耦合生长模型与知识模型的水稻最优播期决策算法——以‘南粳9108’为例

Optimal rice sowing date decision algorithm based on coupling growth model and knowledge model：A case study of ‘Nanjing 9108’

关键词:  作物生长模型  稻麦生育期  茬口限制  温度适宜度  最优播期

Key words:  crop growth model  rice and wheat growth period  stubble constraint  temperature suitability  optimal sowing date

摘要:

【目的】 江苏省区域尺度的水稻播期优化受气候空间异质性和茬口限制存在显著差异。在稻麦轮作体系下，兼顾茬口限制，实现定量化最优播期决策，是亟待解决的问题。【方法】 文章以江苏省为研究区，选用主栽品种‘南粳9108’和‘扬麦158’，利用2010-2020年CMFD v2.0栅格气象数据驱动CropGrow作物生长模型，模拟稻麦轮作系统的生育进程。基于模拟的生育期和气象数据，采用气候适宜度理论计算各播期的温度适宜度，并结合茬口可行性筛选，确定逐年最优播期，进而以多年平均值作为各空间单元的最优播期。【结果】 （1）经农气站点数据进行参数优化后，模型对水稻关键生育期的模拟精度较高，RMSE均在7 d以内。（2）温度适宜度随播期呈先升高后降低的单峰变化，峰值位置年际波动明显。（3）江苏省最优播期集中在DOY 126~149之间，呈现从沿海到内陆逐渐提前的空间格局，且与各地高产栽培实践播期高度吻合（R2=0.75）。【结论】 该研究提出的决策算法在江苏省具有良好泛化能力，揭示了最优播期的空间分布规律。这将区域播期决策从经验指导提升为可量化的科学工具，可为各地因地制宜制定播期方案提供决策依据。

Abstract:

[Purpose] The optimization of rice sowing dates at the regional scale in Jiangsu Province varies significantly due to spatial heterogeneity of climate and crop rotation constraints. How to achieve a quantitative optimal sowing date decision under the rice-wheat rotation system，while considering crop rotation constraints is an urgent problem to be solved.[Method] Taking Jiangsu Province as the study area，the main cultivars ‘Nanjing 9108’ and ‘Yangmai 158’ were selected，and the CropGrow crop growth model was driven by the grid meteorological data of CMFD v2.0 from 2010 to 2020 to simulate the growth process of the rice-wheat rotation system. Based on the simulated growth stages and meteorological data，the temperature suitability of each sowing date was calculated using the climate suitability theory，and the optimal sowing dates were determined year by year through screening based on crop rotation feasibility. The multi-year average was then used as the optimal sowing date for each spatial unit.[Result] （1） After calibration using agrometeorological station data，the model achieved high accuracy in simulating the key growth stages of rice，with RMSE falling within 7 days. （2） The temperature suitability exhibited a unimodal pattern first increasing and then decreasing with delayed sowing dates，while the position of the peak value showed considerable interannual variability. （3） The optimal sowing dates across the province were concentrated between DOY 126 and 149，presenting a spatial pattern of gradual advancement from the coastal areas to the inland regions. These dates were also highly consistent with the local high-yield cultivation practices （R2=0.75）.[Conclusion] The proposed decision-making algorithm has good generalization ability in Jiangsu Province and reveals the spatial distribution pattern of optimal sowing dates. This elevates regional sowing date decisions from empirical guidance to a quantifiable scientific tool，providing a decision-making basis for formulating sowing date schemes tailored to local conditions.