关键词: 光温因子, 一季稻, 生育期模型, 因子分析, 多元线性回归

Abstract:

Based on the observation data of single-season rice phenology, daily average temperature, maximum temperature, minimum temperature, sunshine duration and other meteorological data in 51 mid−altitude areas of Guizhou province from May to September 1991 to 2020, the key climatic factors of the initial period of six stages were determined by using factor analysis and correlation analysis method, including green turning stage, tillering stage, jointing−booting stage, head−flowering stage, filling−ripening stage and harvest stage. In order to explore a simple and easy to operate method for simulating the initial period of single-season rice phenology, multiple linear regression method was applied to construct a simulated model for single-season rice phenology with the above data. The model was verified in observation data of the initial period of single-season rice phenology in 25 experimental stations in Guizhou from 2016 to 2021, as well as temperature and light factors during the same period. The results showed as follows: the meteorological factors that were significantly correlated with the initial period of single-season rice phenology were daily average temperature and daily sunshine duration. There was significant negative correlation between single-season rice phenology and daily average temperature and daily sunshine duration in the first 10 days and the first 20 days, and the correlation coefficients were −0.70 to −0.51 and −0.63 to −0.42, respectively. The correlation coefficient between head-flowering stage and daily average temperature was −0.70 as the largest, and the correlation coefficient between jointing-booting stage and daily sunshine duration was −0.42 as the smallest, all of which passed the 0.01 significance level test, which indicating that the higher temperature and the more sunshine duration before each single-season rice phenology, the shorter the length of each rice phenology and the easier earlier date of each rice phenology. The average coefficient of determination R² of the simulated models of the initial period of single-season rice phenology in the middle altitude area of Guizhou was between 0.88 and 0.94, and except tillering stage and jointing−booting stage passed 0.05 significance level test, other growth stages passed 0.01 significance level test. The observation data of 25 experimental spots in Guizhou was given to verify the model. And when the absolute error was from 3 to 4 days, it was defined as reaching the basic-accurateness level. The simulation accuracy of green turning stage and harvest stage reached above basic-accurateness level was 100%, and that of tillering stage, jointing stage and heading and flowering stage was 83.3%, and that of filling and ripening stage was 100%, which showed that the simulated model could effectively simulate the initial period of rice phenology of the mid-altitude area in Guizhou. It provided scientific and technological support for the meteorological service of rice production in the mid-altitude area of low latitude plateau.

Key words: Light and temperature factor, Single-season rice, Phenology simulated model, Factor analysis, Multiple linear regression