Photosynthetic light response could be expressed as a rectangular hyperbola curve with the fixed parameters A_max and α. Seasonal variations of A_max and α were observed among different ecosystems. In order to eliminate the effects of seasonal variations of A_max and α on simulated result, rectangular hyperbola model was usually fitted at shorter time intervals (e.g. half-month, 10-days, and 5-days). However, this method is tedious and non-mechanism, especially at a shorter time intervals, small amounts of carbon flux or climate data are not enough for simulating accurately rectangular hyperbola model. In this study we tried to elaborate the effects of biotic and abiotic factors on the parameters (A_max and α) as an example of rainfed maize ecosystem. Biotic and abiotic factors may affect the seasonal dynamics of parameters α and A_max. In order to understand mechanism for the influence of these factors on parameters of model, Multiple regression between seasonal variations of parameters (α and A_max) and abiotic factors (e.g. air temperature, soil temperature, relative humidity, soil water content, solar radiation, air vapor pressure deficits(VPD)) and biotic factor (LAI) were evaluated by stepwise regression analysis. The results showed that there was a significant linear relationship between A_max and LAI, and LAI is a main factor affecting seasonal variations of A_max. The relationship between A_max and LAI could be expressed as A_max = a LAI + b (a = 0.64, b = 0.15, R= 0.74, P = 0.002).Thus, a modified model GPP= αPAR(aLAI+b)/αPAR+(aLAI+b) was developed to estimate half-hourly canopy gross primary production (GPP) in maize ecosystem. LAI changed rapidly in a rapid growth phase, but its observed frequency in our experiments was low (at 15-20 days interval). Therefore, we introduced a logistic model to interpolate daily LAI by limited observations. After that, we tested the calculated GPP against all the available observed measurements based on three different simulation methods (the original model with the fixed parameters of α and A_max, respectively, at half-month intervals and throughout the entire growing season; the improved rectangular hyperbola model). Compared with the original model which was fitted throughout the entire growing season, both the original model which was fitted at half-month intervals and our new model are better than the original model throughout the entire growing season. Compared with the original model which was fitted at half-month intervals, our new model has the similar accuracy.The new model introduced the relationship between LAI and A_max, improved accuracy in simulating GPP throughout the growing season, elaborated the mechanism of different accuracy for simulation in GPP at different time scales. The improved model is not only simple, but also easy to explain the continuous variations of parameter A_max. Especially when the amount of missing flux data is large, and the original model is not able to be fitted at short time intervals, it is suitable for using the new model to interpolate.
Evaluation of carbon dynamics in ecosystem is a key issue in global climate change research, the improved hyperbola model could assess ecosystem GPP expediently and accurately. Then, we could evaluate ecosystem carbon budget at regional and global scales expediently by the LAI from remote sensing observations. However, abiotic factors such as temperature, soil moisture, and VPD did not affect the variations of parameters of model in our studies, and these factors may affect GPP by other means. Further coupling the new model with these factors is the next logical step in understanding ecosystem carbon budget.