DeeSse is an advanced multiple-point statistics (MPS) code allowing to simulate spatial fields or time series based on training data sets.
DeeSse is extremely flexible and has been used for a broad range of applications such as geological modeling, groundwater modeling, time series reconstruction, geophysical inversion, remote sensing and climate applications.
One of the reason of this popularity is that DeeSse can handle a wide variety of non stationarity in the training data sets and in the simulation grids such as trends, local or global proportions.
The core technique is to sample directly and randomly patterns from a training data set (Mariethoz et al. 2010). This has many advantages as compared to other MPS methods. First, it allows to treat in the same manner categorical, continuous, or multiple variables. Second it simplifies the algorithm and offers a very large flexibility.
Based on the original direct sampling technique, DeeSse is implemented as a parallel and efficient C code maintained by the University of Neuchâtel and commercialized by Ephesia Consult as a Petrel Plugin. The Direct Sampling algorithm is patented by the University of Neuchâtel.
Continous research and community of users
DeeSse is currently used by a few large companies and more than 30 universities worldwide.
Over the years, DeeSse has been improved and extended with features that most other MPS codes do not offer. For example, when data are available at a scale larger than required for a given application DeeSse can account for these data through block conditioning. This is especially useful for example for downscaling climate data such as rainfall or geophysical inverse modeling. Note that innovation on DeeSse is still ongoing, new features are regularly implemented to answer new needs and offer new capabilities to our users.