On the Statistical Significance of Climatic Trends Estimated From GPS Tropospheric Time Series. Issue 19 (9th October 2018)

• Record Type: Journal Article
• Authors: Alshawaf, F.; Zus, F.; Balidakis, K.; Deng, Z.; Hoseini, M.; Dick, G.; Wickert, J.
• Is Part Of: Journal of geophysical research. Volume 123:Issue 19(2018)
• Page Start: 10, 967
• Page End: 10, 990
• Publication Date: 2018-10-09
• Subjects: Atmospheric physics -- Periodicals; Geophysics -- Periodicals; 551.5
• DOI: 10.1029/2018JD028703
• Languages: English
• ISSNs: 2169-897X

Abstract

Abstract: For more than two decades, Global Positioning System (GPS) tropospheric delays have successfully been exploited to monitor the tropospheric water vapor in near real time and reprocessing mode. Although reprocessed data are considered reliable for climatic research, it is important to address the often present gaps, inhomogeneities, and to use a proper model to describe the stochastic part of the time series so that trustworthy trends are estimated. Having relatively long time series, daily reprocessed tropospheric Zenith Total Delay, precipitable water vapor (PWV), and gradients from the Tide Gauge benchmark monitoring network are used in this work to estimate climatic trends. We use a first‐order autoregressive model AR(1) to describe the residuals after the trend estimation so that a correct trend uncertainty is estimated. Using the same model, we obtain the number of years of PWV data required to estimate statistically significant trends. For comparison, we produce tropospheric parameters at each Tide Gauge station based on ERA‐Interim refractivity fields. We found that 83% of 64 GPS stations show a positive PWV trend below 1 mm/decade independent of the time interval, with approximately half of the trends indicated significant. There is a strong correlation (86%) between the Global Navigation Satellite Systems and ERA‐Interim PWV trends. The trends tend to increase when moving east and south on the European map. The results show a percentage change of PWV of 3–8% per a degree Celsius rise in temperature. The number of years required to detect significant PWV trends varies between 30 and 40 years. Key Points: Climatic trends are estimated from water vapor time series GPS tropospheric products are adequately long to be used for climate studies The statistical significance of the trends should be estimated based on an autoregressive model rather than a white noise process