Exploring the Feasibility of ERA5 ECWMF Reanalysis Data for Estimation of PM10 Surface Concentration Over a Complex Terrain

In an attempt to overcome challenges in PM10 surface concentration estimation imposed by sparse ground-based observations over the complex terrain of Brahmaputra Valley (BV), this study has zeroed in on three objective processes. The first objective is to estimate PM10 using a Multiple Linear Regression Model (MLRM) assimilating planetary boundary layer height (PBLH), meteorological variables [air temperature (AT), relative humidity (RH), wind speed (WS)], and AOD 500 nm for Dibrugarh, a microcosm site of the BV. The second objective is to determine the PBLH from radiosonde observations using the vertical gradient of eight fundamental atmospheric fields. The third objective is to analyze the PM10-meteorology, PM10-PBLH, and PM10-AOD 500 nm relations. The MLRM is fitted with the meteorological variables and PBLH from two sources- ERA5 and ground-based, to assess the model’s performance and ERA5 data feasibility. Results show that the R² for ground-based MLRM (R² = 0.655) is higher than that of reanalysis-fitted MLRM (R² = 0.563) with RMSE 17.761 and 20.753 respectively. The minimum difference in R² and RMSE indicates that ERA5 data performs well in estimating PM10. Out of all the vertical gradients applied, the maximum vertical height of WS and maximum vertical gradient of virtual potential temperature best represent the PBLH.