6 hour) probabilistic convective Wx forecasts coupled with strategic planning by the FAA traffic flow managers and airline personnel to determine how en route traffic should be rerouted so as to avoid sector overloads and minimize the magnitude of the delays that occur.

The objective of this study is to develop a model that relates convective Wx coverage to a surrogate for sector capacity - ATC en route sector blockage (RB). This Wx-RB model can then be coupled with the forecast meteorological validation models to provide a model that translates the probabilistic convective Wx forecasts into forecasts of fractional jet route blockage.

Twenty major events occurring in 2002-03 convective Wx season have been used in this study. These Wx events differ in storm type, intensity, and vertical structure as measured by the Corridor Integrated Weather System (CIWS). The convective Wx spatial patterns collected are treated as sample functions for the overall ensemble of convective Wx that can impact a given ATC en route sector. Ten ATC sectors within the National Airspace System (NAS) have been selected for study based on differing geographic location, size, route orientation, and varying route complexity. From the sampled Wx data that affects each of the chosen ATC en route sectors, the RB for the sector has been calculated using a modified version of the route blockage algorithm available through Lincoln Laboratory's Route Availability Planning Tool (RAPT).

An important issue addressed is the role of echo tops in addition to radar reflectivity in determining RB. We compared RB computed considering both storm reflectivity and echo tops with the RB considering only the storm reflectivity terms and found that the consideration of storm echo tops as well as storm reflectivity reduces the loss of capacity due to convective Wx by roughly a factor of two. This suggests that the sector capacity loss can be significantly reduced by considering echo tops as well as storm reflectivity in assessing the operational impact of convective Wx.

The task of modeling RB is accomplished by statistically relating RB to the 2002-2003 Wx data in the form of fractional coverage of high radar reflectivity cells, Wx type, and Wx vertical structure (echo tops) within an ATC en route sector. A variety of Practical Pattern Classification (PPC) techniques have been investigated as candidates for modeling the compiled distributions of RB in an ATC en route sector as a function of Wx parameters. The PPC techniques also provide an estimate of the relative importance of the various Wx parameters to estimate the RB.

Results show that modeling of RB has been accomplished reasonably well for the low route blockage events (RB less than 20%). Apparent success for low RB is due to large number of sampled convective Wx events associated with low route blockage. This proper sampling effort allows the PPC algorithms to reasonably model of the distribution of low RB events as a function of Wx parameters.

The statistical classification models for Wx events associated with higher route blockage (RB greater and equal to 20%) clearly require a much greater (e.g., 10 fold) amount of Wx data to be operationally useful. Work is underway to accomplish the analysis of route blockage and fractional Wx coverage routinely for a large number of Wx events in the CIWS domain.