Journal of the Meteorological Society of Japan. Ser. II
Online ISSN : 2186-9057
Print ISSN : 0026-1165
ISSN-L : 0026-1165
Bias Correction of Multi-Sensor Total Column Ozone Satellite Data for 1978-2017
Author information
JOURNALS FREE ACCESS Advance online publication

Article ID: 2020-019


 This study constructs a merged total column ozone (TCO) dataset using 20 available satellite Level 2 TCO (L2SAT) datasets over 40 years from 1978 to 2017. The individual 20 datasets and the merged TCO dataset are corrected against ground-based Dobson and Brewer spectrophotometer TCO (GD) measurements. Two bias correction methods are used: simple linear regression (SLR) as a function of time and multiple linear regression (MLR) as a function of time, solar zenith angle, and effective ozone temperature. All of the satellite datasets are consistent with GD within ±2-3%, except for some degraded data from the Total Ozone Mapping Spectrometer/Earth Probe during a period of degraded calibration and from the Ozone Mapping and Profiling Suite (OMPS) provided from NOAA at an early stage of measurements. OMPS data provided from NASA show fairly stable L2SAT-GD differences. The Global Ozone Monitoring Experiment/MetOp-A and -B datasets show abrupt changes of approximately 8 DU coincident with the change of retrieval algorithm. For the TCO merged datasets created by averaging all coincident data located within a grid cell from the 20 satellite-borne TCO datasets, the differences between corrected and uncorrected TCOs by MLR are generally positive at lower latitudes where the bias correction increases TCO because of low effective ozone temperature. In the trend analysis, the difference between corrected and uncorrected TCO trends by MLR shows clear seasonal and latitudinal dependency, whereas such seasonal and latitudinal dependency is lost by SLR. The root mean square difference of L2SAT-GD for the uncorrected merged datasets, 8.6 DU, is reduced to 8.4 DU after correction using SLR and MLR. Therefore, the empirically corrected merged TCO datasets that are converted into time-series homogenization with high temporal-resolution are suitable as a data source for trend analyses as well as assimilation for long-term reanalysis.

Information related to the author
© The Author(s) 2020. This is an open access article published by the Meteorological Society of Japan under a Creative Commons Attribution 4.0 International (CC BY 4.0) license.
Previous article Next article