Abstract

An obstacle to the simultaneous use of near-infrared (NIR) and thermal infrared (TIR) observations from the Measurements of Pollution in the Troposphere (MOPITT) instrument has been a lack of understanding of NIR radiance errors. Retrieval uncertainties produced by optimal estimation-based retrie... Show moreAn obstacle to the simultaneous use of near-infrared (NIR) and thermal infrared (TIR) observations from the Measurements of Pollution in the Troposphere (MOPITT) instrument has been a lack of understanding of NIR radiance errors. Retrieval uncertainties produced by optimal estimation-based retrieval algorithms used for satellite instruments like MOPITT are only meaningful if radiance error statistics are accurately quantified in the measurement error covariance matrix. MOPITT's gas correlation radiometers are subject to a unique form of "geophysical noise" due to the combined effects of (1) translational motion of the instrumental field of view during a single observation and (2) fine-scale spatial variability of surface radiative properties. We describe and demonstrate a new method for quantifying this source of error for each observation. Both TIR and NIR radiance errors due to this effect are highly variable, especially over land, but are qualitatively consistent with the variability of Moderate Resolution Imaging Spectroradiometer radiances in similar spectral bands. In addition, retrieval algorithm modifications are described which adjust the trade-off between smoothing error and retrieval noise within the optimal estimation framework. These modifications are necessary to fully exploit the information in MOPITT's NIR channels. A case study based on MOPITT observations over Minnesota demonstrates significant improvement in retrieval performance as the result of the retrieval algorithm modifications. Show less