On the other hand, the Project Office, which handles logistics, housing, etc. was rocking. More people in, more people out. One of the on-going management issues is keeping track of bicycles and bicycle locks. Even out here a certain amount of "borrowing" goes on, and people frequently turn in bike keys with no identification of the bike to which they belong or where the bike was left.

Following our current schedule, we had a weather forecast briefing in the late afternoon. Some of us had already been assessing the upcoming situation, but at the appointed time four to eight KWAJEX science and operations leaders gathered at the Weather Office (adjacent to the Operations Center) and heard an explanation of the current situation and the prospects for interesting weather for the next few days. There are usually two forecasters present. They've gone over the situation ahead of time together, then one leads the discussion. Most of the presentation is spent showing and interpreting the radar, satellite, and computer model forecasts. This bears a resemblence to some television weather forecasts, but it's aimed at a trained audience and focused on the particular needs of our project. Also, they don't spend time developing fancy graphics. Another difference is that the audience gets to ask questions and offer alternative interpretations. Most of the project members are unfamiliar with forecasting for the West Pacific, so the forecasters are usually doing some teaching, too. One of the important principles in the forecasting process is that the forecaster tries to present the forecast as accurately as possible, including an assessment of uncertainty, while it's the user's responsibility to make decisions based on the forecast. It can be a little tricky for the forecaster to respond to more and more specific questions without biasing the forecast for or against the result that the user wants. For KWAJEX, the best result is widespread rain at overpass time.

Our problem is that the TRMM satellite only sees Kwaj twice a day, and half of those are at night, when flight rules prevent us from operating the aircraft. Of the remaining overpasses, some are from such an angle that the satellite can't see the whole aree covered by the Kwaj radar. That narrows it down to about 20 really good overpasses. Finally, the weather has to cooperate. So far, I think we're getting flyable weather for about 30-40% of the overpasses. Tomorrow is a dead-on overpass at 8:34 a.m. local time, but the forecast didn't really bring the rain in until late afternoon. We alerted the aircraft anyway, just because the forecast might change. We didn't make this decision lightly because it requires all the aircraft-related personnel to get up before dawn.