Sensor choice and VPD flags: Hamfy’s method and findings
How much does a fixed VPD review flag change between two published temperature–humidity streams when both are evaluated over exactly the same available intervals?
Scope of this review
We analysed the complete Reference compartment climate file from WUR’s first 2022 lettuce experiment: 12,384 records across 43 dates. The source is the CEA Open Data distribution processed by Koala Team / Koidra Tech Inc. Hamfy performed the secondary analysis, not the original cultivation experiment.
How we assessed the evidence
We pinned the source version and hash, checked units and missing pairs, normalized timestamps by at most one minute and calculated air VPD for each valid pair. We retained gaps and compared 11,257 common five-minute intervals, excluding any duration after the final timestamp. The 0.4 kPa rule was checked against 0.3, 0.5 and 0.6 kPa alternatives. Download the full reproducible package.
What the evidence supports
Availability must be audited separately
P contains 25 missing pairs; S contains 1,101. The longest missing S block is 69 h 35 min. The paired comparison covers 938 h 05 min; full-stream averages have different coverage.
The same threshold produces different classifications
Below 0.4 kPa: P 29.8%, S 53.1% of matched time. Exactly one stream is below in 39.7% of that time. Mean air VPD is 0.550 versus 0.425 kPa. These describe this export, not universal sensor performance.
The timestamp adjustment has little effect here
Using the original 4/5/6-minute intervals changes the below-threshold shares by less than 0.02 percentage points. Independent calculations and the Hamfy analyser reproduce the published outputs.
What this review does not establish
- No co-located calibration reference or complete sensor placement history is available in this analysis.
- The source timezone is unspecified. UTC+01:00 is an explicit CSV export assumption; no day–night or outdoor-weather inference is made.
- The review threshold has no alarm delay or hysteresis. Classified intervals are neither actual alarm events nor independent experimental replicates.
- One greenhouse compartment cannot establish indoor-farm outcomes, crop losses, energy savings or which sensor is more accurate.
Hamfy’s proposed next step
Use paired measurements to audit a proposed climate rule before deploying it. Preserve missing data, document sensor locations and verify the measurement chain before attributing a discrepancy to the controller.
Evidence to collect for a real project
- Sensor positions, shielding, heights and calibration records.
- A documented side-by-side comparison against a reference instrument.
- Actual controller averaging, delay, hysteresis and operating event logs.
Source notes
- Public measurement dataset · 2022 / 2023International Autonomous Greenhouse Challenge — 3rd edition, 1st experiment: dataset and dictionaryAutonomous Greenhouse Challenge organizers / Koala Team, Koidra Tech Inc. / CEA Open Data
Reference compartment, complete GreenhouseClimate.csv. The pinned distribution declares the MIT license; its permission notice and the unchanged input are included in the download. Hamfy calculated the findings presented here.
- Calculation reference · 1998FAO Irrigation and Drainage Paper 56 — Chapter 3: Meteorological dataFood and Agriculture Organization of the United Nations
Reference for the relationship between air temperature, saturation vapour pressure and relative humidity. It does not validate our selected 0.4 kPa review threshold for this crop.