Sorghum compared: two flights, July 25 and August 17

Comparison between two flights over the same sorghum field at Fazenda Tio Geraldo (Tupaciguara, Brazil): July 25 and August 17, 2026, 23 days apart, 17.06 ha of common area (95% overlap). The goal was to test whether the first flight's recommendation held up and to separate real crop change from instrument drift between flights.

A drift control over road and bare soil (surfaces that do not change) showed a VARI drop of -0.0136 from the instrument alone — nearly half (49%) of the raw drop measured on the crop (-0.0277). After correcting for drift, real drying over 23 days was -0.0140 (-0.0006/day). Additionally, the correlation between standing-crop cells across the two dates fell from r=0.845 (all cells) to r=0.160 (standing crop only) — showing the initially high correlation came from roads and trees, not the crop.

The decisive test: the tercile the July map classified as "driest" (harvest first) was only 3.5% harvested by August, while the "greenest" tercile (leave for last) was already 37.2% harvested — an inverted order, with only 27% agreement (below the 33% chance level, p=0.8). Conclusion: July's spatial pattern was measurement noise, not a real agronomic signal.