Abstract. Estimates of PM 2.5 distributions based on satellite data depend critically on an established relation between AOD and ground level PM 2.5 . In this study we performed an experiment at Cabauw to establish a relation between AOD and PM 2.5 for the Netherlands. A first inspection of the AERONET L1.5 AOD and PM 2.5 data showed a low correlation between the two properties. The AERONET L1.5 showed relatively many observations of high AOD values paired to low PM 2.5 values, which hinted cloud contamination. Various methods were used to detect cloud contamination in the AERONET data to substantiate this hypothesis. A cloud screening method based on backscatter LIDAR observations was chosen to detect cloud contaminated observations in the AERONET L1.5 AOD. A later evaluation of AERONET L2.0 showed that the most data that are excluded in the update from L1.5 to L2.0 were also excluded by our cloud screening, which provides confidence in both our cloud-screening method as well as the final screening in the AERONET procedure. The use of LIDAR measurements in conjunction with the CIMEL AOD data is regarded highly beneficial. Contra-intuitively, the AOD to PM 2.5 relationship was shown to be insensitive to inclusion of the mixed layer height. The robustness of the relation improves dependent on the time window during the day towards noon. The final relation found for Cabauw is PM 2.5 =124.5×AOD−0.34 and is valid for fair weather conditions. The relationship found between bias corrected MODIS AOD and PM 2.5 at Cabauw is very similar to the analysis based on the much larger datasetCorrespondence to: M. Schaap (martijn.schaap@tno.nl) from ground based data only. We applied the relationship to a MODIS composite map to assess the PM 2.5 distribution over the Netherlands for the first time. The verification of the derived map is difficult because ground level artefact free PM 2.5 data are lacking. The validity and utility of our proposed mapping methodology should be further investigated.

Abstract. To acquire daily estimates of PM2.5 distributions based on satellite data one depends critically on an established relation between AOD and ground level PM2.5. In this study we aimed to experimentally establish the AOD-PM2.5 relationship for the Netherlands. For that purpose an experiment was set-up at the AERONET site Cabauw. The average PM2.5 concentration during this ten month study was 18 μg/m3, which confirms that the Netherlands are characterised by a high PM burden. A first inspection of the AERONET level 1.5 (L1.5) AOD and PM2.5 data at Cabauw showed a low correlation between the two properties. However, after screening for cloud contamination in the AERONET L1.5 data, the correlation improved substantially. When also constraining the dataset to data points acquired around noon, the correlation between AOD and PM2.5 amounted to R2=0.6 for situations with fair weather. This indicates that AOD data contain information about the temporal evolution of PM2.5. We had used LIDAR observations to detect residual cloud contamination in the AERONET L1.5 data. Comparison of our cloud-screed L1.5 with AERONET L2 data that became available near the end of the study showed favorable agreement. The final relation found for Cabauw is PM2.5=124.5*AOD–0.34 (with PM2.5 in μg/m3) and is valid for fair weather conditions. The relationship determined between MODIS AOD and ground level PM2.5 at Cabauw is very similar to that based on the much larger dataset from the sun photometer data, after correcting for a systematic overestimation of the MODIS data of 0.05. We applied the relationship to a MODIS composite map to assess the PM2.5 distribution over the Netherlands. Spatial dependent systematic errors in the MODIS AOD, probably related to variability in surface reflectance, hamper a meaningful analysis of the spatial distribution of PM2.5 using AOD data at the scale of the Netherlands.

[1] This study investigates tropical Kelvin wave signatures in the total ozone column data from the Global Ozone Monitoring Experiment (GOME) instrument. A new approach for spectral analysis is introduced by generalizing an unequally spaced data technique from one to two dimensions. This enables the handling of satellite data containing gaps. The simple statistical behavior of the method furthermore allows an easy determinination of the statistical significance of any observed spectral features. Seven years of GOME data (1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002) have been analyzed in which we have identified three periods of high Kelvin wave activity in 1996, 1998, and 2000. The periods are in conjunction with westward equatorial zonal winds at 30 hPa and show eastward propagating waves 1-2 with periods of $12-15 days. The induced Kelvin wave signatures in the ozone concentrations are around 2-4 DU peak-to-peak and can be attributed to ''slow'' Kelvin waves. The results are shown to be significant. Our study provides an important contribution to the study of Kelvin waves by introducing the bidimensional unequally spaced data spectral analysis and is the first to demonstrate the potential of the GOME ozone data set to contribute to a global description of equatorial Kelvin wave activity.

No abstract

[1] This study investigates the vertical structure of the Kelvin wave signals previously found in total ozone column measurements from the Global Ozone Monitoring Experiment (GOME) instrument. For this, zonal wind and temperature measurements from the European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis data set are analyzed by using the same bidimensional spectral method as was used to analyze the GOME total ozone columns. These fields are available on 60 levels from the surface to 0.1 hPa. For the three high Kelvin wave activity periods identified in the GOME data we found spectral features in the ECMWF fields associated with Kelvin waves with zonal wave numbers 1 or 2 and periods around 15-20 days. These characteristics correspond to the characteristics of the Kelvin waves detected in GOME. The signals are significant throughout the lower stratosphere between $100 and 10 hPa and, depending on the period, are largest around 15, 45, or 65 hPa. There is a good correlation between the Kelvin wave signals in the ECMWF zonal wind and temperature and the GOME total ozone column. The induced fluctuations in zonal wind and temperature are, respectively, up to 8 m/s and 2 K. From these induced zonal wind fluctuations, expected total ozone column fluctuations of around 1 DU are calculated, corresponding to the ozone fluctuations found in the GOME data. The results indicate that the analyzed total ozone column fluctuations are mainly caused by transport effects in the lower stratosphere. This study shows that combined use of ECMWF Re-Analysis data and GOME ozone columns provides a possibility to study the three-dimensional structure of Kelvin wave activity.