Abstract
The quadratic polynomial fertilizer response model (QPFM) is the primary method for implementing quantitative fertilization in crop production, but the success rate of this model's recommended fertilization rates in China is low because the model contains a high setting bias. This paper discusses a new modelling method for expanding the applicability of QPFM. The results of field experiments with 8 levels of N, P, or K fertilization showed that the dynamic trend between rice yield increases and fertilizer application rate exhibited a typical exponential relationship. Therefore, we propose a unary non-structural fertilizer response model (NSFM). The responses of 18 rice field experiments to N, P, or K fertilization indicated that the new models could significantly predict rice yields, while two experimental fitting results using the unary QPFM did not pass statistical significance tests. The residual standard deviations of 13 new models were significantly lower than that of the unary QPFM. The linear correlation coefficient of the recommended application rates between the new model and the unary QPFM reached a significant level. Theoretical analysis showed that the unary QPFM was a simplified version of the new model, and it had a higher fitting precision and better applicability.