Estimation of inter-satellite and inter-track biases of satellite altimetry missions over lakes and reservoirs using surface area from satellite imagery

Abstract

Including lakes, reservoirs, and rivers, inland water bodies cover only a small portion of the Earth’s surface. However, they play an important role in maintaining life on Earth, the global water cycle, and climate change. Due to the declining number of gauge stations that provide the in-situ data, the muti-mission satellite altimetry has been applied to the monitoring of medium to large lakes and reservoirs, which enables computing a water-level time series with impro ved temporal and spatial resolution. However, inter-satellite and inter-track biases are still a problem for multi-mission. There have been studies conducted to determine absolute altimetry biases at calibration locations and global altimetry biases. But we still don’t know everything about how satellites are biased over inland waters. This thesis is dedicated to developing a method to resolve the biases between satellites and tracks over lakes and reservoirs. Our strategy for calculating the biases between overlapping and non-overlapping time series of water levels from various missions and tracks is based on satellite-derived time series of water area. With the help of the estimated area by the image ry, the relative biases can be estimated by modeling the area-height relationship. With water level observations and water area observations, the Gauss-Helmert model is chosen to ad just the area-height relationship. Due to the possible interpolating error and the gross error in both observations, two robust estimation methods have been used to deal with outliers. One is the expectation maximization method, which provides a robust estimate by iteratively down weighting the observation with large residuals, and another one is an outlier rejection method based on Baardas’ data snooping, which detects outliers in the observations with statistical hypothesis tests. In the end, we also discuss the influence of the topography on the inter-track and inter-satellite biases. We calculate the standard deviation of the DEM of the intersaction area between the 2 km region along the track and the 5 km region along the lake to determine the relationship between topography and biases. The results show a high correlation between the inter-track biases and the topography. We have employed the developed methodology on a number of lakes and reservoirs, and the findings are compared to in-situ water level data. The results reveal the existence of the inter-satellite and inter-track biases, which vary from the global bias estimates.