YIELD VARIABILITY ANALYSIS IN TWO FOUR-CROP-ROTATION FIELDS (1996-1999) IN SACRAMENTO VALLEY CALIFORNIA

摘要

Yield variability within a field and its inter-annual correlation is one of the main concerns today in precision agriculture. Data was analyzed from two 30 ha commercial four-crop-rotation fields in Sacramento Valley, California. One of the fields hada wheat-tomato-bean-sunflower rotation from 1996 to 1999, whereas the other had a wheat-tomato-sunflower-corn rotation during the same period. After georeferencing, interpolation and correction of monitored-yield data, correlation was measured between the years using 20 x 20m cells. Six two-year comparisons were performed in each field to estimate the congruence of inter-annual variation in defining yield patterns within the fields. Very low correlation values were found using this approach, but afterstandardizing and adding the yields, a larger correlation resulted. This indicates that areas with the same average performance tend to be clustered together. To further the study from another perspective we constructed clusters using standardized yielddata. This provided a different spatial configuration of clusters but also a high level of spatial autocorrelation. Preliminary relations with soil characteristics were developed through Classification and Regression Trees (CART) and regression analysis,using soil core samples data from a 60m grid. Some relatively stable to very stable soil traits such as textural components and organic matter content were found to be important in explaining the average yield over the four year period.