MEMORANDUM
TO: LFA/Eric Bernhardt, Deputy Project Manager
FROM: Navigation Committee/Dorn Crawford, Chair
SUBJECT: Runway Use Estimates/ARTS Data
The Committee has been studying your Sep 14 draft response to its concerns on runway use assumptions, but has now received an Oct 4 working paper incorporating this material into a base case for analysis. These circumstances prompt us to respond to the draft now, rather than awaiting final copy. The draft, taken as an authoritative account, is a useful extension of the original Aug 27 memo, and resolves many of the concerns the committee expressed in its Sep 3 inquiry. Generally following the order of discussion in your draft, here are some thoughts we can share on remaining issues at this point:
Runway 11: With every respect for the analytical points raised, it remains unwise to exclude an active runway from the modeling construct. Anomalies seem inevitable. The Oct 4 working paper offers good early examples, listing flight tracks for Runway 11 in Tables 6-9 that don’t appear on the corresponding Figures 7-10. The general optics of such an exclusion are likewise unfortunate, particularly for residents in the Rangeland area. Present-day computer capacity too easily accommodates sparse-data cases like these to accept the added risks and costs of exclusion. The airport staff has the same recommendation for updating its runway use program.
Deviations: There’s an invitation in the draft response on this item to a semantic debate we’d rather avoid. The preferred daytime runway use configuration is north to south; when planes fly south to north, that’s a deviation from the preferred daytime runway use configuration. And it occurs in the sample at about double the rate in the forecast, as noted before. Your indication that winds in the sample occurred from the north at nearly twice the ten-year average is adequate to explain the difference, assuming these winds were of sufficient strength to exceed the thresholds for preferred use.
Design problems: You’ve gone to considerable lengths to illustrate the complexities involved in runway use decisions, and your conclusion that "many other factors than wind speed alone are important in the determination of runway use configurations" is persuasive. What’s not clear is how that leads to the conclusion that wind speed alone is adequate as a modeling parameter. The distributions you’ve constructed from the ARTS data are instructive in this regard. For the daytime case, the average of the data is reassuring, but their apparent variability less so. Your notation, if following standard form, indicates a standard deviation of 1.5 knots in the daytime data (though an eyeball inspection of the graph would suggest easily twice that value). Nighttime data, as you indicated, reflect little if any dependence on wind conditions. Both situations accordingly beg for additional determinants to reduce the variability of the forecasts.
You’ve indicated that you conducted a sensitivity analysis on this parameter, but haven’t reported the results – i.e., how would noise contours be affected by a change in value from, say, 3 knots to 4 or 5 in the daytime for tailwind, and corresponding variations for nighttime and crosswind values? As noted in the committee’s last memo, insights like these would add needed perspective to an assessment of the criticality of these forecasts.
Data Anomalies: It seems axiomatic that sample data depicted for presentational purposes would either be representative of the parent population, or that systematic departures from estimated population parameters would be identified and explained. Your earlier amplification on abnormal wind conditions skewing the north-south mix in the sample is an excellent example of the power of such explanatory factors. A better reason for analyzing portions of the sample selected for presentation is difficult to fathom. The associated tasks laid out in the project’s scope of work, 5.3, "Conduct ARTS Analysis…," and 3.3, "Comparison to ARTS Analysis," certainly seem to anticipate fundamentals like these, as indeed they should.
Runway Maintenance: We need to reflect further on maintenance practices and closure rates. On its face, it would appear that ten closure incidents covering 40 hours over a four-week period would have a non-negligible impact on runway use. Perhaps there are additional considerations to relax the implication.
Most of the committee’s remaining observations had to do with apparent peculiarities in the flight tracks observed in the ARTS data, and your draft helps clear up a lot of the problems noted. There are still some things to study further, to be sure. Navigation information presented to the committee, for example, has held consistently that arriving aircraft must assume runway heading at least seven miles from the airport, which the data, and your penultimate response, don’t appear to support. And the last issue discussed, on apparent skew of nighttime arrivals depicted, seems fully explained by the observation on skewed winds in the sub-sample. This begs the question, though, of remaining presentations from this same sub-sample that did not give the same skewed appearance: nighttime departures, daytime arrivals and daytime departures. These we plainly hope to explore further; but at bottom, the data are the data; ultimately, in better understanding what we’ve observed, we all benefit. Thanks for your continued attention to these issues.