This work focuses on estimating trends in ozone partial pressure, temperature, and water vapour in the upper troposphere/lower stratosphere (UT/LS) region using a new unified ozonesounding profile database and a novel homogenised dataset named RHARM (Radiosounding HARMonization). Studying temperature, water vapour, and ozone trends is key for studying climate change and climate variability. Temperature changes in the UT/LS are related to both internal processes, for example, changes in sea surface temperature (SST) and external forcing, such as greenhouse gases (GHGs) and ozone-depleting substances (ODS) (Randel et al., 2009) . Various studies have been conducted to estimate the trends for different climate variables at a regional and global scale. The radiative effects of rising GHGs and changes in stratospheric ozone as a response to human emissions of ODS have led to net warming of the troposphere and cooling of the stratosphere (Hartmann et al., 2013) . The impact of ODS on tropical upwelling, revealed by the absence of lower-stratospheric cooling, has been reported since 1998 (Polvani et al., 2017). Moreover, an increased tropopause temperature in the period 2001–2011 associated with a weaker tropopause inversion layer, due to the weakened upwelling in the Tropics, was found using Global Positioning System Radio Occultation (GNSS-RO) data and simulations with the National Center for Atmospheric Research's Whole Atmosphere Community Climate Model (WACCM). Such changes in the thermal structure of the UT/LS may have important implications for climate, such as a possible rise in water vapour in the lower stratosphere (Wang et al., 2013). More recently, balloon-borne radiation measurements proved that the stratosphere is warming after years of cooling (Philipona et al., 2016). Whether a slowdown or change in temperature sign in the UT/LS will persist in the future is an open question. This work estimates ozone trends using balloon-borne measurements from three existing datasets: the Southern Hemisphere Additional OZonesondes (SHADOZ), Network for the Detection of Atmospheric Composition Change (NDACC), and World Ozone and Ultraviolet Radiation Data Centre (WOUDC) networks, developed respectively by NASA, NOAA, with the collaboration of many different institutes and PIs around the world, and WMO. These datasets are merged to provide appropriate data coverage at different latitudes and increase sampling for improving the calculation of anomalies and trends in the ozone concentration at the global scale. The resulting unified dataset removes duplicated profiles. Duplication for ozonesounding profiles often occurs when measurements from the same station are submitted to several networks, which in theory should be identical but are often provided for different periods, using different data formats, and providing a different amount of individual data points. Metadata may also differ. This also means that the different networks do not always report the same number of ozone levels for the same profile. A range of selection criteria has been applied to overcome this issue and harmonise the existing ozonesounding datasets to refine the quality and ensure the identification of outliers by applying a series of quality checks (QC) listed below: Plausibility checks: reported values should be within plausible physical range; Completeness check: on a monthly basis to verify that all variables are complete; Outliers check: using the Inter-Quartile Range method as follows: median-3∙IQR≤observation≤median+3∙IQR, Vertical coverage checks: on a monthly basis to verify if ozone profiles reach 10 hPa; Vertical completeness checks: to ensure a minimum number of reports are available for each vertical region covered by the ozonesoundings; Statistics of missing values: to check the coherency with the source datasets. The unified dataset is then grouped according to their monthly coverage to quantify sampling uncertainties in the trend calculation. The 155 available stations are separated into three different clusters: Long Coverage (LC): 26 stations (with a data time series of at least 20 years). Medium coverage (MC): 23 stations (with a data time series between 10 and 20 years). Short coverage (SC): 106 stations (with data time series of less than 10 years). The first two clusters are the only ones with sufficient data coverage for estimating anomalies and trends. The latter is estimated from monthly mean anomalies using LC, MC, and their combination. Different regression methods are used for estimating trends to provide for quantification of structural uncertainties in the trend calculation, including: Least-square linear regression (Reinsel et al. 2002) ; LOTUS regression (Petropavlovskikh et al., 2019; Godin-Beekmann et al., 2022 ); Least Absolute Deviation (LAD) regression (Rice and White, 1964; Barrodale, 1968; Wong and Schneider Jr, 1989; Calitz and Rüther, 1996; Santer et al., 2000 ); Theil-Sen regression (Theil, 1950 ; Siegel and Benson, 1982 ; Helsel and Hirsch, 1992 ); The Mann-Kendal (MK) test (Kendall, 1975; Mann, 1945 ) is also used to statistically assess if there is a significant trend of the variable of interest over time. The comparison shows that comparing trends estimated from LC data and a combination of LC and MC provides very similar percentage trends. For the 50-1 hPa layer, for example, the differences, comparing the different regressors, range from 0.6%/decade to 1.2%/decade. These represent an estimate of sampling uncertainty in cases where the trends are significant. Therefore, the LC data, representing the highest quality data according to the above criteria, is used to estimate trends. The estimates on this cluster show, for the Northern Hemisphere mid-latitudes (NH): a negative trend of 5% for the period pre-2000 at 50-1 hPa layer, reducing to 1% for the period post-2000 at 50-1 hPa, a negative trend of 10% for the period pre-2000 at 100-50 hPa, in contrast to a positive trend of 4% for the period post-2000 at 100-50 hPa. For the Tropics (TR) sector, for the period pre-2000, a positive trend of about 8% at 50-1 hPa, in contrast to a positive trend of 2% at 50-1 hPa for post-2000, for the period pre-2000, a positive trend reached 10% at 100-50 hPa on the other hand, for the post-2000 period the estimated trend did not pass the MK test so it is not significant. The estimates for the NH sector are consistent with those presented by Petropavlovskikh et al. (2019): for the pre-2000 time series, in the lower stratosphere, there is a negative trend of 5% per decade that also reaches a negative trend of 10% at 100 hPa; for the post-2000 time series, in the lower stratosphere, there is a small negative trend of 1%, and at 100 hPa, a negative trend of 2%. This value has an uncertainty of ±7%, which makes the result produced in this work (about 4% positive trend) within the uncertainty range. For the TR, the pre-2000 time series, in the lower stratosphere, the trends presented in Petropavlovskikh et al. (2019) show a negative trend of 2% per decade and more, reaching a negative trend of 10% at 100 hPa, in contrast to the estimate shown in this work that presents a positive trend of about 8% in the lower stratosphere, reaching 10% at 100 hPa. This discrepancy is probably due to the small number of stations (only 4) used in this work for trend estimation. However, in the post-2000 time series, there is a positive trend of 2% in the lower stratosphere, as also shown in this work, and, at 100 hPa, a positive trend of 8%. This value, like the NH, has an uncertainty limit of ±7%, however, the trend evaluated in this work for the 100-50 hPa layer was not found to be significant by the MK test. Finally, for temperature and water vapour trends, this study uses a novel dataset, named Radiosounding HARMonization (RHARM), providing a homogenized data set of temperature, humidity, and wind profiles along with an estimation of the measurement uncertainties for 697 radiosounding stations globally. The RHARM method has been used to adjust twice daily (0000 and 1200 UTC) radiosonde data holdings at 16 pressure levels in the range of 1,000–10 hPa, from 1978 to the present, provided by the Integrated Global Radiosonde Archive. Relative humidity data are limited to 250 hPa. The applied adjustments are interpolated to all reported levels. RHARM is the first data set to provide a homogenized time series with an estimation of the observational uncertainty at each sounding pressure level. By construction, RHARM-adjusted fields are not affected by cross-contamination of biases across stations and are fully independent of reanalysis data. RHARM shows warming trends of 0.39 K/decade at 300 hPa in the NH and 0.25 K/decade in the TR. The RHARM adjustments also reduce differences with the European Centre for Medium-Range Weather Forecast ERA5 reanalysis, with the strongest effect in the NH for temperature and relative humidity.

Study of ozone, temperature, and water vapour in the UT/LS using upper-air measurements / Marra, Fabrizio. - (2024 Feb 26).

Study of ozone, temperature, and water vapour in the UT/LS using upper-air measurements.

MARRA, FABRIZIO
2024-02-26

Abstract

This work focuses on estimating trends in ozone partial pressure, temperature, and water vapour in the upper troposphere/lower stratosphere (UT/LS) region using a new unified ozonesounding profile database and a novel homogenised dataset named RHARM (Radiosounding HARMonization). Studying temperature, water vapour, and ozone trends is key for studying climate change and climate variability. Temperature changes in the UT/LS are related to both internal processes, for example, changes in sea surface temperature (SST) and external forcing, such as greenhouse gases (GHGs) and ozone-depleting substances (ODS) (Randel et al., 2009) . Various studies have been conducted to estimate the trends for different climate variables at a regional and global scale. The radiative effects of rising GHGs and changes in stratospheric ozone as a response to human emissions of ODS have led to net warming of the troposphere and cooling of the stratosphere (Hartmann et al., 2013) . The impact of ODS on tropical upwelling, revealed by the absence of lower-stratospheric cooling, has been reported since 1998 (Polvani et al., 2017). Moreover, an increased tropopause temperature in the period 2001–2011 associated with a weaker tropopause inversion layer, due to the weakened upwelling in the Tropics, was found using Global Positioning System Radio Occultation (GNSS-RO) data and simulations with the National Center for Atmospheric Research's Whole Atmosphere Community Climate Model (WACCM). Such changes in the thermal structure of the UT/LS may have important implications for climate, such as a possible rise in water vapour in the lower stratosphere (Wang et al., 2013). More recently, balloon-borne radiation measurements proved that the stratosphere is warming after years of cooling (Philipona et al., 2016). Whether a slowdown or change in temperature sign in the UT/LS will persist in the future is an open question. This work estimates ozone trends using balloon-borne measurements from three existing datasets: the Southern Hemisphere Additional OZonesondes (SHADOZ), Network for the Detection of Atmospheric Composition Change (NDACC), and World Ozone and Ultraviolet Radiation Data Centre (WOUDC) networks, developed respectively by NASA, NOAA, with the collaboration of many different institutes and PIs around the world, and WMO. These datasets are merged to provide appropriate data coverage at different latitudes and increase sampling for improving the calculation of anomalies and trends in the ozone concentration at the global scale. The resulting unified dataset removes duplicated profiles. Duplication for ozonesounding profiles often occurs when measurements from the same station are submitted to several networks, which in theory should be identical but are often provided for different periods, using different data formats, and providing a different amount of individual data points. Metadata may also differ. This also means that the different networks do not always report the same number of ozone levels for the same profile. A range of selection criteria has been applied to overcome this issue and harmonise the existing ozonesounding datasets to refine the quality and ensure the identification of outliers by applying a series of quality checks (QC) listed below: Plausibility checks: reported values should be within plausible physical range; Completeness check: on a monthly basis to verify that all variables are complete; Outliers check: using the Inter-Quartile Range method as follows: median-3∙IQR≤observation≤median+3∙IQR, Vertical coverage checks: on a monthly basis to verify if ozone profiles reach 10 hPa; Vertical completeness checks: to ensure a minimum number of reports are available for each vertical region covered by the ozonesoundings; Statistics of missing values: to check the coherency with the source datasets. The unified dataset is then grouped according to their monthly coverage to quantify sampling uncertainties in the trend calculation. The 155 available stations are separated into three different clusters: Long Coverage (LC): 26 stations (with a data time series of at least 20 years). Medium coverage (MC): 23 stations (with a data time series between 10 and 20 years). Short coverage (SC): 106 stations (with data time series of less than 10 years). The first two clusters are the only ones with sufficient data coverage for estimating anomalies and trends. The latter is estimated from monthly mean anomalies using LC, MC, and their combination. Different regression methods are used for estimating trends to provide for quantification of structural uncertainties in the trend calculation, including: Least-square linear regression (Reinsel et al. 2002) ; LOTUS regression (Petropavlovskikh et al., 2019; Godin-Beekmann et al., 2022 ); Least Absolute Deviation (LAD) regression (Rice and White, 1964; Barrodale, 1968; Wong and Schneider Jr, 1989; Calitz and Rüther, 1996; Santer et al., 2000 ); Theil-Sen regression (Theil, 1950 ; Siegel and Benson, 1982 ; Helsel and Hirsch, 1992 ); The Mann-Kendal (MK) test (Kendall, 1975; Mann, 1945 ) is also used to statistically assess if there is a significant trend of the variable of interest over time. The comparison shows that comparing trends estimated from LC data and a combination of LC and MC provides very similar percentage trends. For the 50-1 hPa layer, for example, the differences, comparing the different regressors, range from 0.6%/decade to 1.2%/decade. These represent an estimate of sampling uncertainty in cases where the trends are significant. Therefore, the LC data, representing the highest quality data according to the above criteria, is used to estimate trends. The estimates on this cluster show, for the Northern Hemisphere mid-latitudes (NH): a negative trend of 5% for the period pre-2000 at 50-1 hPa layer, reducing to 1% for the period post-2000 at 50-1 hPa, a negative trend of 10% for the period pre-2000 at 100-50 hPa, in contrast to a positive trend of 4% for the period post-2000 at 100-50 hPa. For the Tropics (TR) sector, for the period pre-2000, a positive trend of about 8% at 50-1 hPa, in contrast to a positive trend of 2% at 50-1 hPa for post-2000, for the period pre-2000, a positive trend reached 10% at 100-50 hPa on the other hand, for the post-2000 period the estimated trend did not pass the MK test so it is not significant. The estimates for the NH sector are consistent with those presented by Petropavlovskikh et al. (2019): for the pre-2000 time series, in the lower stratosphere, there is a negative trend of 5% per decade that also reaches a negative trend of 10% at 100 hPa; for the post-2000 time series, in the lower stratosphere, there is a small negative trend of 1%, and at 100 hPa, a negative trend of 2%. This value has an uncertainty of ±7%, which makes the result produced in this work (about 4% positive trend) within the uncertainty range. For the TR, the pre-2000 time series, in the lower stratosphere, the trends presented in Petropavlovskikh et al. (2019) show a negative trend of 2% per decade and more, reaching a negative trend of 10% at 100 hPa, in contrast to the estimate shown in this work that presents a positive trend of about 8% in the lower stratosphere, reaching 10% at 100 hPa. This discrepancy is probably due to the small number of stations (only 4) used in this work for trend estimation. However, in the post-2000 time series, there is a positive trend of 2% in the lower stratosphere, as also shown in this work, and, at 100 hPa, a positive trend of 8%. This value, like the NH, has an uncertainty limit of ±7%, however, the trend evaluated in this work for the 100-50 hPa layer was not found to be significant by the MK test. Finally, for temperature and water vapour trends, this study uses a novel dataset, named Radiosounding HARMonization (RHARM), providing a homogenized data set of temperature, humidity, and wind profiles along with an estimation of the measurement uncertainties for 697 radiosounding stations globally. The RHARM method has been used to adjust twice daily (0000 and 1200 UTC) radiosonde data holdings at 16 pressure levels in the range of 1,000–10 hPa, from 1978 to the present, provided by the Integrated Global Radiosonde Archive. Relative humidity data are limited to 250 hPa. The applied adjustments are interpolated to all reported levels. RHARM is the first data set to provide a homogenized time series with an estimation of the observational uncertainty at each sounding pressure level. By construction, RHARM-adjusted fields are not affected by cross-contamination of biases across stations and are fully independent of reanalysis data. RHARM shows warming trends of 0.39 K/decade at 300 hPa in the NH and 0.25 K/decade in the TR. The RHARM adjustments also reduce differences with the European Centre for Medium-Range Weather Forecast ERA5 reanalysis, with the strongest effect in the NH for temperature and relative humidity.
26-feb-2024
Ozone; temperature; water vapor; trend; ozonesounding; radiosounding; UT/LS; SHADOZ; NDACC; WOUDC; RHARM
Study of ozone, temperature, and water vapour in the UT/LS using upper-air measurements / Marra, Fabrizio. - (2024 Feb 26).
File in questo prodotto:
File Dimensione Formato  
Thesis_PhD_Fabrizio_Marra.pdf

accesso aperto

Descrizione: Tesi di dottorato
Tipologia: Tesi di dottorato
Licenza: Non definito
Dimensione 12.62 MB
Formato Adobe PDF
12.62 MB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11563/177616
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact