区域PM2.5时空回归建模与预测

基于区域PM_(2.5)时空建模和预测的需要及PM_(2.5)浓度呈现明显的时空分布趋势的状况,以苏南地区2014年PM_(2.5)日监测数据为实验数据,使用回归克里格对区域PM_(2.5)进行时空建模和估值。利用最小二乘法建立了PM_(2.5)与时空位置的三元二次回归趋势模型,建模点趋势值与实测值间的平均误差接近于0,表明趋势模型拟合效果较好;拟合了样点残差的理论变异函数模型,表明该地区PM_(2.5)的空间和时间相关性范围分别为150 km和4 d;基于该模型,使用时空普通克里格对残差进行时空插值;插值结果与趋势项相加,得到PM_(2.5)回归克里格估值结果;通过对比不考虑趋势的时空普通克里格估值结果,发现考虑时空趋势的时空回归克里格法精度提高了1. 29%。对所提方法进行了创新性分析,并对不足之处进行了讨论。

Modeling and Prediction of PM2.5 Distribution Using Spatiotemporal Kriging with Trend Model

PM_2.5 concentration exhibit significant space-time trends due to geographical and seasonal differences. Thus, concentrations trends cannot be ignored when performing spatiotemporal PM_2.5 predictions in an area. In this paper, daily monitoring data of PM_2.5 concentration in southern Jiangsu Province in 2014 were used for modeling and prediction of PM_2.5 distribution by Spatiotemporal Kriging with Trend model (STKT). Firstly, PM_2.5 trends were represented by a space-time polynomial functions. The mean error of trend model closed to 0, indicating that the fitting of trend model is excellect. Secondly, the variogram analysis of the residual PM_2.5 concentrations indicated that there is a varying ST dependence between the PM_2.5concentrations in the domain defined by the 150 km, 4 days ranges. Based on the variogram and spatiotemporal trend model, the spatiotemporal distribution of PM_2.5 in the study area was obtained. Plots of the predicted spatiotemporal PM_2.5 distributions revealed a marked tendency of the PM_2.5 concentrations to increase from the summer and autumn to spring and winter, and from east to west of the study area. STKT can reduce prediction errors in practically and statistically significant ways. A numerical comparison of the STKT technique vs the mainstream Spatiotemporal Ordinary Kriging (STOK) technique showed that STKT can increase accuracy by 12.9% than STOK.