New PROSPECT-S model tracks wheat spike growth for better spectral simulations

Researchers in China have developed PROSPECT-S, a radiative transfer model that links winter wheat spike growth with optical properties during grain filling. The model aims to improve hyperspectral simulation, trait retrieval and yield-related analysis across the crop life cycle.

Why it matters: - Winter wheat spikes change shape and structure as grains fill, which changes how they absorb and reflect light. - PROSPECT-S is designed to capture those growth-driven changes, making spectral simulations of spikes more realistic. - The model could improve estimates of light interception and spike photosynthetic contribution, both of which are often left out of crop yield models. - The framework may support non-destructive retrieval of spike traits from UAV- and satellite-based hyperspectral data.

What happened: - Researchers from Chang’an University, Beijing Academy of Agriculture and Forestry Sciences, Beijing Normal University, Henan Polytechnic University and Nanjing Agricultural University published the study on 15 May 2026 in the Journal of Remote Sensing. - The paper introduced PROSPECT-S, a new radiative transfer model for winter wheat spikes across the 400–2,500 nm range. - The study was published under DOI 10.34133/remotesensing.1048. - The work was supported by China’s National Key Research and Development Program, grant 2023YFD2000100.

The details: - PROSPECT-S replaces the unmeasurable structural parameter N with measurable spike width and dry matter. - The model links those traits to accumulated growing degree days through logistic curves. - PROSPECT-S also models the refractive index with a Beer-Lambert law based on spike width. - Validation used 88 spike samples from five cultivars. - The model reached R² of 0.92 and RMSE of 0.052 in validation. - PROSPECT-S retrieved chlorophyll content with RMSE of 6.20 μg cm⁻² and nRMSE of 25.8%. - The model retrieved equivalent water thickness with RMSE of 0.010 g cm⁻² and nRMSE of 16.8%. - The model retrieved dry matter content with RMSE of 0.007 g cm⁻² and nRMSE of 20.4%. - Field experiments ran in Beijing and Henan, China, and covered flowering plus two grain-filling stages. - The team measured spike length, width, fresh weight, dry weight and chlorophyll content. - AGDD was calculated from ERA5 reanalysis temperature data. - Spike width followed a sigmoidal growth pattern as AGDD increased, and the pattern was fitted separately by cultivar. - The structural parameter Nₛ was estimated from reflectance at 1,131 nm and then linked to width and dry matter through a quadratic function. - The width-linked refractive index was expressed as nr(λ,W) = n₀(λ) × (1 – e^{–W×k(λ)+d}). - Specific absorption coefficients for chlorophyll, water and dry matter were recalibrated separately for 400–800 nm and 800–2,500 nm. - Global sensitivity analysis found AGDD was the dominant driver of model variability in the near-infrared region, explaining up to 90% of the variance. - Cross-validation with three partitioning methods produced validation R² values of 0.91 to 0.92. - AGDD errors of ±25°C and ±50°C affected flowering-stage accuracy more than late filling-stage accuracy. - Spike reflectance was measured with an ASD FieldSpec4 spectrometer and a leaf clip for consistent illumination. - Destructive sampling was used to measure fresh weight, dry weight and spike surface area. - Chlorophyll content was measured by spectrophotometry at 440, 649 and 665 nm after ethanol extraction. - Spike surface area was calculated as length × width × 3.8.

Between the lines: - The model moves spike optics away from static leaf-based assumptions and toward a developmental framework tied to real measurements. - That matters because cereal reproductive organs can contribute to photosynthesis, but many crop models still emphasize leaves and ignore spikes. - The strong sensitivity of the near-infrared signal to AGDD suggests developmental stage is a major control on spectral behavior, not just pigment or water content. - The explicit use of measurable traits could make the model easier to apply in field studies than older approaches that depended on hidden structural parameters.

What's next: - The research team said future work will extend the model to rice spikes. - Planned updates include adding anatomical traits following the RSPECT leaf model. - The team also plans to account for the spike’s three-dimensional architecture. - Those additions could improve monitoring of cereal reproductive organs and sharpen stress-related crop assessments.

The bottom line: - PROSPECT-S gives researchers a more dynamic way to simulate wheat spike reflectance during grain filling, which could improve remote sensing, trait retrieval and crop yield analysis.