Comparative Study on Main Crop Yield Separation Methods

Abstract

Crop yield separation is one of the important steps in analyzing the impact of meteorological factors on yield. Statistical rice yield data for 1985-2018 from 24 counties in Jiangsu are used to analyze the rationality of different separation methods. Six separation methods are 3-year moving mean, 5-year moving mean, five-point quadratic smoothing, quadratic exponential smoothing, HP filter and year-to-year increment. Consistencies and differences are analyzed from aspects of trend yield and meteorological yield. In order to select better methods that could accurately capture the yield variation caused by meteorological factors, the meteorological yield based on different methods are compared with the typical annual increase and decrease of rice yield records. Finally, as mentioned above, the selected methods are calibrated by the rationality of the relationship between meteorological factors and yield. Results show that the trend yield curves fitted by different methods are in line with the process of social technology development. Compared with the average trend yield, almost all the consistency correlation coefficients are greater than 0.5. It suggests that different methods do not differ much in trend fitting. Characteristics of meteorological yield separated by 3-year moving mean, 5-year moving mean, five-point quadratic smoothing and quadratic exponential smoothing in each county are simultaneously increasing or decreasing. And their standard deviation values are significantly smaller than HP filter method and year-to-year increment method. The result suggests that the rationality of separating the meteorological yields by 3-year moving mean, 5-year moving mean, five-point quadratic smoothing, and quadratic exponential smoothing is higher than the other two methods. Five-point quadratic smoothing method and 3-year moving mean method can capture almost 100% of typical annual meteorological yield changes in the whole research area. Further verification results show that the positive and negative effects of meteorological factors captured by 3-year moving mean and five-point quadratic smoothing method are more consistent with the response to meteorological factors. Overall, separation methods of five-point quadratic smoothing method and 3-year moving mean method are more suitable for this research area and match well with meteorological factors.

Keywords

• climate change
• rice
• trend yield
• meteorological yield
• separation methods