Data and models
The DGM-deep v4.0 model (2014) was used as basis for ThermoGIS v2.0. This model consists of depth and thickness grids at a (Main) Group level. The layers of DGM-deep are based on extensive seismic interpretation, and are calibrated with about 800 wells. A detailed description of the creation of the model can be found on the NLOG website. For ThermoGIS v2.0, the areal extent of the Rotliegendes, the Zechstein and the lower Cretaceous were updated with respect to DGM-deep v4.0 using the interpretation of a confidential well near Middenmeer, which was released by the operator for this purpose.
The depth and thickness of the aquifers that exist within the (Main) Groups were modelled for ThermoGIS v2.0 using information of 3855 wells, which is significantly more than the 800 wells that were used for the calibration of DGM-deep v4.0. Therefore, discrepancies may exist between the two models.
The Nieuwerkerk Formation (SLDN), containing the Delft Sandstone and Alblasserdam members, was mapped using unpublished interpretations. The thickness of these units may therefore deviate from the thicknesses published on the NLOG website.
Out of the selection of 3855 wells, several have been excluded for geological and/or technical reasons:
- Geological: unreliable interpretation, well intersected by faults, well with thrust fault exceeds the DGM-deep main units
- Technical: poor quality of unlogged well, unreliable deviation survey, etc.
The wells of the resulting data set were split in four types:
- Type = 1: aquifer fully drilled - the entire thickness is used for modelling
- Type = 2: well ends (TDs) in the modelled aquifer - the modelled thickness of the aquifer at this location must be equal to or exceed the thickness encountered in the well
- Type = 3: the modelled aquifer was not encountered in the well, but an older stratigraphic unit was - a zero thickness is modelled for this location
- Type = 4: the modelled aquifer was not encountered in the well, but the (Main) Group containing the aquifer was; in combination with regional knowledge it is assumed that the aquifer is present. This is usually the case when the stratigraphic labelling of the well is incomplete - this well is not used in the modelling
The thickness of the aquifers is calculated from the well data using the kriging algorithm. A by-product is the kriging variance. The square root of the variance is the standard deviation, which is used for calculating the P10 and P90 thickness maps. Note: the kriging variance represents the uncertainty of the modelling, but tells nothing about the uncertainty of the determined thickness in the well.
Standard workflow: consistent
The areal extent of the aquifers is constructed using the extent of the relevant (Main) Groups of the DGM-deep v4.0 model and the relevant wells from the set of 3855 wells. For each well, the difference between the base of the aquifer and the base of the (Main) Group is calculated. The difference is interpolated and added to the base of the (Main) Group. The resulting grid represents the base of the aquifer. The thickness of the aquifer is then added to the base, resulting in the depth of the top of the aquifer. Figure 1 shows that sometimes the modelled aquifer does not fit within the alotted space in the (Main) Group. This is due to the fact that for the aquifers, more wells are used than for the (Main) Groups, which are based on a combination of seismic data and wells. The top and base of an aquifer cannot cross the top and base of the (Main) Group. Therefore the top and base of both were compared and, when required, made "consistent" with each other. The DGM-deep v4.0 layers took precedence, therefore the aquifer grids were modified by cutting the top. The areal extent was then determined using a thickness cutoff of >0 meters. This method may lead to an underestimation of the aquifer thickness and thus its geothermal potential.
Salt domes of the Zechstein Group in the northern part of the Netherlands may disturb the overlying Paleogene aquifers, resulting in their being absent.
Alternative workflow: inconsistent
Not all grids were made consistent. For some aquifers this would lead to an incorrect loss of data due to the cutting of the top. Figure 1 illustrates the effect. Here, the Rijnland Group is defined by a convex base and a flat top, adapted from DGM-Deep v4.0. In the example, the Group contains four aquifers (aq1-4). A well that was not used in DGM-Deep v4.0 may reveal that the actual top of Aquifer 4 lies above the DGM-Deep v4.0 In the consistent case, the DGM-Deep v4.0 surface will prevail and the top of the Aquifer 4 will (erroneously) be cut (1). In the inconsistent case the information from the extra well will prevail, and the top of the modelled Aquifer 4 will lie above the DGM-Deep v4.0 Base Chalk surface (2). This method of inconsistent gridding may lead to an overestimation of the geothermal potential.
The following aquifers are inconsistent with DGM-deep v4.0:
- Middle and Lower North Sea Groups (Someren, Voort, Steensel, Basal Dongen Sand, Reusel and Heers Members): due to the lack of sufficient data points and/or incomplete interpretations, while the aquifer is supposed to be present.
- Nieuwerkerk Formation: due to the use of a separate data set that is possibly inconsistent with the data published on NLOG.
- Rotliegend (Slochteren Formation and Upper Slochteren Member): the thickness of the aquifer was subtracted from the depth of the base of the Zechstein in order to prevent local underestimation.
In order to keep the thickness within reasonable limits (regarding border effects) during modelling, some aquifers require a minimum or maximum thickness to be set (for inconsistent grids). Additional data points (pseudo-wells) were sometimes created to guide the interpolation of aquifers having a very limited amount of data points (for instance, the Tubbergen Formation).