Accurate near-surface air temperature is demanded for climate change research. To reduce the air temperature observation error, this paper presents a novel radiation shield. First, a computational fluid dynamics (CFD) method is applied to obtain an optimum design of the radiation shield. Next, the CFD method is used to obtain quantitative radiation errors. Then, a neural network model is used to obtain a radiation error correction equation. Finally, observation experiments are conducted to vertify the actual performance of the shield and the corresponding correction equation. Experimental results show that the mean radiation error of the shield proposed in this paper is approximately 0.04 °C. In addition, the comparison between the radiation errors provided by the experiments and the radiation errors given by the correction equation show that the mean absolute error (MAE) and the root mean square error (RMSE) are 0.012 °C and 0.015 °C, respectively. The radiation error of the radiation shield proposed in this paper may be 1–2 orders of magnitude lower than the radiation errors of the traditional instruments.

中文翻译：

---

用于监测地表气温的有效、低成本、自然通风辐射屏蔽的设计和实验研究

气候变化研究需要准确的近地表气温。为了减少气温观测误差，本文提出了一种新型的辐射屏蔽。首先，应用计算流体动力学 (CFD) 方法来获得辐射屏蔽的最佳设计。接下来，使用 CFD 方法获得定量的辐射误差。然后，使用神经网络模型获得辐射误差校正方程。最后通过观测实验验证了屏蔽的实际性能及相应的修正方程。实验结果表明，本文提出的屏蔽的平均辐射误差约为0.04°C。此外，实验提供的辐射误差与修正方程给出的辐射误差的比较表明，平均绝对误差（MAE）和均方根误差（RMSE）分别为0.012℃和0.015℃。本文提出的辐射屏蔽的辐射误差可能比传统仪器的辐射误差低1-2个数量级。