G6高速公路乌兰察布到呼和浩特段路面温度变化特征及预报模型

Characteristics of Road Surface Temperature and Prediction Model of the G6 Freeway from Ulanqab to Huhhot

利用G6高速公路乌兰察布到呼和浩特段卧佛山隧道出口和土默特左旗收费站两套交通气象站数据和全球大气再分析资料,统计分析了两站不同季节、不同海拔高度和不同天空状况下的路面温度日变化特征,基于路面温度与气象因子间的相关性建立多元线性回归模型和Elman神经网络模型,并对模型预报效果进行对比检验。结果表明:G6高速公路乌兰察布到呼和浩特段各季节路面温度在日出后快速升高,午后达到峰值后快速下降。白天气温先上升后下降,峰值滞后于路面温度,提前于路基温度(-10 cm)。夜间气温和路面温度要先于路基温度(-10 cm)达到谷值。较低海拔站点路面温度高于较高海拔站点,春季平均差值最大,冬季最小。与晴到多云天气对比,夏季阴雨天路面温度与气温的差值较小,日变化不明显。Elman神经网络模型对路面温度的预报效果优于多元线性回归模型,其中,对路面最低温度的预报效果更为良好,对路面最高温度的预报模型还需进一步优化。

Based on the data of the two traffic meteorological stations in Wofo Tunnel Exit and Tumot Left Banner Toll Station from Ulanqab to Huhhot section of the G6 freeway and the reanalysis data of ECMWF, the daily change characteristics of the road surface temperature in different seasons, different altitudes and different sky conditions were analyzed. Based on the correlation between road surface temperature and meteorological factors, a linear regression model and an Elman neural network model were established, and the prediction effects of two models were compared and tested. The results showed that in the Ulanqab to Huhhot section of the G6 freeway, the road surface temperature rose rapidly after sunrise in each season, and then dropped rapidly after reaching the peak in the afternoon. In the daytime, the air temperature first rose and then fell, and the peak value lagged behind the road surface temperature and advanced the roadbed temperature(-10 cm). In the nighttime, the air temperature and road surface temperature reached the valley value before the roadbed temperature(-10 cm). The road surface temperature of the low altitude station was higher than that of the high altitude station, with the largest average difference in spring and the smallest in winter. Compared with sunny to cloudy in the summer, the difference between road surface temperature and air temperature in rainy days was smaller, and the daily change was not so obvious. The prediction effect of the Elman neural network model on road surface temperature was better than the multiple linear regression model. The prediction effect of the minimum road surface temperature was better, while the prediction model of the maximum road surface temperature needed to be further optimized.