Presentation is loading. Please wait.

Presentation is loading. Please wait.

Intercomparison of tropospheric ozone measurements from TES and OMI

Similar presentations


Presentation on theme: "Intercomparison of tropospheric ozone measurements from TES and OMI"— Presentation transcript:

1 Intercomparison of tropospheric ozone measurements from TES and OMI
Lin Zhang1, Daniel J. Jacob1, Xiong Liu2,3, Jennifer A. Logan1, and the TES Science Team 1. Harvard University 2. GEST, UMBC 3. Harvard-Smithsonian Center for Astrophysics I will show you the intercomparison between TES and OMI. The ozone data set from OMI is from Xiong Liu at UMBC. TES Science Team Meeting (Feb.24, 2009)

2 Concurrent ozone measurements from IR and UV
Nadir-looking instrument measuring backscattered solar radiation ( nm) Daily global coverage at a spatial resolution of 13 x 24 km2 at nadir Retrieve ozone at 24 ~2.5 km layers TES Infrared-imaging Fourier transform spectrometer ( µm) 16 orbits of nadir vertical profiles at a spatial resolution of 5x8 km2 and spaced 1.6° along the orbit track every other day. Retrievals of ozone and CO at 67 levels from surface up to 0.1 hPa, version 3 data Both TES and OMI are aboard Aura. They provide concurrent measurements of tropospheric ozone at 1:30 pm, but from different spectrum region. OMI measures backscattered UV radiation, while TES measurements IR. Do they provide consistent measurements of tropospheric ozone? What can we learn by comparing both measurements with chemical transport models?

3 1.8 DOFS in the troposphere 1.0 DOFS in the troposphere
Vertical sensitivity of TES and OMI ozone retrievals Both retrievals are obtained from the optimal estimation method [Rodgers, 2000]: Retrieval Averaging kernel Aug 6, 2009 58W, 28N Both of the measurements are obtained from the optimal estimation method. The retrieval can be expressed by this equation, where x_head is the satellite ozone retrieval; x is the true profile, and x_a is a priori representing our current understanding. The averaging kernel (A) shows the vertical sensitivity of the retrieval to the true profiles. As John’s paper shows that combining the two spectra measurements would greatly improve the ability to retrieve boundary layer ozone. In this study, we will focus on the 500 hPa where both TES and OMI are relatively sensitive. 1.8 DOFS in the troposphere 1.0 DOFS in the troposphere

4 2006 ozone at 500 hPa averaged on 4ox5o resolution
Tropospheric ozone from TES and OMI 2006 ozone at 500 hPa averaged on 4ox5o resolution Here shows TES and OMI measurements of tropospheric ozone at 500 hPa where both instruments are relatively sensitive. This shows the year 2006 averaged over each season. Notice here all the data are reprocessed with a single fixed a priori, which means all the geographic and seasonal variations are obtained from the satellite information and differences between the instruments reflect the measurement itself. TES and OMI are showing very similar geographic features and seasonal variability. They both show the high ozone regions in the summertime of northern hemisphere, and also the evolution of the biomass burning regions over the Africa. But we do see some differences between the two. OMI concentrations are lower than TES at these high ozone regions. OMI observations are selected along TES pixels. The data are reprocessed with a single fixed a priori.

5 2006 ozone at 500 hPa averaged on 4ox5o resolution
Tropospheric ozone from TES, OMI and GEOS-Chem 2006 ozone at 500 hPa averaged on 4ox5o resolution Here we add model simulations. The model simulation is in 4x5 resolution, and sampled along TES observations at the observing time, and then smoothed by corresponding averaging kernels. The one with OMI averaging kernel is overall lower than the one the TES averaging kernels in high ozone regions. This shows that some of the difference can be explained by different vertical sensitivities. The data and model results are reprocessed with a single fixed a priori. GEOS-Chem simulation in 4ox5o resolution is sampled along the TES/OMI pixels, and then smoothed by corresponding averaging kernels.

6 Validation with ozonesonde
Ozonesonde data from , available at AURA AVDC Coincidence Criteria: < 2o longitudes & Latitudes, < 10 hours 60oS-60oN, 500 hPa: TES has a positive bias of 5.4 ± 9 ppbv OMI has a positive bias of ± 5 ppbv This figure put together the validation of TES and OMI ozone measurements at 500 hPa. We use the ozonesonde from As we can see from the sparse sonde comparison, in the middle troposphere, both TES and OMI have a positive bias of 3-5 ppbv. There is some variability around the world, but we can not identify any geographical structure from the comparison.

7 Methods for the intercomparison
Sparse in time and space Validation Validation Generally we cannot directly compare two satellite measurements, because they have different vertical sensitivity. There are three methods to do the comparison. The first method is comparison through the sonde measurements. We can validate both satellite measurements with the sonde, and then infer their differences, but as we know the sonde measurements are very limited in time and space, and hence the comparison has a large uncertainty. 1. Sonde method: Validation with ozonesonde measurements

8 Methods for the intercomparison
Sparse in time and space Validation Validation The second method is direct comparison after adjusting their different a priori and averaging kernels. This is introduced by Rodgers and Conner. The method is to use the retrieval with higher vertical sensitivities smoothed by the retrieval with lower vertical sensitivity. In this case, we apply OMI averaging kernels to TES retrievals, and then compare with OMI retrievals. This method works well when one retrieval has a very high sensitivity. But here both TES and OMI have quite low vertical sensitivity, so this method may also have large uncertainties. Direct comparison (Rodgers and Conner, 2003) 1. Sonde method: Validation with ozonesonde measurements 2. Direct method: Compare OMI/TES directly after considering their different a priori constrains and vertical sensitivity (Apply OMI averaging kernels to TES retrievals)

9 Methods for the intercomparison
Sparse in time and space Validation Validation Evaluation Interpretation Evaluation Evaluation Interpretation Interpretation The third method is to use chemical transport models as comparison platform. we use both TES and OMI measurements to evaluate the model, and see whether they provide consistent constrains on the model simulation. Further we can use the model to interpret the observations. So now we are not just use the satellite observations to compare with the model, but can use model to transfer information between two and provide confidence to the satellite measurements. Direct comparison (Rodgers and Conner, 2003) 1. Sonde method: Validation with ozonesonde measurements 2. Direct method: Compare OMI/TES directly after considering their different a priori constrains and vertical sensitivity (Apply OMI averaging kernels to TES retrievals) 3. CTM method: Use GEOS-Chem CTM as a comparison platform

10 What do the methods actually compare?
Let Sonde method: TES – sonde/TES AK = bTES OMI – sonde/OMI AK = bOMI TES – OMI = bTES – bOMI 2. Direct method: AOMIbTES – bOMI + AOMI(ATES – I)(X – Xa) (Rodgers and Conner, 2003) Here we separate the systematic bias from the total observational errors of the retrievals (b_tes and b_omi). 3. CTM method: (TES – CTM/TES AK) – (OMI – CTM/OMI AK) = bTES – bOMI + (ATES – AOMI)(X – XCTM)

11 Quantify differences between TES and OMI
1. TES – OMI (sonde) = bTES – bOMI 2. TES (OMI AK) – OMI = AOMIbTES – bOMI + AOMI(ATES – I)(X – Xa) 3. TES – OMI (GC) = bTES – bOMI + (ATES – AOMI)(X – XCTM) 500 hPa 160 TES/OMI/sonde coincidences for 2006 CTM method Direct method In the direct method, slopes < 1 reflect application of AOMI reduce the sensitivity to diagnose the bias. The CTM method preserves the variability of the differences in the comparison. TES – OMI Sonde method [ppbv] We can test the three methods by using the 160 coincidences for here shows the equations for the three comparison methods. The left panels show the comparison of the direct method with the sonde method. The slopes are generally smaller than 1, This reflects smoothing the TES bias with the OMI averaging kernels will reduce the ability to diagnose the differences. Especially at 850 hPa, OMI has very low sensitivity. In the CTM method, the right panels, it shows a high correlation, and the slopes are close to 1. The small offset from 1:1 line reflects CTM bias. But what’s more important, all the variability are preserved. 850 hPa Direct method CTM method TES – OMI Sonde method [ppbv]

12 Difference between TES and OMI at 500 hPa
TES – OMI Sonde method Direct method CTM method Now broaden the picture. Mostly at 500 hPa the bias is smaller than 10 ppbv, and the average is close to 0. The average bias from sonde is higher, because they are averaged differently. TES – OMI Mean ± 1 sigma 2.6 ± 6.6 ppbv -0.1 ± 3.6 ppbv -0.3 ± 5.0 ppbv Total error for both TES and OMI

13 Difference between TES and OMI at 850 hPa
TES – OMI Sonde method Direct method CTM method In the lower troposphere, at 850 hPa, we see large differences among the three comparisons. The direct comparison doesn’t show any of the large differences between TES and OMI as shown from the other two methods. As we proved, in the direct method, application of the OMI averaging kernels to TES biases will reduce the variability of the differences. TES – OMI Mean ± 1 sigma 3.3 ± 6.8 ppbv -0.3 ± 1.9 ppbv 2.7 ± 5.5 ppbv

14 For 2006 and averaged on 4ox5o resolution
Differences with GEOS-Chem at 500 hPa For 2006 and averaged on 4ox5o resolution Minus 3 ppbv from both TES and OMI measurements. Regions with the bias between TES and OMI larger than 10 ppbv are masked as black. GC – sonde GC/TES AK – (TES– 3) GC/OMI AK – (OMI– 3) Now we show the comparison of TES and OMI ozone measurements with GEOS-Chem at 500 hPa. Here we minus 3 ppbv from both TES and OMI measurements, and we only show regions where TES and OMI are within 10 ppbv.

15 For 2006 and averaged on 4ox5o resolution
Differences with GEOS-Chem at 500 hPa For 2006 and averaged on 4ox5o resolution Minus 3 ppbv from both TES and OMI measurements. Regions with the bias between TES and OMI larger than 10 ppbv are masked as black. GC – sonde GC/TES AK – (TES– 3) GC/OMI AK – (OMI– 3) We can see both TES and OMI show model underestimation of ozone concentrations over the over the tropical regions, some of them are due to low biomass burning emissions in the model. We also see an overestimation over the North America in the summer. The sonde comparison show model largely overestimates ozone in the wintertime of the northern hemisphere. It is much weaker in the satellite comparison, because they both have lower sensitivity in the winter.

16 Thank you

17 2006 ozone at 500 hPa averaged on 4ox5o resolution
Tropospheric ozone measurements from TES and OMI 2006 ozone at 500 hPa averaged on 4ox5o resolution OMI observations sample along the TES pixels OMI observations are sampled along the TES pixels. Convert the different a priori to a fixed a priori:

18 Vertical sensitivity of TES and OMI ozone retrievals
Both retrievals are obtained from the optimal estimation method [Rodgers, 2000]: Averaging kernel July 2006 TES Degrees of Freedom for tropospheric ozone OMI The first question we may have is whether the two ozone retrieval have different vertical sensitivity. Zonal average of Diagonal terms of AK


Download ppt "Intercomparison of tropospheric ozone measurements from TES and OMI"

Similar presentations


Ads by Google