Rowing Across an Ocean of Contours

Today I'm very happy to feature a guest post from my friend Angie Pendergrass, a PhD student in the Atmospheric Sciences department here at UW. She's got the mother of all side-projects: using her data analysis and climate modeling skills to help predict ocean currents for a group of crazy brave rowers...

Imagine you're rowing a boat across the ocean. You can't row very fast, just 3 knots if you try your hardest (that's a little slower than walking speed.) You have 4000 nautical miles nautical miles of sea to cover. Scattered across the ocean surface, you know there are some favorable currents and some eddies, spinning whirlpools of water about 50 nautical miles across, with current up to at least one knot. They can help you if you can ride them favorably, or they can suck -- a lot -- if they're against you. You really need to know what these eddies are doing!

Luckily you have a team of weather forecasters and a navigator ashore helping you out. There are actually some guys doing this row, called OAR Northwest. They set out from Dakar, Senegal in late January for Miami, Florida, and now they're halfway across the middle of the Atlantic; check them out at oarnorthwest.com. Your guest-author is their lead weather forecaster.

So now you should be super excited to hear that the National Center for Environmental Prediction (part of NOAA) runs a model of the ocean that diagnoses the ocean surface currents. Great, this is perfect! Let's take a look!

Well, I want to zoom in. I use some matplotlib tools (a python toolbox) - quiver to plot arrows, and contourf to plot filled contours. As it happens, if you just plot the current as a vector with quiver allowing the arrow size to get bigger for faster currents, the plot is nearly unreadable. Using color contours to show current speed and arrows to show direction is nice, I think.

You hopefully see the stars. One of them shows where the rowers are. How am I figuring this out? The boat sends back a data stream of its location and I find the most recent point. It's also really important for forecasting to get an idea of where they will be, for example, tomorrow. I find their location 24 hours ago and plot that. Then I extrapolate that line to project where the boat is headed.

I repeat that for forecast snapshots at every 6 hours out to 72. Then we (my forecast team) look at these plots and send forecasts every day or two, and send a summary to the rowers. Is there an eddy ahead? Will it help or hurt you? Should you go for it or avoid it? It would be kind of miserable for them to row against an eddy, and they may not even be able to make progress. On the other hand, it'd be a shame to miss a nice boost.

Check out this video of the rowers, the chief scientist, and my plots on Canadian TV!

Here's a harder question: are these current forecasts any good? This model was set up focusing mostly on more well-traveled coastal areas. It is probably untested for our purpose. No boats, even sailboats, go at speeds this slow. These currents are unlikely to make much of a difference to most mariners. Of course, oceanographers care, and they set out drifting buoys to measure currents by floating in it that should pick this up. However, this model is initialized from satellites, which should see eddies but won't pick up smaller-scale phenomena that the rowers notice.

So, next time you're setting out to row across the ocean, you'll know to make your surface current maps at a nice scale. And then row, row, row, regardless of what the forecasters tell you.