Models are crucial for simulating complex systems and decision-making, but they have uncertainties that must be characterized and understood. One uncertainty that has been overlooked in agroecosystem assessments is that arising from the temporal disaggregation of temperature and solar radiation. Our study used data from an agricultural weather station network to investigate (a) the errors associated with hourly temporal disaggregation of daily temperatures and solar radiation, (b) how these input errors impact two agroecosystem models, (c) the sensitivity of change assessments to disaggregation errors, and (d) how high-temporal-resolution weather station networks can be leveraged to correct disaggregation errors in daily gridded meteorological data products. Our findings demonstrate that temporal temperature disaggregation errors can have a significant impact on agroecosystem model output, with large errors in sunburn risk estimation (>100% median deviation percentage) but minimal effects on chill accumulation (<5% median deviation percentage). However, we were able to achieve significant reductions in error (>75% error reduction in sunburn risk assessment in majority of cases) by integrating simple monthly statistics from station observations into the disaggregation process. Our study highlights the importance of understanding uncertainties in agroecosystem models stemming from temporal disaggregation of temperature, and the potential benefits of utilizing simple adjustments leveraging weather station networks to improve model accuracy and applicability for decision-making.

中文翻译：

---

温度时间分解的错误可能导致农业生态系统风险评估中不可忽视的偏差

模型对于模拟复杂系统和决策至关重要，但它们具有必须表征和理解的不确定性。农业生态系统评估中被忽视的一种不确定性是由温度和太阳辐射的时间分解引起的。我们的研究使用来自农业气象站网络的数据来调查（a）与每日气温和太阳辐射每小时时间分解相关的误差，（b）这些输入误差如何影响两个农业生态系统模型，（c）变化评估的敏感性分解误差，以及 (d) 如何利用高时间分辨率气象站网络来纠正每日网格化气象数据产品中的分解误差。我们的研究结果表明，时间温度分解误差会对农业生态系统模型输出产生重大影响，晒伤风险估计误差较大（>100% 中值偏差百分比），但对寒意积累的影响最小（<5% 中值偏差百分比）。然而，通过将站点观测的简单月度统计数据整合到分解过程中，我们能够显着减少错误（在大多数情况下，晒伤风险评估的错误减少>75%）。我们的研究强调了理解农业生态系统模型中因温度时间分解而产生的不确定性的重要性，以及利用气象站网络进行简单调整来提高模型准确性和决策适用性的潜在好处。