Multimodal in situ rice leaf physiology dataset for uncertainty-aware adaptive sampling and measurement-quality assessment

This dataset supports the manuscript “Uncertainty-aware adaptive sampling for multimodal field measurement: Physics-based consistency, repeatability, and temporal alignment in rice leaf phenotyping.” Dataset overview The dataset contains 1,290 leaf-level in situ observations collected from 86 farmer-managed irrigated rice plots within a 51.2-ha agricultural landscape in Plaosan, Tlogoadi Village, Mlati District, Sleman Regency, Special Region of Yogyakarta, Indonesia. Measurements were conducted during three campaign windows—11–13 February, 24–25 February, and 4–5 March 2026—covering seven field dates and producing 258 plot–date measurement units. Five leaves were measured within each plot–date unit. Rice age ranged from 24 to 103 days after transplanting. Measurement system SPAD indications were acquired using a SPAD-502 Plus chlorophyll meter. Gas-exchange and light-adapted fluorescence quantities were measured using an LI-600 porometer/fluorometer. The primary physiological quantities comprise SPAD index, stomatal conductance (gsw), effective quantum yield of photosystem II (PhiPS2), and electron transport rate (ETR). The workbook contains one worksheet and 44 variables covering observation identifiers, days after transplanting, GPS and LI-600 timestamps, geographical coordinates, elevation, stomatal, boundary-layer and total conductance, apparent transpiration, water-vapour quantities, relative humidity, leaf and reference temperatures, leaf vapour-pressure deficit, atmospheric pressure, sample flow, steady-state fluorescence (Fs), maximum light-adapted fluorescence (Fm′), effective PSII yield, ETR, and 1-, 2-, and 4-s acquisition-stability indicators. The DAT field encodes the plot and leaf replicate. Because this identifier is repeated between campaigns, the acquisition date and plot component of DAT must be combined when constructing the plot–date analysis unit. HST denotes days after transplanting. Data integrity and quality control Original leaf-level values and sensor outputs were retained to preserve analytical provenance. Missing observations, physically invalid values, and −9999 stability-channel sentinel values remain in the dataset so that the variable-specific quality-control procedure can be reproduced. Sentinel and out-of-domain values must not be interpreted as physiological measurements. Invalidity in one sensor channel does not necessarily invalidate simultaneously acquired quantities from another channel; complete-case deletion across all measurement modalities is therefore not recommended. GPS and LI-600 timestamps originate from independent device clocks and must not be assumed to be temporally synchronized. The dataset can be used to assess date-specific clock offsets and residual within-session drift. The five leaves within each plot–date unit are biological sampling replicates rather than repeated instrument measurements of the same leaf. Within-unit variability therefore combines leaf-to-leaf heterogeneity and repeatability of the implemented field-measurement procedure. Recommended uses The dataset supports research on multimodal measurement-quality assessment, physics-based computational-consistency checking, timestamp harmonization, hierarchical variance decomposition, field-protocol repeatability, fixed-size leaf-sampling evaluation, measurement uncertainty, and uncertainty-aware adaptive sampling. Plot–date grouping should be retained during resampling or validation because the 1,290 rows are not statistically independent. Scope and limitations The measurements represent one irrigated rice landscape, one seasonal observation window, and one instrument configuration. The dataset does not contain an independent calibration reference and cannot establish measurement trueness or metrological traceability. The five-leaf plot–date mean should be interpreted as an internal protocol consensus rather than the true physiological value of a plot. External validation is recommended before transferring sampling recommendations to other cultivars, seasons, locations, instruments, or management conditions. License The dataset is released under the Creative Commons Attribution 4.0 International license. Users may share and adapt the material provided that the dataset creators are appropriately credited and the Zenodo DOI is cited.