"Luiz de Queiroz" College of Agriculture, University of São Paulo, Piracicaba, SP, 13418-900, Brazil.
Int J Biometeorol. 2021 Nov;65(11):1881-1893. doi: 10.1007/s00484-021-02145-6. Epub 2021 May 10.
High-quality measured weather data (MWD) are essential for long-term and in-season crop model applications. When MWD is not available, one alternative for crop simulations is to employ gridded weather data (GWD), which needs to be evaluated a priori. Therefore, this study aimed to evaluate the impact of weather data from two GWD sources (NASA and XAVIER), in the way that they are available for end users, on simulating sugarcane crop performance within the APSIM-Sugar model at traditional sites where sugarcane is grown in Center-South Brazil, compared to simulations with MWD. Besides, this study also evaluated the impact of replacing GWD rainfall by the site-specific measured data on such simulations. A common sugarcane cropping system was repeatedly simulated between 1997 and 2015 for different combinations of climate input. Both NASA and XAVIER appear to be interesting for applications that only require temperature and solar radiation for predictions, such as crop phenology and potential yield. Nonetheless, GWD should be used with caution for crop model applications that rely on accurate estimation of crop water balance, canopy development, and biomass accumulation, at least with crop models that run at a daily time-step. The replacement of gridded rainfall with measured rainfall was pivotal for improving sugarcane simulations, as observed for cane yield, by increasing both agreement (NASA d index from 0.67 to 0.90; XAVIER d from 0.73 to 0.93) and R (NASA from 0.35 to 0.76; XAVIER from 0.43 to 0.79) and reducing root mean square errors (RMSE) from 32.8 to 16.3 t/ha when simulated with other variables of NASA data and from 27.9 to 12.7 t/ha when having XAVIER data as input. Therefore, while using both GWD sets without any correction, it is recommended to replace gridded rainfall by measured values, whenever possible, to improve sugarcane simulations in Center-South Brazil.
高质量的实测气象数据（MWD）是长期和季节性作物模型应用的基础。当 MWD 不可用时，作物模拟的一种替代方法是使用网格化气象数据（GWD），但需要事先进行评估。因此，本研究旨在评估两种 GWD 数据源（NASA 和 XAVIER）的气象数据，根据其可用性，对在巴西中南部传统的甘蔗种植区使用 APSIM-Sugar 模型模拟甘蔗作物表现的影响，与使用 MWD 的模拟进行比较。此外，本研究还评估了用特定地点实测数据替代 GWD 降雨数据对这些模拟的影响。对于不同气候输入的组合，1997 年至 2015 年期间，对常见的甘蔗种植系统进行了多次重复模拟。NASA 和 XAVIER 似乎都适用于仅需要温度和太阳辐射进行预测的应用，例如作物物候和潜在产量。然而，对于依赖于作物水分平衡、冠层发育和生物量积累的准确估计的作物模型应用，GWD 应谨慎使用，至少对于在每日时间步长运行的作物模型而言。用实测降雨替代网格化降雨对于提高甘蔗模拟至关重要，就像甘蔗产量一样，通过提高一致性（NASA d 指数从 0.67 提高到 0.90；XAVIER d 从 0.73 提高到 0.93）和 R（NASA 从 0.35 提高到 0.76；XAVIER 从 0.43 提高到 0.79），并降低均方根误差（RMSE）（从模拟 NASA 数据其他变量时的 32.8 吨/公顷降低到 16.3 吨/公顷，从输入 XAVIER 数据时的 27.9 吨/公顷降低到 12.7 吨/公顷），从而改善了巴西中南部的甘蔗模拟。因此，在不进行任何校正的情况下使用这两组 GWD 时，建议尽可能用实测值替代网格化降雨，以改善巴西中南部的甘蔗模拟。
