Evaluating Long-Term HIRS-based Near Surface Marine Air Temperature Dataset

Satellite instruments such as the High-resolution Infrared Radiation Sounder series (HIRS) onboard polar-orbiting environmental satellites operated by NOAA and other international agencies since the late 1970s are used to monitor the temperature of the Earth including the surface temperature. These provide data and additional coverage for atmospheric models and climatologies where traditional in situ measurements are uneven and sparse. A global dataset of atmospheric temperature and humidity profiles has been derived based on inter-satellite calibrated HIRS data using a neural network approach. The HIRS-based temperature profile data provides temperature estimates at 10 different atmospheric pressure levels and near-surface for both ocean and land surfaces globally. To ensure the quality of the long-term satellite-based temperature estimates for climate and marine applications, this project focuses on the validation of a subset of the retrievals that includes temperatures at the two-meter level over sea surface and the skin temperature at the sea surface for the period of 1986-2017. It uses buoy and ship data from the latest version of the International Comprehensive Ocean-Atmosphere Data Set (ICOADS) that has been adjusted to a common height above the sea surface. Statistics and visualizations are used to compare the HIRS retrievals to both the original and height adjusted ICOADS dataset. The project has found strong agreement between the ICOADS dataset and HIRS retrievals and resulted in the improvement of the HIRS temperature profile retrievals by removing erroneous neural network retrievals caused by the missing data from HIRS brightness temperature in certain satellites. After the removal of the erroneous HIRS-based temperature estimates, the bias of HIRS-based surface temperature estimates for each satellite was found to be 0.302 ± 0.339°C and the RMSE was found to be 3.155 ± 0.475°C.