%0 Journal Article %1 Berendrecht2006959 %A Berendrecht, W.L. %A Heemink, A.W. %A van Geer, F.C. %A Gehrels, J.C. %D 2006 %J Advances in Water Resources %K myown %N 7 %P 959 - 973 %R http://dx.doi.org/10.1016/j.advwatres.2005.08.009 %T A non-linear state space approach to model groundwater fluctuations %U http://www.sciencedirect.com/science/article/pii/S0309170805002113 %V 29 %X A non-linear state space model is developed for describing groundwater fluctuations. Non-linearity is introduced by modeling the (unobserved) degree of water saturation of the root zone. The non-linear relations are based on physical concepts describing the dependence of both the actual evapotranspiration and the percolation rate on the degree of saturation of the root zone. Precipitation and reference evaporation are the input variables. Recharge from the root zone is transferred through the percolation zone to the water table by means of a linear reservoir model. The flux in the percolation zone is assumed to be independent of the groundwater level. Errors due to model assumptions and parameter uncertainties are modeled as a noise process. The parameters of the resulting stochastic model are calibrated on time series by combining an extended Kalman filter with a maximum-likelihood criterion. The model is tested at two locations and compared with a linear time series model. It is shown that the non-linear model estimated extreme groundwater levels better than the linear time series model. The variance of the stochastic component reduced significantly. This implies that effects from natural influences can be separated from non-natural influences with less uncertainty. Another advantages of the non-linear model is that it also gives an indication of fluctuations in the degree of saturation in the root zone.

@article{Berendrecht2006959, abstract = {A non-linear state space model is developed for describing groundwater fluctuations. Non-linearity is introduced by modeling the (unobserved) degree of water saturation of the root zone. The non-linear relations are based on physical concepts describing the dependence of both the actual evapotranspiration and the percolation rate on the degree of saturation of the root zone. Precipitation and reference evaporation are the input variables. Recharge from the root zone is transferred through the percolation zone to the water table by means of a linear reservoir model. The flux in the percolation zone is assumed to be independent of the groundwater level. Errors due to model assumptions and parameter uncertainties are modeled as a noise process. The parameters of the resulting stochastic model are calibrated on time series by combining an extended Kalman filter with a maximum-likelihood criterion. The model is tested at two locations and compared with a linear time series model. It is shown that the non-linear model estimated extreme groundwater levels better than the linear time series model. The variance of the stochastic component reduced significantly. This implies that effects from natural influences can be separated from non-natural influences with less uncertainty. Another advantages of the non-linear model is that it also gives an indication of fluctuations in the degree of saturation in the root zone. }, added-at = {2015-10-02T12:57:09.000+0200}, author = {Berendrecht, W.L. and Heemink, A.W. and van Geer, F.C. and Gehrels, J.C.}, biburl = {https://www.bibsonomy.org/bibtex/2716b78623249d4921fee895d123e144a/wlberendrecht}, doi = {http://dx.doi.org/10.1016/j.advwatres.2005.08.009}, interhash = {82a6dc537da76ffa37160b903fb4698c}, intrahash = {716b78623249d4921fee895d123e144a}, issn = {0309-1708}, journal = {Advances in Water Resources }, keywords = {myown}, number = 7, pages = {959 - 973}, timestamp = {2015-10-02T12:57:09.000+0200}, title = {A non-linear state space approach to model groundwater fluctuations }, url = {http://www.sciencedirect.com/science/article/pii/S0309170805002113}, volume = 29, year = 2006 }