After last year’s dismal number of flyable tasks, the organisers of the annual Porterville paragliding competition have been questioning if we need a change of dates. A survey has been sent out to all previous participants requesting their preferred dates. I was recently contacted by a friend, Francois Miog (paragliding pilot and instructor), for my thoughts on when we should host it. As a data-loving person I wanted to respond with a well thought out answer.
My first idea was to analyse the RASP forecasts that I’ve been running and storing for the past year. However the downfall with this approach would be that I’d only have one summer season of data to analyse which isn’t much. So I looked elsewhere, and it turns out that windguru has an archive of its forecasts for which I was able to pull the weather forecasts for Porterville for past 3 years. I wondered whether we could pick up any trends that would help with answering the question of when should we host the Porterville paragliding competition?
Limitations of this analysis:
Before getting into the numbers lets appreciate a few things. Firstly I’m using forecast data in this analysis, not actual historical data. Still, I’m ok with this as forecasts tend not to be hugely different from actuals. Secondly, 3 years worth of data points isn’t a lot – essentially it’s 3 seasons which is 3 data points per period of analysis. Thirdly, the available data did not include wind direction for Porterville – I don’t think this is a major issue as we have both the Porterville and Piketberg paragliding takeoff sites available to us for the competition. Piketberg covers anything with an easterly component in it, Porterville covers anything with a westerly component, and sunshine/thermals usually mean that one can takeoff from one of these provided the wind strength is not too strong and it’s not raining. Finally, my method of splitting the months into quarters isn’t strictly correctly because competition dates start and end on a weekend.
Method:
Figure 1: Snapshot of raw data sourced from Windguru archives. Note the data is from the GFS13km weather forecast model
Windguru archive data provided the following useful variables: wind speed, gust speed, and rain for 2 hour intervals for the past years (see figure 1 above for an example of wind speed data). Using these variables we can define what would be acceptable in terms of task flying. For the purposes of this analysis I defined a taskable period as one in which the maximum wind speed was 10kts, the maximum gust speed was 15kts, and there was zero rain. Using this definition we can then use each of the 2 hour intervals forecasts to find out if that specific 2 hour period was good for flying a task. From there we can find out if the day was good enough for flying a competition task. I defined a taskable day as one in which both the 2pm and 4pm periods were both taskable. This means that 4 hours ending at 4pm were forecast as being good for task flying. Once this is done we can then count the number of taskable days for a given period of the month. To do this I split each month up into quarters resulting in 4 quarters per month with some quarters having 7 days and some 8 days. Finally, I plotted the data using boxplots because I like boxplots. They show the spread of data and the ability to add colours to the results (in this case differentiating the months of the year) makes looking at data more fun.
Result
Figure 2: Boxplots of the number of taskable days per quarter of the month
The results of this analysis are plotted in figure 2 above. In terms of answering our original question, we’re looking for the quarters of the month that are higher up on the chart. I must say, I was surprised to see some sort of trend here. The trend being that we have a peak in the 2nd quarter of November, followed by a low point in the 2nd quarter of December, followed by the highest peak at the end of January and another low point in the 3rd quarter of February. NB. The lack of boxes e.g. in December Q1, is because the number of taskable days for this quarter was the same for the past 3 years i.e. 3 days.
When is the best time to have the Porterville Paragliding Competition?
Looking at the past 3 summer seasons, the 4th quarter of January comes out on top, followed by a tie in second place being the 2nd quarter of November and the 4th quarter of February. Of course there’s other factors which the organisers will have to take into account such as clashes with dates of other international competitions, holidays (Christmas and New Years), back to school, business year end, etc.
If you really had to press me for answer I’d go with the second of quarter of November for the following reasons: the weather looks decent, it’s far enough from business year end so hopefully taking leave shouldn’t be too much of an issue, it’s not yet peak tourist season so international and local flight prices shouldn’t be too bad, its not yet peak season in the tandem flying industry so we should get more of Cape Town’s tandem flight instructors competing, and finally it’s my birthday in the second quarter of November which means that you’ll still have to be nice to me even if the weather is crap and I influenced the moving of the competition dates 😉
NB. Three seasons, or rather three data points per quarter is a rather tiny sample, way too little for this analysis to carry any serious weight behind it. I think there’s other reasons for the date moving, as noted in the previous paragraph.
So, 10th to 18th November. What do you think? Comment below
