A multispectral-based tea garden fertilization decision model and method

The fertilization decision model, which combines multispectral data with tea garden phenology, solves the problems of spatial differences and temporal fluctuations within tea gardens, achieving accuracy and feasibility in tea garden fertilization, reducing environmental risks, and supporting traceable and controllable fertilization decisions.

CN122172558APending Publication Date: 2026-06-09江西省经济作物研究所

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
江西省经济作物研究所
Filing Date
2026-02-27
Publication Date
2026-06-09

AI Technical Summary

Technical Problem

Existing tea garden fertilization technologies are ill-suited to the spatial differences and temporal fluctuations within tea gardens, leading to over-fertilization, incorrect formulations, or inappropriate fertilization windows. They also lack verifiable mechanisms, pose environmental risks and implementation uncertainties, and make it difficult to achieve traceable and controllable fertilization decisions.

Method used

By combining multispectral data with tea garden phenological stages and agricultural event constraints, and through mechanisms such as stress decoupling and conflict gating, active sampling calibration, digital twin scenario extrapolation, causal attribution and counterfactual assessment, and robust risk budgeting, a diagnostic closed loop, a prescription closed loop, and a governance closed loop are formed, and an executable variable fertilization prescription is output.

Benefits of technology

Significantly improves the accuracy and feasibility of fertilization decisions, ensures that fertilization strategies are consistent with agricultural rhythms, reduces the risk of misjudgment and misapplication, improves fertilizer utilization efficiency and reduces environmental risks, and achieves traceable and auditable cross-seasonal self-iterative optimization.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

This invention discloses a multispectral-based tea garden fertilization decision-making model and method. Using the management unit as the smallest object, it acquires multispectral images and, through stitching, calibration, and orthorectification, forms data frames with quality labels, calculating equipment / data health scores. Based on a phenological-agricultural event state machine, it generates fertilization window tokens for hard gating, constructs a state cube to extract stress features and performs conflict determination, triggering minimum sampling for supplementary evidence and rapid calibration. It establishes a multi-scenario digital twin model of nutrients, water, canopy, and runoff, combining causal attribution and dual-caliber risk budgeting to generate dose-layer + formulation / distribution-layer prescriptions and mapping them to execution instructions. Verifiable iteration is achieved through consistency verification, control verification, evidence chain auditing, and sandbox replay grayscale release and rollback.
Need to check novelty before this filing date? Find Prior Art