Rebuilding a Kapadokya Trip Plan Around Balloon Cancellations

In March I helped two friends plan four days in Göreme. The plan was built around one balloon flight on the second morning. The operator cancelled it at 05:10 the same day, citing wind at 1,800 ft. They had a flight booked again for the next morning, which was also cancelled. The third morning went ahead. The whole trip had been scheduled as if the balloon was a fixed appointment, and it cost them two early rises and a rearranged day.

My approach was to treat the balloon as a probabilistic event and build a plan that survives two misses. First, I looked at the constraint dates, not the wish list: arrival date, departure date, and any non-refundable booking. Then I reserved balloon slots on the first, second, and third mornings where possible, with the operator’s cancellation policy in writing. The useful part was a small script that pulled historical wind forecasts for the Göreme area and printed the mornings with the lowest chance of a cancel. I did not have reliable cancellation data, so I used forecast wind speed at 850 hPa and 700 hPa as a proxy.

import requests
from datetime import date, timedelta

lat, lon = 38.6431, 34.8289
start = date(2024, 3, 11)
for i in range(4):
    d = start + timedelta(days=i)
    url = ("https://api.open-meteo.com/v1/forecast"
           f"?latitude={lat}&longitude={lon}"
           f"&hourly=wind_speed_850hPa,wind_speed_700hPa"
           f"&start_date={d}&end_date={d}&timezone=Europe/Istanbul")
    r = requests.get(url, timeout=10).json()
    speeds = [v for v in r["hourly"]["wind_speed_850hPa"][5:7] if v is not None]
    print(d, round(sum(speeds) / len(speeds), 1) if speeds else "no data")

This is not a cancellation predictor. It is a way to rank mornings and decide which one to protect. The mechanism is simple: balloon operators care about wind at altitude more than surface wind, because surface wind can look calm while the launch site is unusable. The cost is that the forecast is coarse, the API has no local launch-site data, and I was ranking four mornings, not making a guarantee.

Where it fails: spring in Kapadokya can have calm forecast wind and still cancel because of a local gust or a decision made for the whole region. I saw one morning where the forecast was 8 km/h at 850 hPa and the flight still did not go. Also, booking three mornings with one operator is often not possible in peak season, and booking with three operators means three deposits. The script only tells you which morning to prioritise; it does not solve the booking constraint.

A concrete question for anyone who has done this: when you have only two mornings in Göreme, do you book the first and third mornings and leave the second as a buffer, or do you book the two most promising forecast mornings and accept the gap?