Application of Robust Estimation Method in Study of Relationship Between Lake’s Water Area and Water Level

Abstract

Better understanding of the relationship between lake’s water area and water level plays an important role for the monitoring of floods and droughts. At present, there are many analysis and research on the relationship between water area and level, but the estimation criteria are mostly based on ordinary least square, which has no ability to resist the gross error. In this paper, the equivalent weight estimation method with robustness ability is introduced. We use the proposed method in study of the Poyang Lake which is the largest freshwater lake in China. The area extraction data and water level records spans the period from 2009 to 2013. Four models (linear, exponential, logarithmic and quadratic polynomial) of water area and water level are constructed by regression analysis. Meanwhile, two kinds of common weight function factors are applied to analyze the quadratic curve model. The experimental results show that the quadratic polynomial fitting performs best, and the solution of the equivalent weight function method is closer to the realism than least square. We note that the proposed robust estimation method can dynamically monitor the water area and water level, which provides a theoretical basis for similar research.