Development of a three-dimensional variational assimilation system for lidar profile data based on a size-resolved aerosol model in WRF-Chem model v3.9.1 and its application in PM<sub>2.5</sub> forecasts across China

Abstract

Abstract. For the aerosol variables in the model for simulating aerosol interactions and chemistry (MOSAIC)-4bin chemical scheme in the Weather Research and Forecasting–Chemistry (WRF–Chem) model, this study presents an observation forward aerosol extinction coefficient (AEC) and aerosol mass concentration (AMC) operator and corresponding adjoint based on the interagency monitoring of protected visual environments (IMPROVE) equation, and then a three-dimensional variational (3-DVAR) data assimilation system (DA) is developed for lidar AECs and AMCs. DA experiments are conducted based on AEC profiles measured by five light detection and ranging (lidar) systems as well as mass concentration (MC) data measured at over 1,500 ground environmental monitoring stations across China for particulate matter 2.5 µm or less in diameter (PM2.5) and PM between 2.5 and 10 µm in diameter (PM10). An experiment comparing assimilated and without assimilated measurements finds the following. While only five lidars were available within the simulation region (approximately 2.33 million km2 in size), assimilating lidar AEC data alone can effectively improve the accuracy of the initial field of the WRF–Chem as well as its forecast performance for PM2.5MCs. Compared to the without assimilated experiment, DA reduces the root mean square error of surface PM2.5MCs in the initial field of the model by 10.5 μg/m3 (17.6 %). Moreover, the positive effect resulting from the optimization of the initial field for AMCs can last for more than 24 h. By taking advantage of lidar aerosol vertical profile information and the near-surface PM MC observations, assimilating lidar AEC and surface PM2.5 (PM10) simultaneously can effectively integrate their observed information and generate a more accurate 3D aerosol analysis field.

An aerosol vertical data assimilation system (NAQPMS-PDAF v1.0): development and application

“…However, the performance of only assimilating surface PM 2.5 measurements on the surface aerosol simulations is better than that of only assimilating ground-based lidar measurements. This could be explained by the relatively sparser distribution of lidar sites compared with surface PM 2.5 measurement sites and the uncertainty in the spatial representation of lidar data (Liang et al, 2020), as well as the errors in the lumped variables of extinction coefficients with multiple contributions by different aerosol components. Moreover, the problem can also be attributed to the discordant relationship between aerosol mass concentration and extinction coefficients both in the simulation and measurements, which was noticed by Ma et al (2020), and a simple bias correction method was proposed to fix this problem.…”

“…Since then, a few studies have been conducted on the application of lidar DA: studies on assimilating virtual lidar measurements based on an observing system simulation experiment (OSSE) (Wang et al, 2013), studies on assimilating spaceborne Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) (Sekiyama et al, 2010;Zhang et al, 2011;, studies on investigating the short-term (no more than 12 h) performance of analyses and subsequent forecasts (Wang et al, 2014a, b;X. Cheng et al, 2019;Liang et al, 2020), studies based on static background DA methods (Wang et al, 2013(Wang et al, , 2014aZheng, 2018;Xiang, 2018;X. Cheng et al, 2019;Liang et al, 2020), and studies on ground-based lidar concentrated on a few model grids (Ma et al, 2020).…”

“…Cheng et al, 2019;Liang et al, 2020), studies based on static background DA methods (Wang et al, 2013(Wang et al, , 2014aZheng, 2018;Xiang, 2018;X. Cheng et al, 2019;Liang et al, 2020), and studies on ground-based lidar concentrated on a few model grids (Ma et al, 2020).…”

An aerosol vertical data assimilation system (NAQPMS-PDAF v1.0): development and application

“…However, the performance of only assimilating surface PM2.5 measurements on the surface aerosol simulations is better than that of only assimilating groundbased lidar measurements. This could be explained by the relatively sparser distribution of lidar sites compared with surface PM2.5 measurement sites and the uncertainty in the spatial representation of lidar data (Liang et al, 2020), as well as the errors in the lumped variables of extinction coefficients with multiple contributions by different aerosol components. Moreover, the problem can also be attributed to the discordant relationship between aerosol mass concentration and extinction coefficients both in the simulation and measurements, which was noticed by , and a simple bias correction method was proposed to fix this problem.…”