Study on spatial associations of wheat yield based on global and local statistic

摘要

The objective of this study was to explore the spatial associations of wheat yield and yield pattern changes under five weather scenarios with the application of both global and local associations. This research was conducted on an 8.4 ha wheat field of Xiaotangshan National Experiment Station of Precision Agriculture from 2001 through 2006. The yield data was collected by CASE AFS (Advanced Farming System) in the harvest seasons. After that, the error of yield data was analyzed, and spatial associations were described in terms of global Moran's I and local LISA (Local Indicator of Spatial Association). The results showed that there were considerable errors in the raw data. In order to reduce the error of raw data, filling-time had been calculated by fitting the grain flows curves. After data processing, the C.V. (Coefficient of Variable) fell from 32.3% to 14.8%. The relationship of wheat yield among different year had also been analyzed. But there was no regular relationship had been found. In this paper, local spatial statistics were use to identify the influences from individual positions and the trends between neighboring positions. The results indicated that yield patterns were highly affected by weather conditions. In Beijing, wheat yields were highly correlated with the average temperature of April in this study. Wheat yield pattern was affected by weather patterns. While the monthly average temperature increment of April, the spatial association of wheat yield had the trend of decrease. Statistically, wheat yields were highly spatially correlated in cold years. With global and local spatial statistics, LISA map of yield difference in two weather patterns, stable and unstable yield areas were identified. The northeast and west region of the field tended to be unstable between different weather patterns. In conclusion, since the recognition of spatial associations on yield could help identify if a yield zone was stable or not, it would benefit the decision-making processes in site-specific management systems.