Please use this identifier to cite or link to this item:
Study on homogenization of synthetic GNSS-Retrieved IWV time series and its impact on trend estimates with autoregressive noise
Title: Study on homogenization of synthetic GNSS-Retrieved IWV time series and its impact on trend estimates with autoregressive noise
Authors: Klos, AnnaVan Malderen, RoelandPottiaux, EricBock, OlivierBogusz, JanuszChimani, BarbaraElias, MichalGruszczynska, MartaGuijarro Pastor, José Antonio ORCID RESEARCHERID Autor AEMETZengin Kazancı, SelmaNing, Tong
Keywords: Integrated Water Vapour; Global Positioning System; International GNSS Service
Issue Date: 2017
Citation: EGU General Assembly (2017)
Abstract: A synthetic benchmark dataset of Integrated Water Vapour (IWV) was created within the activity of “Data homogenisation” of sub-working group WG3 of COST ES1206 Action. The benchmark dataset was created basing on the analysis of IWV differences retrieved by Global Positioning System (GPS) International GNSS Service (IGS) stations using European Centre for Medium-Range Weather Forecats (ECMWF) reanalysis data (ERA-Interim). Having analysed a set of 120 series of IWV differences (ERAI-GPS) derived for IGS stations, we delivered parameters of a number of gaps and breaks for every certain station. Moreover, we estimated values of trends, significant seasonalities and character of residuals when deterministic model was removed. We tested five different noise models and found that a combination of white and autoregressive processes of first order describes the stochastic part with a good accuracy. Basing on this analysis, we performed Monte Carlo simulations of 25 years long data with two different types of noise: white as well as combination of white and autoregressive processes. We also added few strictly defined offsets, creating three variants of synthetic dataset: easy, less complicated and fully complicated. The synthetic dataset we present was used as a benchmark to test various statistical tools in terms of homogenisation task. In this research, we assess the impact of the noise model, trend and gaps on the performance of statistical methods to detect simulated change points.
Description: Póster presentado en: EGU General Assembly celebrada del 23 al 28 de abril de 2017 en Viena, Austria.
Appears in Collections:EGU General Assembly

Files in This Item:
  File Description SizeFormat 
2,37 MBAdobe PDFThumbnail
Show full item record

Items in Arcimis are protected by Creative Commons License, unless otherwise indicated.

Arcimis Repository
Nota Legal Contacto y sugerencias