temperature correction for omi no2 slant column …€¦ · temperature correction for omi no 2...
TRANSCRIPT
Temperature correction for OMI NO2 slant column densities (SCD) [405-465 nm]
Revisiting the temperature correction factor for OMI
1. DOMINO-2: GOME data
fitted in 425-450 nm
1. Reevaluation
1. Reprocessed OMI NO2
SCDs (v2.0) fitted with cross-sections at 220°K, 240°K, 294°K.
1. Non-linear fit to the
nominal temperature correction factors of 220°K, 240°K, 294°K.
Uncertainty Analysis on OMI NO2 SCDs: • Intercomparison between DOMINO-2, DOMINO-3,
and QA4ECV SCDs
• Theory vs Practice (T5.5)
statistics on SCDs
DO
AS
Reprocessed OMI NO2 Slant Columns
DOMINO-3
1. Account for wavelength dependency of OMI slit function.
1. Wavelength
calibration window optimized.
1. Reference spectra
update. 1. Inclusion of
absorption by O2-O2 and H2Oliq.
QA4ECV
1. Account for viewing angle dependency of OMI slit function.
2. Different fitting equation (optical density vs. intensity fitting).
3. Inclusion of intensity offset term (1/I0).
DOMINO-3- QA4ECV SCDs
in agreement within 5%
QA4ECV SCDs reduced by 15%
Methodology
• Selection of 2° x 2° boxes over the Pacific
2°
2°
• AMF variability check
Slant Column Uncertainty: DOMINO-2, DOMINO-3, QA4ECV SCDs
6% lower SCD uncertainty for QA4ECV SCDs
SCD uncertainty =
Fitting uncertainty + Stripiness
Slant Column Uncertainty: DOMINO-2, DOMINO-3, QA4ECV SCDs
6% lower SCD uncertainty for QA4ECV SCDs
SCD uncertainty =
Fitting uncertainty + Stripiness
Theory vs Practice: Uncertainty of the Uncertainty
15%
DOMINO-2
45% 22%
QA4ECV
DOMINO-3
Higher agreement for DOMINO-3 & QA4ECV SCDs
Conclusions
• Revisit on the temperature correction for OMI NO2 SCDs; implementation in DOMINO-3: temperature sensitivity drops from 4.8%/°K to 3.2%/°K.
• Consistency between new DOMINO-3- and QA4ECV- NO2 SCDs within 5%.
• Statistical evaluations suggest NO2 SCD uncertainty reduced by 6%.
• Better match between DOAS-fit- and statistical NO2 SCD uncertainty;
better understanding of the DOAS fitting uncertainty.(∗∗)
T5.5 Slant column QA (evaluate theoretical precision against statistical uncertainty over
reference sector)
(∗∗)
Temperature Correction Factors so far
I.
III.
Boersma et al. [2002] & Bucsela et al. [2013] (405-465 nm):
II.
Boersma et al.[2004] (425-450 nm):
Present Work (425-450 nm)
SP2
DOMINO-2
DOMINO-3
Backup Slides
Simplest approach: Scaled Cross-Sections (X)
240°K– and 294°K- differential X are a scaled version of 220°K- differential X: A as a scaling factor
To = 220 K
A240 = 0.939
A294 = 0.789
Differences between Scaled X: contributor to the ‘DOAS fitting residuals’
min.
Backup Slides
A240 = 0.939 A294 = 0.789
Temperature correction factors, C, are considerably close to scaling factors, A.
C(T) magnitude dominated when #observations is highest
Backup Slides
Wavelength Window
A (240°K) C (240°K) A (294°K) C (294°K)
405 – 465 nm 0.939 0.938 0.789 0.785
425 – 450 nm 0.935 0.773
The different wavelength window
explains ≈ 25%
Good agreement between the scaling factors, especially for lower temperatures
Backup Slides
Underestimation
Overestimation
Figure courtesy of Henk Eskes (KNMI)
Backup Slides
AMF variability
Backup Slides
Backup Slides