为了验证Cryosat-2/SIRAL数据监测湖泊水位的能力，提高其提取湖泊水位变化的精度，以青海湖为研究对象，利用主波峰重心偏移法、主波峰阈值法、主波峰5-β参数法、传统重心偏移法、传统阈值法和传统5-β参数法6种算法对Cryosat-2/SIRAL LRM 1级数据进行波形重跟踪，提取青海湖2010—2015年湖泊水位，对比不同算法获取水位的精度，并结合Envisat/RA-2 GDR数据，延长水位变化时间序列，获得青海湖2002年—2015年的水位变化信息。结果表明，主波峰5-β参数法提取湖泊水位的精度最好，均方根误差为0.093 m；对于GDR产品中LRM模式的3种数据，基于Refined OCOG算法的数据更适合湖泊水位的提取；青海湖2002年—2015年水位整体上涨，水位平均变化趋势为0.112 m/年，年内水位变化呈现明显的季节性。

Lake is an important water resource and a sensitive indicator of climate and environment change. Satellite altimetry has been used as an alternative tool to measure lake levels since the 1990s. With the development of satellite altimetry technology, different altimetry thatcan be used for lake level monitoring has been launched. This paper aims to verify Cryosat-2/SIRAL data capabilities of monitoring lake level, improve the extraction accuracy of lake level changes, and monitor the water level change of Qinghai Lake. The boundary of the lake was first extracted using the image of MODIS13Q1 close to the date altimeter visited to ensure the observation points in the lake. This study used six kinds of algorithms to retrack Cryosat-2/SIRAL LRM level 1 data in order to extract the Qinghai Lake water levels from 2010 to 2015, including the primary peak Offset Center of Gravity (OCOG), primary peak threshold, primary peak 5-β parameter, traditional OCOG, traditional threshold, and traditional 5-β parameter methods. Furthermore, the Cryosat-2/SIRAL GDRs of LRM mode provides three different retrackers: UCL, refined CFI, and refined OCOG. The accuracy of all these different algorithms in extracting water level was then compared with the measured water level of the hydrological station using the indexes, such as the difference, correlation coefficient, and root mean square error (RMSE). The 2002 to 2015 water level time series of Qinghai Lake was obtained and combined with the Envisat/RA-2 GDR data by adding the differences between the lake levels extracted from Envisat/RA-2 and Cryosat-2/SIRAL. The seasonal and inter-annual variation features of Qinghai Lake water level were then analyzed. The results showed that the primary peak 5-β parameter retracker for Qinghai Lake performed the best with the least RMSE 0.093 m and a maximum correlation coefficient (0.956) among these retrackers. Generally, the water level extraction accuracy of the retrackers based on the primary peak is better than the retrackers based on the entire waveform. While for these waveforms which are influenced by land echo information, the primary peak OCOG algorithm and primary peak threshold algorithm presented were better than others. Comparing the three kinds of Cryosat-2/SIRAL GDR products for LRM patterns, the data based on the refined OCOG algorithm was more suitable for extraction of lake level. In addition, the water level of Qinghai Lake generally rose from 2002 to 2015 with the overall increasing trend of 0.112 m/a, with marked seasonal changes in a year. The water level began to rise in May and December each year, with respectively high peaks in September and January. Based on the preceding experiments and analysis, the Cryosat-2/SIRAL LRM data can be used to extract lake levels with high precision at approximately 1 dm. Retracking for altimetry level 1b data could improve the water level extraction accuracy. The best adaptive retracking algorithm for different types of lakes is often different because they show different echo waveforms. The analysis in the paper is rough; hence, the next step is selecting different types of lakes to obtain a detailed comparative analysis.