The SPRINTARS version 3.80/4D-Var data assimilation system: development and inversion experiments based on the observing system simulation experiment framework

摘要

We present an aerosol data assimilation system based on a global aerosolclimate model (SPRINTARS – Spectral Radiation-Transport Model for Aerosol Species) and a four-dimensional variational dataassimilation method (4D-Var). Its main purposes are to optimize emissionestimates, improve composites, and obtain the best estimate of the radiativeeffects of aerosols in conjunction with observations. To reduce the hugecomputational cost caused by the iterative integrations in the models, wedeveloped an offline model and a corresponding adjoint model, which aredriven by pre-calculated meteorological, land, and soil data. The offlineand adjoint model shortened the computational time of the inner loop by morethan 30%.By comparing the results with a 1 yr simulation from the original onlinemodel, the consistency of the offline model was verified, with correlationcoefficient R > 0.97 and absolute value of normalized mean biasNMB < 7% for the natural aerosol emissions and aerosol opticalthickness (AOT) of individual aerosol species. Deviations between theoffline and original online models are mainly associated with the timeinterpolation of the input meteorological variables in the offline model;the smaller variability and difference in the wind velocity near the surfaceand relative humidity cause negative and positive biases in the wind-blownaerosol emissions and AOTs of hygroscopic aerosols, respectively.The feasibility and capability of the developed system for aerosol inversemodelling was demonstrated in several inversion experiments based on theobserving system simulation experiment framework. In the experiments, weused the simulated observation data sets of fine- and coarse-mode AOTs fromsun-synchronous polar orbits to investigate the impact of the observationalfrequency (number of satellites) and coverage (land and ocean), and assignedaerosol emissions to control parameters. Observations over land have anotably positive impact on the performance of inverse modelling as comparedwith observations over ocean, implying that reliable observationalinformation over land is important for inverse modelling of land-bornaerosols. The experimental results also indicate that information thatprovides differentiations between aerosol species is crucial to inversemodelling over regions where various aerosol species coexist (e.g.industrialized regions and areas downwind of them).