Abstract
Retrieval of land surface variables and atmospheric variables over land from passive microwave remote-sensing data sets has been a challenge for many years. A lot of progress has been made in these quests such as using cloud-resolving models and data assimilation. Data assimilation allows the integration of observations (including observation errors) into imperfect models, thereby yielding more improved model forecasts. In this work, a coupled data assimilation framework (CDAF) is proposed and applied to predict the evolution of land surface and atmospheric conditions. CDAF comprises a coupling of two data assimilation schemes, namely a land data assimilation scheme (LDAS) and an ice microphysics data assimilation scheme (IMDAS). This system has been developed and evaluated using data for the Tibetan Plateau. In this framework, both low-frequency and high-frequency passive microwave brightness temperatures (T(B)s) are assimilated. Low-frequency TBs are assimilated in the LDAS subsystem and used to obtain land surface conditions, which are subsequently used as improved initial conditions together with high-frequency TBs and assimilated in the IMDAS subsystem to obtain atmospheric conditions. The retrieved land surface variables and integrated atmospheric variables are demonstrated to show good agreement with observed land and atmosphere conditions such as those derived from point measurements of temperature and soil moisture (using the Soil Moisture and Temperature Measurement System (SMTMS)), sonde, Advanced Infrared Sounder (AIRS) and Global Precipitation Climatology Project (GPCP) products. The distribution of integrated cloud liquid water and cloud ice is shown to follow the observed cloud distribution over the study area. It is shown that by using IMDAS with modifications to account for precipitation and a good description of land surface emission, it is possible to obtain precipitation information of high fidelity over the land surface. Retrieved integrated water vapour using IMDAS shows correspondence with 'corrected' AIRS total precipitable water product. It is also shown that the relative humidity profile obtained from IMDAS agrees with the corresponding sonde profile. From the simulations, it is clear that by using the CDAF, there is marked improvement in the forecast conditions compared with the non-assimilation scenario for all of the variables considered. Comparisons with observed land surface conditions and inferences of atmosphere state from the Geostationally Operational Environmental Satellite Series 9 (GOES- 9) InfraRed Channel 1 (IR1) brightness temperatures and the GCPC's cumulative daily precipitation indicate that the CDAF is able to generate reliable forecasts that agree with observation-derived products.