Influence of pumping head data in the conditional estimation of a transmissivity field

A contribution to inverse modelling within a stochastic framework is presented. Insights are provided into the type and level of information associated with head data resulting from permanent pumping tests in a two-dimensional aquifer. A conditioning methodology based on Monte-Carlo conditional simulations is developed and applied to numerical test cases taking different data density subsets into account. Results show that satisfactory reconstruction of the original transmissivity field requires a rather high data density (regular grids of mesh size two times the correlation length of transmissivity). These results are in accordance with results formerly obtained by conditioning on head data measured within a uniform flow field. Differences between these two types of data are analysed. But a major advantage of pumping tests is that additional information is easily obtained by changing the pumping location within a given set of boreholes. This was done here and preliminary results show that important additional variance reductions are obtained.