南方城市住区气温分布插值方法研究

Interpolation Methods for Temperature Distribution of Urban Communities in Southern China

通过有限的样本点数据和合理插值方法绘制分布图是观测和研究大范围对象常用的方法。有关城市住区气温插值方法的研究还比较缺乏。本文选取广州、南宁2个南方城市的3个典型住区开展夏季晴日的气温测试,以住区日间气温平均值为目标,利用测试数据和Golden Software Surfer13软件,比较11种插值方法的计算性能。结果显示,最近邻点插值法、多元回归法和移动平均法作出的温度等值线分布图均为条装或块状,自然邻点法和线性插值三角网法作出的温度等值线分布图出现边界气温缺失、边界等值线不平滑和等值线全闭合,不适合反映南方城市住区夏季晴日的气温分布情况。反距离加权插值法计算误差小(RMSE≤0.71℃),更能准确反映分布特征和高低值点,是南方城市住区夏季晴日气温插值的合理方法。除插值方法外,测点布置对城市住区气温插值结果的影响也十分显著。

It has been a common way to draw distributions of large objects by means of limited sample data and interpolation method. However,the interpolation method for air temperature of urban community was rarely studied. This paper conducts a field test on air temperature in three typical communities in Guangzhou and Nanning in the summer sunny days,and compares performances of 11 interpolation methods in daily mean air temperature distribution by using the test data and the Golden Software Surfer12. The results show that the air temperature contour maps are strips or blocks by using methods of Nearest Neighbor,Polynomial Regression and Moving Average,and missing boundaries, not smooth or fully closed by using methods of Natural Neighbor and Triangulation with Linear Interpolation,indicating that the methods are not suitable for drawing air temperature distribution in summer sunny days. The paper confirms that the Inverse Distance to a Power method is appropriate with smaller errors（RMSE≤0. 71 ℃） and outperforms others in performances on distribution and high or low point identifications. In addition,the paper also indicates that the measuring points are of significance for interpolation results.