Estimating crop coefficients of winter wheat based on canopy spectral vegetation indices

Abstract: At present, many studies have been carried out on crop coefficients and its variation over years under local climate conditions, but little attention has been given to its estimation method for a regional scale, which plays a key role in the regional application of the FAO 56 crop coefficient approach in crop evapotranspiration and transpiration estimation. In this work, experiments including five nitrogen (N) treatments were conducted in the 2008-2009 and 2009-2010 seasons to investigate the relationships between the crop coefficient (Kc), basal crop coefficient (Kcb) and eight common canopy vegetation indices (VIs) of winter wheat, as well as the effects of N and water stress on them. In addition, the feasibility and the performances of VIs on Kc and Kcb estimation of winter wheat were analyzed. Results demonstrated that high levels of N were associated with high Kcb and low Ke, and vice versa, which resulted in no obvious regular differences in Kc among different N treatments. Crop Kc was weakly correlated with VIs (the coefficient of determination R2 = 0.094 ~ 0.150, p < 0.01, n=195) due to the variations in soil evaporation and soil background, while Kcb had strong correlations with VIs (R2 = 0.511 ~ 0.685, p < 0.01, n=195). In addition, the water stress before resulting in an obvious sign on crop canopy spectral characteristics can introduce considerable scatter in the relations between Kcb and VIs, while N stress had no effects on them. Validation results showed that VIs performed well in crop Kcb estimation, and the enhanced vegetation index (EVI) gave the best accuracy (R2 = 0.765 ~ 0.864, n=150). The proposed method would be more favorable for regional application, since VIs can be easily collected by means of remote sensing. However, it should be pointed out that the method may have some limitations under the conditions with water stress but is not severe according to the above analysis, and as in this case, additional water stress information collected from other sources like thermal images and ground-based wireless sensor network observation would be needed.