Improved Dust Forecast by Assimilating MODIS IR-Based Nighttime AOT in the ADAM2 Model

Abstract

A data assimilation (DA) system employing day- and nighttime aerosol optical thickness (AOT) was developed for the Asian Dust Aerosol Model 2 (ADAM2), using the optimal interpolation (OI) method. The DA system assimilated nighttime AOT for dust retrieved from MODIS infrared (IR) measurements with an artificial neural network (ANN) approach. An Asian dust case that occurred during 14-18 March 2009 was simulated using ADAM2. To examine the impact of the inclusion of nighttime AOT on forecasts of the data assimilation system, experiments were performed with different assimilation cycles (i.e., DA1: 24-hour cycle with daytime MODIS AOT only, DA2: 12-hour cycle with additional nighttime AOT). A control simulation was also performed without data assimilation (CTL). Forecasts were assessed using MODIS-derived AOT distributions as well as ground-based skyradiometer, PM10, and lidar observations. The model-estimated vertical distribution of the dust extinction coefficient was also compared with lidar measurements. Both experiments (DA1, DA2) were found to have improved forecasting, but DA2 outperformed DA1. Results suggest that the ANN-based nighttime AOT contributes more positively to the forecasting through better temporal coverage for data assimilation.