The reasons for this are as follows:
1. With the twin configuration, the measuring conditions change from measuring point to measuring point only at the moving electrodes. As a result, the measurement is significantly less noisy and of higher quality.

2. The extremely short distance between the two current electrodes in the square configuration compared to the twin configuration leads to a shallower volume detection. As a result, elevations and depressions on the earth's surface have a greater effect and, for example, disturbing line formations are more likely to occur.

3. Although the square arrangement saves carrying the connecting line to the distant pair of electrodes, this argument is only of essential importance if the measuring surface is provided with many obstacles.

Monument type
Villa Rustica, farm with outbuildings and extensive pipe system
 

Survey data
Area investigated: 36.200 m2
Resolutions:    1 MP/m2 (a = 1,0 m): 36.200 m2
                       4 MP/m2 (a = 0,85 m):  13.900 m2

Array configuration: square-array
Measuring instruments: Lippmann 4point light MC
Imaging software: SURFER (Golden Software)
Survey period: 08.2014 -04.2015
(additions: Sept. & Nov. 2015)

Comments
A few, only faintly recognizable, negative cropmarks visible on Google Earth, as well as some thin sherd-scatterings found during the first inspection, led to the discovery of the site in 2014.

In the course of the following extensive prospecting campaign, the  villa area could presumably be recorded completely, but the prospecting results do not allow a reliable addressing of the four recognized areas (A, B, C, D) due to the wrongly chosen measurement configuration.
The reason for using a square arrangement was, on the one hand, to avoid carrying a connecting line, as is usual with the previously used twin configuration, because of the unusually large measuring area, and, on the other hand, the desire to exploit the potential of this previously untested array-configuration in the get to know practice. In a first phase (top picture) the entire accessible area was prospected with a square array with an electrode separation of one meter (a = 1.0 m) and a corresponding grid of one meter.

The strip-shape of the survey area  which was divided into nine plots of land with different types of agricultural use and processing led to a very inhomogeneous result and was therefore difficult to interpret.

In the hope of being able to make the structures more visible through a higher resolution, all areas that showed a recognizable or possible foundation structure (red, green, blue) were prospected in a second measurement phase . The square array used here had a slightly smaller electrode-separation (a = 0.85 m) than that used in the first phase, and the areas were prospected with a grid of 0.5 m. This resulted in an overlapping measurement with a resolution of 4MP/m2.

The visibility of the structures improved only minimally in the second phase and the disturbing influence of the agricultural processing lines increased significantly.

Subsequent line removal (e.g. GIMP filter, enhance, destripe) made the result a little more pleasant for the untrained eye, but did not bring any additional knowledge.

To assess the potential of a possible repeat prospecting with the previously used and proven twin-probe configuration, the area C was surveyed with an electrode separation of 0.75 m. (Phase 3)

Although this measurement - compared to the unsuccessful measurements of the 2nd phase - was carried out with the slight advantage of a 10 cm or 12% smaller electrode separation, it delivered a very much higher gain in image quality and recognition than the square configuration of phase 2 and made a significant difference addressing possible structures:

Two structural complexes are recognizable. The complex on the left has a portico structure at the top of the image, which possibly continues downwards in the center of the grid between the two parts of the building. While the complex on the left has a small-scale subdivision with individual floors, the right part of the building shows large-scale preserved floors or hypocaust structures. Along the right edge of the grid there are five water conduits that can also be seen in the first phase of the prospection (Phase 1.)

The repeat survey planned in the near future should be carried out with a twin-probe array (a = 0.5m) and a corresponding grid of
0.5 m x 0.5 m.

The assumption that an increase in resolution solely by overlapping the measurement points could improve the recognizability of the structures is incorrect. A higher resolution can only lead to an improvement in the recognisability of the structures by reducing the electrode-separation and thus the measurement volume, because the structure is broken down into smaller parts. Merely increasing the number of measuring points does not provide any additional information because: Many equally weak images side by side do not result in a stronger image.
A "limit value problem" arises when the measurement resolution is successively increased due to the resulting smaller acquisition volume and the associated smaller acquisition depth. The measurement result becomes more and more detailed with ever shorter electrode-separation, but if the structure sought is below the detection depth, it is no longer detected....

(see also Marnheim)