Weather routing & ensembles
One great thing about running a blog, is that you get to meet some really smart people. Sailors, sailmakers, navigators, coaches and inventors. One of them is Nils Melsom Kristensen, meteorologist, navigator and router. Some of you might know him from Team Bergen, Cookson 50 Camilla or Swan66s Godot.
I met Nils when we arranged the “Advanced navigation with Expedition” training in Stockholm and Göteborg, and he was clearly fascinated by the possibilities to further develop the methods used in routing. So for the geeks out there, this is an interesting reed. And I’m sure Nils would love some feedback on this.
Weather routing is widely known and used within the sailing community. The most widespread use of this technique is via so called deterministic weather routing, where a weather forecast and a polar diagram describing the boats performance, are put into a routing program, which in turn the calculates the fastest route from point A to point B.
It is widely known, both within the sailing community and the meteorological community, that weather forecasts are not always perfect. Therefore, the use of ensemble forecasting systems are becoming more and more popular. These systems consist of, basically, a model that is run with slightly different initial conditions, to see how the develops and evolve differently as time progresses. The goal of the system, is to be able to capture the possible developments of the atmosphere in a way that the real state of the atmosphere is always “within” the spread of the ensemble.
As these ensemble systems become more and more common and easily available on-line, one would like to incorporate these forecasts into the routing process, to be able to calculate routes that performs well under multiple weather scenarios, and to give the offshore navigator some kind of “measurement” of the risk involved when choosing a route over another.
During the work on my Master Thesis, I tried to come up with such a technique. I came up with a technique of averaging the optimal routes for each ensemble member of the ensemble i space, and the calculate how this route performed compared to the route suggested by the deterministic deterministic forecast alone. In three out of the four cases I studied, the average route performed better than the deterministic. This is thoroughly described in my thesis. I acknowledge that this technique is far from perfect, or optimal, but it is a first shot at using an entire weather ensemble in the routing, and that way being able to reduce the risk in the route decision when the weather forecast is uncertain.
