GPS Reflectometry: Innovative Flood Monitoring at the Mekong Delta
- Location of the GPS-R experimental set-up in Vietnam
Jamila Beckheinrich (Successful
finalization June 2016)
Faculty VI - Planning Building Environment, Technical University Berlin
Ground-based gauge instrumentation enables a high altimetric accuracy with high temporal resolution, but for a point location only. However, the number of gauge instruments is decreasing worldwide due to high maintenance costs. Global Positioning System-Reflectometry (GPS-R) reveals new perspectives for water level monitoring, since water surfaces show a high reflectivity for the GPS L-band signal. To test the possibility of using this innovative technique as a gauge instrument, two field campaigns were conducted in Vietnam, in February 2012 and March 2013 respectively, within the Water related Information system for the Sustainable Development Of the Mekong delta (WISDOM) project. As the use of phase observations has the potential to offer more accurate results, a new generation of GNSS Occultation, Reflectometry and Scatterometry (GORS) receiver has been successfully tested. Several reflection traces on the 150 m wide Can Tho River section have been recorded.
GPS-R phase-based altimetry implies continuous coherent phase
observations. Due to the high sampling rate of the recorded data (200
Hz), a new automated algorithm, based on an ellipse fitting, is
proposed to extract coherent phase observations. A hit rate of 82%
could be reached. To test the geometrical impact of the antenna
position on quality and quantity of the recorded coherent
observations, two different antenna heights were used. For the first
time, a detailed analysis of the recorded observations was performed
and correction techniques were developed and applied. The results of
the analysis show that the roughness of the water surface, caused
primarily by ship traffic, had a major influence on loss of coherency.
Additionally, the surroundings of the antennas and the river geometry
restrict the use of reflection events within the interval of 3 to 29
deg. The analysis also showed the presence of multipath effects other
than the water surface in the direct and the reflected signals, thus
deteriorating the results. A phenomena that was already found in other
research activities but not resolved. These multipath effects are
mitigated and filtered based on an adjusted Empirical Mode
Decomposition Method showing an improvement of several centimetres in
the obtained water level results. The data also reveals the strong
presence of cycle slips that distort the results, so that a
preprocessing of the data is mandatory. A cycle slip detection
strategy was therefore proposed. The challenge here, was to develop an
algorithm based on GPS L1 observations only as they form 95% of the
coherent phase observations. The detected cycle slips reveal a strong
correlation with the surroundings of the antennas.
To extract water level changes, a Least-Squares method is used. As the accuracy of the extracted altimetric heights are strongly dependent on the accurate calculation of the geometrical excess paths between the direct and the reflected signal, effort was made to analyse the impact of systematic errors that influence the signals. For this purpose, different tropospheric correction strategies are compared, showing that atmospheric height layer and total zenith delay have to be precisely calculated. Additionally, the impact of the mostly ignored phase wind-up effects in GPS-R applications is underlined. As the recorded coherent phase observations are ambiguous, an ambiguity fixing strategy for different satellite redundancy is proposed. Mostly, only L1 coherent phase observations from a single satellite are present. The most challenging problem in this case arises due to the lack of redundancy and slowly changing satellite geometry. In this case, water level heights within the set goal of decimetre level of accuracy can be calculated with an interval of 10 min. In the most desirable case, with the presence of more than one satellite tracked simultaneously, water level heights could be estimated at the centimetre level of accuracy.
To test the general applicability of the proposed algorithm, recorded observations during a measurement campaign, conducted 2014 within the Programme Marocco-Allemand de Recherche Scientifique (PMARS) project in Midelt, Morocco, were used. The obtained results within the decimetre level of accuracy, underline the applicability of the proposed algorithm to other ground and phase-based GPS-R altimetry applications.
All the obtained results are verified by comparing them with tide gauge measurements in the vicinity of the antennas.