Abstract: 【Objective】Leaf shape coefficient (α) provides a simple and fast way for the measurement of crop leaf area and LAI (leaf area index) in field. However, the selection of values of this coefficient was very arbitrary and there was no unique standard to follow. In addition, this coefficient was usually considered as a constant regardless of its variations through the whole lifetime of a given crop. 【Method】 In this study, the temporal and spatial variations of α values of summer maize were investigated through a field experiment conducted from June to October in 2015. A total of six maize cultivars with different properties of ripening were involved. The whole growth season of maize was divided into six different stages, i.e. trefoil, jointing, heading, flowering, silking, and maturity. Maize plants were randomly sampled every six days and all leaves were cut off and measured for their length, width, and area with a digital leaf area scanner. Then, α value was calculated for each leaf. The variations of α values were analyzed for different growth stages and among different leaf positions within a single maize plant. Finally, five different models of leave area estimation, which belong to the linear, quadratic, and logarithmic types, were established to estimate the area of each maize leaf. Three different statistics of RMSE (root mean square error), RRMSE (relative root mean square error), and ARE (absolute relative error) were used to represent the estimation accuracy. 【Result】 Based on linear regression analysis between leaf areas and products of leaf length and width of 760 leaf samples of six different maize cultivars, the general average value of &alpha was about 0.78. Then, when estimating maize leaf areas with the model of LA=0.78×L×W, the relative root mean square error (RRMSE) and absolute relative error (ARE) were 9.50% and 6.96%, respectively. The accuracy was the highest among the five different models investigated for the estimation of maize leaf area. The results showed that α value ranged from 0.72 to 0.87 and varied at different growth stages. It increased with fluctuations from trefoil to flowering stage, and then decreased. At flowering stage, the α value was significantly different from other before-flowering stages, while showed no significant difference at silking and maturity stages. For different ripening cultivars, α value only showed a significant difference at flowering and silking stages The α value varied for different leaf shapes in the whole growth season. From the trefoil to before-jointing stage, α value of wide-short leaves was higher than that of thin-long leaves. From then on, α value of wide-short leaves became lower than that of thin-long ones. Within a single maize plant, α values varied remarkably for leaves at different positions. At the flowering, silking, and maturity stages, α values were higher at top and bottom than in the middle. The average value of three-ear-leaves was 0.78 and the standard deviation was less than 0.05. However, α value was about 0.87 for leaves at the top and bottom of maize canopy, with standard deviations from 0.03 to 0.10. The α values at different leaf positions only showed significant difference at flowering, silking, and maturity stages. 【Conclusion】 When estimating maize leaf areas with the model of LA=0.78×L×W, the accuracy was the highest among the five different models investigated. The leaf shape coefficient of 0.78 improved the estimation accuracy of maize leaf area by about 3.86%, when compared with the estimation resulted from a coefficient of 0.75. In general, it is better to adopt various α values at different growth stages and different leaf positions so as to improve the accuracy of simulation and prediction of leaf area of summer maize.

Key words: summer maize, leaf area, leaf shape coefficient, leaf length, leaf width, variation