A physics-based atmospheric precipitable water vapor retrieval algorithm by synchronizing MODIS near-infrared and thermal infrared measurements

Abstract

This study proposed an innovative joint inversion algorithm that synchronized Moderate Resolution Imaging Spectroradiometer (MODIS) near-infrared (NIR) and thermal infrared (TIR) radiance data for accurate estimates of clear-sky precipitable water vapor (PWV). The algorithm consists of three parts: (1) simplifying the NIR radiative transfer equation by assuming linear reflectance change with wavelength in the 0.85–1.25 μm range, facilitating NIR water vapor absorption channel top-of-atmosphere (TOA) radiance simulation without explicit reflectance; (2) partial derivatives of NIR-TIR TOA radiance with respect to the background fields were derived by applying the one-term variational theorem to the radiative transfer equation; (3) optimization approach was employed to adjust the background fields, minimizing the discrepancy between simulated and observed NIR-TIR TOA radiances. The refined water vapor profile was integrated to derive PWV. Three years in situ measurements from the 473 GPS sites and 122 sun photometers in North America were utilized for PWV validation. Additionally, the MODIS MYD05 and MYD07 PWV products were validated using the same in situ measurements. Validation results indicated that the root mean square error (RMSE) of PWV retrieval using the NIR-TIR joint inversion algorithm ranged from 2.40 mm in summer to 1.67 mm in winter, and the mean bias and RMSE were − 0.55 mm and 2.08 mm, respectively, outperforming MODIS PWV products. The bias and RMSE were 3.84 mm and 4.86 mm for MYD05, and 0.41 mm and 4.60 mm for MYD07. Overall, the NIR-TIR joint inversion algorithm provides an effective way to generate comprehensive, long-term, high-resolution PWV data records.