POSTER PAPER SESSION PP-40 The 10th FIG International Symposium on Deformation Measurements A TESTING PROCEDURE FOR SUBSIDENCE ANALYSIS

Land subsidence is an important topic in The Netherlands, due to its situation below sea level and the resulting large-scale waterworks this requires. Except for natural causes, like post-glacial rebound and compaction of the soft peat soil, man-made causes for land subsidence, like the extraction of ground water, oil, gas and salt, have to be monitored with great care. However, the determination of land subsidence with magnitudes of less than one centimeter per year requires high-accuracy measurement and modelling methods, and scrupulous data processing. Therefore, the monitoring of land subsidence is based on an extensive adjustment and testing procedure of the levelling data. The optimal deformation model, e.g. a kinematic model per benchmark or a spatial-temporal model for a subsidence bowl, is determined by testing the actual solution (null hypothesis) against several alternative hypotheses, resulting in adaptation of the data and/or the functional or stochastic model in case of rejection. The paper discusses this testing procedure and various alternative hypotheses like datasnooping, identification tests, tests for deviant behavior of individual benchmarks and epoch tests. The most likely alternative hypothesis indicates an adaptation of the model or data, resulting in an improved solution acting as new null hypothesis. The testing procedure is demonstrated on a multi-epoch data set