TY - JOUR
T1 - Characterizing and minimizing the effects of noise in tide gauge time series
T2 - Relative and geocentric sea level rise around Australia
AU - Burgette, Reed J.
AU - Watson, Christopher S.
AU - Church, John A.
AU - White, Neil J.
AU - Tregoning, Paul
AU - Coleman, Richard
PY - 2013/7
Y1 - 2013/7
N2 - We quantify the rate of sea level rise around the Australian continent from an analysis of tidegauge and Global Positioning System (GPS) data sets. To estimate the underlying linear ratesof sea level change in the presence of significant interannual and decadal variability (treatedhere as noise), we adopt and extend a novel network adjustment approach. We simultaneouslyestimate time-correlated noise as well as linear model parameters and realistic uncertaintiesfrom sea level time series at individual gauges, as well as from time-series differences computedbetween pairs of gauges. The noise content at individual gauges is consistent with acombination of white and time-correlated noise. We find that the noise in time series from thewestern coast of Australia is best described by a first-order Gauss-Markov model, whereas eastcoast stations generally exhibit lower levels of time-correlated noise that is better describedby a power-law process. These findings suggest several decades of monthly tide gauge dataare needed to reduce rate uncertainties to <0.5mmyr-1 for undifferenced single site timeseries with typical noise characteristics. Our subsequent adjustment strategy exploits the moreprecise differential rates estimated from differenced time series from pairs of tide gaugesto estimate rates among the network of 43 tide gauges that passed a stability analysis. Weestimate relative sea level rates over three temporal windows (1900-2011, 1966-2011 and1993-2011), accounting for covariance between time series. The resultant adjustment reducesthe rate uncertainty across individual gauges, and partially mitigates the need for century-scaletime series at all sites in the network. Our adjustment reveals a spatially coherent pattern ofsea level rise around the coastline, with the highest rates in northern Australia. Over the timeperiods beginning in 1900, 1966 and 1993, we find weighted average rates of sea level riseof 1.4 ± 0.6, 1.7 ± 0.6 and 4.6 ± 0.8mmyr-1, respectively. While the temporal pattern ofthe rate estimates is consistent with acceleration in sea level rise, it may not be significant, asthe uncertainties for the shorter analysis periods may not capture the full range of temporalvariation. Analysis of the available continuous GPS records that have been collected within80 km of Australian tide gauges suggests that rates of vertical crustal motion are generallylow, with the majority of sites showing motion statistically insignificant from zero. A notableexception is the significant component of vertical land motion that contributes to the rapidrate of relative sea level change (>4mmyr-1) at the Hillarys site in the Perth area. Thiscorresponds to crustal subsidence that we estimate in our GPS analysis at a rate of -3.1 ±0.7mmyr-1, and appears linked to groundwater withdrawal. Uncertainties on the rates ofvertical displacement at GPS sites collected over a decade are similar to what we measure inseveral decades of tide gauge data. Our results motivate continued observations of relative sealevel using tide gauges, maintained with high-accuracy terrestrial and continuous co-locatedsatellite-based surveying.
AB - We quantify the rate of sea level rise around the Australian continent from an analysis of tidegauge and Global Positioning System (GPS) data sets. To estimate the underlying linear ratesof sea level change in the presence of significant interannual and decadal variability (treatedhere as noise), we adopt and extend a novel network adjustment approach. We simultaneouslyestimate time-correlated noise as well as linear model parameters and realistic uncertaintiesfrom sea level time series at individual gauges, as well as from time-series differences computedbetween pairs of gauges. The noise content at individual gauges is consistent with acombination of white and time-correlated noise. We find that the noise in time series from thewestern coast of Australia is best described by a first-order Gauss-Markov model, whereas eastcoast stations generally exhibit lower levels of time-correlated noise that is better describedby a power-law process. These findings suggest several decades of monthly tide gauge dataare needed to reduce rate uncertainties to <0.5mmyr-1 for undifferenced single site timeseries with typical noise characteristics. Our subsequent adjustment strategy exploits the moreprecise differential rates estimated from differenced time series from pairs of tide gaugesto estimate rates among the network of 43 tide gauges that passed a stability analysis. Weestimate relative sea level rates over three temporal windows (1900-2011, 1966-2011 and1993-2011), accounting for covariance between time series. The resultant adjustment reducesthe rate uncertainty across individual gauges, and partially mitigates the need for century-scaletime series at all sites in the network. Our adjustment reveals a spatially coherent pattern ofsea level rise around the coastline, with the highest rates in northern Australia. Over the timeperiods beginning in 1900, 1966 and 1993, we find weighted average rates of sea level riseof 1.4 ± 0.6, 1.7 ± 0.6 and 4.6 ± 0.8mmyr-1, respectively. While the temporal pattern ofthe rate estimates is consistent with acceleration in sea level rise, it may not be significant, asthe uncertainties for the shorter analysis periods may not capture the full range of temporalvariation. Analysis of the available continuous GPS records that have been collected within80 km of Australian tide gauges suggests that rates of vertical crustal motion are generallylow, with the majority of sites showing motion statistically insignificant from zero. A notableexception is the significant component of vertical land motion that contributes to the rapidrate of relative sea level change (>4mmyr-1) at the Hillarys site in the Perth area. Thiscorresponds to crustal subsidence that we estimate in our GPS analysis at a rate of -3.1 ±0.7mmyr-1, and appears linked to groundwater withdrawal. Uncertainties on the rates ofvertical displacement at GPS sites collected over a decade are similar to what we measure inseveral decades of tide gauge data. Our results motivate continued observations of relative sealevel using tide gauges, maintained with high-accuracy terrestrial and continuous co-locatedsatellite-based surveying.
KW - Australia
KW - Global change from geodesy
KW - Sea level change
KW - Space geodetic surveys
KW - Time-series analysis
UR - http://www.scopus.com/inward/record.url?scp=84880321005&partnerID=8YFLogxK
U2 - 10.1093/gji/ggt131
DO - 10.1093/gji/ggt131
M3 - Article
SN - 0956-540X
VL - 194
SP - 719
EP - 736
JO - Geophysical Journal International
JF - Geophysical Journal International
IS - 2
ER -