The predictor code seems to make use of positions that have null position (e.g. lat/lon=0), resulting in dodgy ascent rate calculations if the payload has poor GPS lock.
The query used to get recent positions for a payload should probably:
Filter out any positions where a 'sats' variable exists, and is equal to 0
For the moment I've just set it to filter if either lat or lon is exactly 0. I think in practice it's unlikely we'll ever receive exactly 0 from the sondes.
The predictor code seems to make use of positions that have null position (e.g. lat/lon=0), resulting in dodgy ascent rate calculations if the payload has poor GPS lock.
The query used to get recent positions for a payload should probably: