摘要
利用随机森林和向前逐步选择算法筛选关键气象因素,基于广义加性模型等构建6类甘蔗糖分含量模拟模型,以关键气象因子为自变量模拟甘蔗成熟期的糖分含量,通过模拟值与实测值对比分析,对模拟效果进行评价,并量化关键气象因素对甘蔗糖分含量的影响。结果表明:不同地区、不同时间尺度气象因素对甘蔗糖分含量均有影响,关键气象因素包括气温日较差、最低地温、日照时数、最高气温、平均气压、降水量及平均相对湿度,气温日较差对甘蔗糖分含量的影响呈近似稳定的线性趋势,其余关键气象因素对甘蔗糖分含量的影响呈复杂波动趋势。基于关键气象因素建立模拟甘蔗糖分含量的6种模型,广义加性模型模拟效果最优,拟合R^(2)达0.97,模型验证的绝对误差为0.12个百分点,相对误差为0.86%,均方根误差为0.15。说明模拟值与实测值具有较好的一致性与相关性,模型可实现对广西甘蔗糖分含量的预测,同时,量化关键气象因素对甘蔗糖分含量的影响,也为调控生态系统提供依据。
Statistical analysis was used to analyze the influences of meteorological factors such as temperature and precipitation on the sugar content of sugarcane.Random forest and forward stepwise selection algorithms were utilized to screen the key meteorological factors.Based on the generalized additivity model and so on,six sugar content simulation models were constructed to simulate the sugar content of sugarcane at the maturity stage with the key meteorological factors as independent variables.Evaluate the simulation effects and quantify the influences of key meteorological factors on the sugar content of sugarcane by comparing and analyzing the simulated and measured values.The results showed that the meteorological factors in different regions and time scales affected the sugar content of sugarcane.The key meteorological factors include diurnal variation of temperature,minimum ground temperature,sunshine hours,maximum temperature,average air pressure,precipitation,and average relative humidity.The influence of diurnal variation of temperature on the sugar content of sugarcane showed a nearly stable linear trend,while the influences of the other key meteorological factors on the sugar content of sugarcane showed a complex fluctuation trend.Based on the key meteorological factors,six models were established to simulate the sugar content of sugarcane.The generalized additive model had the best simulation effect,with the fit R^(2) up to 0.97,the absolute error was 0.12 percent points,the relative error was 0.86%,and the root mean square error was 0.15 in the model validation.It indicates that the simulated values have good consistency and correlation with the measured values,and the model can realize the prediction of the sugar content of sugarcane in Guangxi.Meanwhile,quantifying the effects of key meteorological factors on the sugar content of sugarcane also provides a basis for regulating the ecosystem.
作者
郑国俊
祝光湖
王勤龙
黄海荣
李翔
罗霆
马鹏程
王泽平
ZHENG Guo-jun;ZHU Guang-hu;WANG Qin-ong;HUANG Hai-rong;LI Xiang;LUO Ting;MA Peng-cheng;WANG Ze-ping(School of Mathematics and Computing Science,Guilin University of Electronic Technology/Center for Applied Mathematics of Guangxi(GUET),Guilin 541004,China;Sugarcane Research Institute,Guangxi Academy of Agricultural Sciences/Key Laboratory of Sugarcane Biotechnology and Genetic Improvement(Guangxi),Ministry of Agriculture and Rural Affairs,Nanning 530007)
出处
《中国农业气象》
CSCD
2024年第6期609-618,共10页
Chinese Journal of Agrometeorology
基金
中央引导地方科技发展资金项目(桂科ZY21195033)
国家自然科学基金项目(32260715)
广西重大科技专项项目(桂科AA22117004)
广西科技计划项目(2022AC06002)。
关键词
甘蔗糖分含量
气象因素
随机森林
回归模型
模拟
Sugar content of sugarcane
Meteorological factors
Random forest
Regression model
Simulation
