Performance evaluation of different satellite radar altimetry missions for monitoring inland water bodies

Abstract

Inland water bodies, e.g. lakes and rivers, play vital roles in society and in nature. Moreover, these water bodies can be considered as integrators of environmental change to study climate effects and hydrological cycle at global and regional scales. Because changes in the water level of lakes and rivers indicate changes in climatic parameters, such as precipitation and evaporation, it is necessary to monitor water level variation of inland water bodies continuously to understand long term changes. Traditional methods, e.g. using in-situ gauges, provide precise water level determination. But they can not monitor these water bodies in a way that today’s human needs are to be satisfied, because in-situ gauge networks do not cover all inland water bodies and their data are not publicly available. Furthermore, they are expensive to install and to maintain, especially in remote areas. In-situ gauge networks follow national policy and there is not a unified data base of their measurements. Satellite altimetry as a space-borne technology helps us to partially solve the issue of water level monitoring. This technique was originally designed to observe ocean water surface. But due to advances in satellite radar systems and in data processing methodologies, the application of satellite altimetry has been extended to monitor small lakes and narrow rivers over the past 20 years. So far, studying water level variations of inland water bodies has been a challenge for satellite altimeters in terms of spatial and temporal resolution as well as accuracy of water level determination. Due to a relatively large radar footprint, the illuminated area inside the footprint can be inhomogeneous, i.e. consisting of water, land and vegetation. Therefore, responses to the radar pulses from such a surface are complex and lead to multi-peak waveforms (corrupted waveforms). Seriously corrupted waveforms need to be analyzed to extract optimal ranges. Retracking is an effective method to improve the accuracy of the range measurement from contaminated waveforms and, consequently, to determine a more accurate water level. The design of an optimal retracking algorithm appropriate for a specific inland water body is very important in this respect. The quality of retracked water level depends on the type of altimeters and on the algorithm that is used in the retracking process. Moreover, the shape and size of the inland water bodies can affect the quality of the water level determination. In this thesis, we analyzed the waveforms in two different ways: full-waveform and sub-waveform retracking. For this purpose, different physical and empirical retracking algorithms have been employed to retrack the waveforms. In full-waveform retracking, for a given waveform one retracked range correction is estimated. But in sub-waveform retracking more than one retracked range correction can be calculated. We analyze all sub-waveforms in a given waveform and select the optimal one to retrack and consequently to determine water level variations. Three different analyses have been performed to select the optimal sub-waveform. In the first analysis we retracked only the first sub-waveform for all of the waveforms. In the second analysis all detected sub-waveforms in a given waveform are retracked to calculate the mean retracked range correction. In the last analysis we retrack the sub-waveform that provides the water level with minimum RMS with respect to model fits. For a given satellite, first we determine the water level according to on-board retrackers. The results of the on-board retrackers have been validated against available in-situ gauge data to find the best on-board retracker. Then, the full and sub-waveforms have been processed by different retracking algorithms to define the retracked water level. The retracked water level derived from different retracking scenarios have been compared with in-situ gauge data to evaluate the accuracy of each scenario. Finally, the results of the best on-board retracker were compared with the results from post-processing the waveforms to find the most accurate water level estimator. Radar characteristics and geometry of the satellite orbit, that affect on the altimeter’s performance, are designed based on main objectives of a given mission. Monitoring inland water bodies have not been the main objectives for the altimetry missions till now. We therefore do our analysis over data from different altimeters and evaluate their performance in water level monitoring of different inland water bodies. To complete our analysis, a comparison between different satellite altimeters has been performed to assess the performance of each altimeter in continental water level determination. We selected challenging objects with different shapes and sizes in different continents. For a given object, two or three satellite altimetry data sets have been analyzed to study water level variations. We used different satellite altimetry missions in our study, divided into pulse-limited and beam-limited altimeters. For the pulse-limited altimeters we selected Envisat, Jason-2, SARAL and CryoSat-2 LRM and for the beam-limited ones we used CryoSat-2 SAR and SARI n modes and I CES at satellite altimeters. GDR and SGDR data of these altimeters have been analyzed over four lakes: Neagh (Northern Ireland), Nasser (Egypt), Urmia (Iran) and Qinghai (China). We also analyzed the same data type of Envisat, Jason-2 and SARAL missions over different sections of the Danube river. We have found that over inland water bodies it is necessary to retrack the waveforms to achieve a qualified water level determination. Comparing the results from the on-board retrackers with those of the post-processed waveforms indicates that there tracked water level is more accurate. Our numerical results of the waveform retracking show that the sub-waveform outperforms the full-waveform especially over small lakes and complex shape (even large) lakes as well as over narrow rivers, e.g. Danube river. Over lakes Neagh and Nasser the beam-limited altimeters show better performance than the pulse-limited altimeters. In the case of Urmia lake, we analyzed only pulse-limited altimeters. Envisat provides the water level more accurately than CryoSat-2 LRM . Over Qinghai lake, covered by beam- and pulse-limited altimeters, both Envisat and CryoSat-2 LRM have the same performance. They show better performance than I CES at. Over Danube river, Envisat and SARAL show the same performance which is better than that of Jason-2. If we compare the results of all retracking scenarios for all missions, we can conclude that the mean sub-waveform retracked with the threshold retracker is the best retracking scenario to monitor small and complex shape inland water bodies. The first sub-waveform retracked with this retracker is an alternative scenario for the inland water bodies.