RT doctoral thesis
T1 The gulf of cadiz as a natural laboratory for the validation and exploitation of coastal altimetry and model data
A1 Aldarias Martos, Ana Isabel
A2 Física Aplicada
K1 coastal altimetry
K1 synthetic aperture radar (SAR) data
K1 high posting rate
K1 Sentinel-3
K1 sea level anomaly
K1 Fully Focused SAR altimetry
K1 Gulf of Cadiz
K1 Delft3D mode
K1 calibration
K1 validation
K1 Guadalquivir River estuary
K1 fluvial discharge
K1 water levels
K1 altimetrymodel comparison
AB Hydrodynamic models and satellite altimetry can be complementary tools for the study ofsea level variations. Nowadays, users of these tools demand high quality products in coastalzones. In this sense, this doctoral dissertation focusses on the validation of innovativeproducts that entail an advance in the understanding of sea level variation in coastal areas.The study was carried out in the Gulf of Cadiz (GoC), an important region that connectsthe Atlantic Ocean and the Mediterranean Sea, although other study areas were also used tostrengthen the analysis. The research was performed using: in-situ data, sea level altimetrymeasurements from Sentinel-3 SRAL, and observations from a hydrodynamic modelimplemented in the main study area. The in-situ data were used to validate the altimetrymeasurements, as well as to calibrate and validate the numerical model.Different specific objectives were proposed. The first was to determine the quality ofaltimetric data in coastal areas from the new Sentinel-3 space mission. Altimetry data ofSentinel-3A SRAL (S3A) were validated at the sampling frequency of 80 Hz. The data wereobtained from the European Space Agency (ESA) Grid Processing On Demand (GPOD)service over three coastal sites in Spain: Huelva (GoC), Barcelona (Western MediterraneanSea), and Bilbao (Bay of Biscay). Two tracks were selected at each site: one ascending andone descending. Data were validated using in-situ tide gauge (TG) data provided by theSpanish Puertos del Estado. The altimetry Sea Level Anomaly (SLA) time series wereobtained using the corrections available in GPOD. The validation was performed usingtwo statistical parameters, the Pearson correlation coefficient (r) and the root mean squareerror (rmse). In the 5–20 km segment with respect to the coastline, the results obtainedwere 6–8 cm (rmse) and 0.7–0.8 (r) for all of the tracks. The 0–5 km segment was alsoanalysed in detail to study the effect of land on the quality of altimetry data. Results showedthat the track orientation, the angle of intersection with the coast, and the land topographyconcur to determine the nearest distance to the coast at which the data retain a similar levelof accuracy than in the 5–20 km segment. This ‗distance of good quality‘ to shore reaches aminimum of 3 km for the tracks at Huelva and the descending track at Barcelona.In addition, altimetry sea level data of Sentinel-3A and Sentinel-3B SRAL (S3A and S3B)were also validated at the sampling frequency of 80 Hz. Two tracks of S3A and two of S3Bwere selected at seven different coasts around the Iberian Peninsula. The altimetry SLAtime series obtained were compared with TG in-situ data using the standard deviation of the difference (sdd) and the normalized sdd (sdd_n). Two tidal models were used: TPXO8and TPXO9. They were previously validated with in-situ data and then used in the S3assessment. Contrary to expectations, a more current version of the tide model did notalways lead to improvements for all of the coasts studied. The same data availability andaccuracy results (mean sdd <9cm) were obtained for both satellites, as the radar altimeteron-board S3A and S3B are identical instruments. The sdd_n results were generally rangedbetween 0.1 and 0.25 cm, higher values were obtained in coastal areas with a complexhydrodynamic.The second specific objective was to implement the Delft3D model in the estuary of theGuadalquivir River and part of the GoC continental shelf with the aim of studying theinfluence of its discharges on the sea level variability. Details of the Delft3D FLOWmodule implementation are given in the manuscript. The model was calibrated andvalidated along the river estuary comparing the output with in-situ observations of waterlevel and current velocities during normal and high-discharge events. In order to obtainthe best possible adjustment, the friction coefficient and bathymetry were used asadjustment parameters. The determination coefficients attained mean values of R2= 0.9and R2=0.8, for calibration and validation, respectively. Moreover, the model wascalibrated for the continental shelf during normal discharge conditions using data fromthree current meters, obtaining good correlation results (rmse= 3.0 cm·s-1and R2=0.7, onaverage). The model simulations were able to reproduce the low salinity plume-inducedover-elevations at the mouth of the estuary and its adjacent inner shelf during periods ofhigh river discharge from the head dam (> 400 m3·s-1). These over-elevations were alsoidentified in a qualitative comparison with altimetry data. Despite the good resultsobtained, there are some improvements that could be made for future work: includingwind, coupling the wave module, updating the bathymetry, etc.The aim of this thesis last section was to apply the new Fully Focused SAR (FF SAR)processing technique for the Sentinel-3 altimetry signal. The accuracy and precision of thisnovel product were analysed in order to provide the best quality product close to the coast(0-5 km track segment). FF SAR processing is similar to SAR altimetry, but with anunprecedented high along-track resolution which goes up to the theoretical limit equal tohalf the antenna length (~0.5 m). Two FF SAR algorithms still in development were usedin this work: FF SAR Back Projection (BP) (S3 prototype version of Kleinherenbrink et al., 2020), and FF SAR Omega-Kappa (WK) (Guccione et al., 2018), as well as differentretracking algorithms. Two tracks from Sentinel-3A and two from Sentinel-3B wereprocessed, at 80 Hz. The products were validated by comparing time series of SLA withthose obtained from a tide gauge in the Gulf of Cadiz. The accuracy was analysed using thePercentage of Cycles for High Correlation (PCHC) and the standard deviation of thedifference (sdd); and the precision was determined by calculating the along-track noise. FFSAR and unfocused SAR products were compared. The results showed improvementsusing Adaptative Leading Edge Subwaveform (ALES+) retracker for unfocused SAR,although FF SAR BP with ALES+ was the most precise product for all the tracks. Inaddition, highly accurate SLA measurements were also obtained with FF SAR products.The greatest advantage of FF SAR is that it produces good quality data closer to the coast(1-2 km) than unfocused SAR (3-4 km).Finally, these results highlight the potential of the implementation of validated altimetrydata and hydrodynamic models in sea level studies. Furthermore, the methodologydescribed here can be useful to validate altimetry data, as well as to implement the Delft3Dmodel in other coastal areas.
YR 2022
FD 2022
LK http://hdl.handle.net/10498/28930
UL http://hdl.handle.net/10498/28930
LA eng
DS Repositorio Institucional de la Universidad de Cádiz
RD 21-sep-2026