Comparison of microwave radiative transfer models and Retrieval of Humidity using Satellite Microwave Data

Abstract

Water vapor plays an important role in tropical climate through its impact on Earth’s radiation budget. Accurate knowledge of the spatial and temporal distribution of water vapor is a fundamental requirement for understanding Earth’s climate. In this study, two microwave radiative transfer (RT) models, namely the NCAR (National Center for Atmospheric Research) RT model and the EUMETSAT (European Organisation for the Exploitation of Meteorological Satellites) RT model, were considered for simulating brightness temperatures at the top of the atmosphere. The simulated brightness temperatures in AMSU B (Advanced Microwave Sounding Unit B) water vapor channels were inter compared using radiosonde profiles of temperature and relative humidity for nadir viewing (zenith angle = 0°). The radiosonde profiles were obtained from BOBMEX (Bay of Bengal Monsoon Experiment), ARMEX (Arabian Sea Monsoon Experiment) and ISRO satellite validation research cruises. The two RT models agreed with each other, with a root mean square (RMS) difference of 1.14 K to 1.6 K, which is comparable to the noise equivalent brightness temperature difference (NEAT) of the sensor. Both RT models were compared with AMSU B satellite observed brightness temperatures using collocated radiosonde launches in time and space. The comparison was done only for non precipitating days, identified using METEOSAT 5 IR brightness temperature and observed rainfall data. The RMS error between satellite observations and RT simulations ranged from 1 K to 5 K across different channels. At high IR brightness temperatures (Tb > 280 K), both models showed simulation errors of about 2 K. The NCAR RT model was used for sensitivity analysis because of its higher vertical resolution. The sensitivity study was carried out for the water vapor sounding channels of a new microwave humidity sounder called SAPHIR, to be launched by a French agency. Observed profiles of temperature and relative humidity were used as base states. Key findings include:

A change in temperature lapse rate by 0.15 K/km resulted in brightness temperature changes smaller than NEAT in all channels. Error in humidity retrieval was higher in moist atmospheres than in dry atmospheres. Error in specific humidity retrieval was about three times larger in moist atmospheres compared to dry atmospheres at the same heights. In similar channels, the difference in retrieval error between dry and moist atmospheres ranged from 0.10 g/kg to 0.20 g/kg. In the presence of cloud liquid water up to 70 g/m³, the change in brightness temperature was smaller than NEAT. Sensitivity to surface emissivity changes was negligible for all channels except 172.31 GHz, where a change of 0.015 in emissivity resulted in a 0.5 K change in brightness temperature, potentially leading to a 10% humidity retrieval error.

Finally, relative humidity retrieval was performed using a linear algorithm with both RT models. AMSU B brightness temperatures were used to retrieve humidity, which was then compared with observed radiosonde relative humidity. Differences between retrieved and observed humidity were larger in the upper troposphere (about 20–25%) and smaller (<10%) in the lower troposphere. With three AMSU B channels, relative humidity could be retrieved at three distinct atmospheric levels, whereas radiosondes provide higher resolution vertical profiles. Therefore, more channels are needed on the wings of the strong water vapor absorption line. The proposed SAPHIR sounder has six channels around the 183.31 GHz water vapor absorption band and is therefore capable of retrieving humidity at six distinct atmospheric levels.