Normalized Cross-Gradient Screening and Consistency Assessment of ERT and GPR Data for Near-Surface Structural Mapping

Understanding what lies just beneath our feet is a fundamental challenge for anyone building, farming, or managing water in a specific location. In the world of geophysics, scientists use two primary tools to peer into the shallow ground without digging a single hole. The first tool, known as electrical resistivity tomography, works by sending an electrical current into the earth to measure how much the soil resists that flow. Wet clay conducts electricity easily, while dry sand or rock resists it, allowing researchers to map out layers of soil and rock based on their moisture and composition. The second tool, ground-penetrating radar, acts like a flashlight for the subsurface, sending high-frequency radio waves down and listening for the echoes that bounce back when they hit a change in material. While the first method is excellent at seeing deep changes in moisture and soil type, the second is incredibly sharp at spotting thin layers and small objects near the surface. The problem is that these two tools speak different languages; one measures resistance to electricity, and the other measures the reflection of radio waves. When scientists try to combine their pictures, they often end up with two maps that look similar but don't quite line up, leaving them to guess whether a feature seen in one is the same as the one seen in the other.

A team of researchers at Hasanuddin University in Indonesia set out to solve this puzzle by creating a new, objective way to check if these two different maps are actually showing the same underground structures. They conducted their survey on their own university campus, an area with complex soil conditions including weathered earth, fill material, and varying moisture levels. Instead of simply placing the two images side by side and hoping they matched, the team developed a rigorous process to test how well the boundaries between different soil layers aligned in both datasets. They started by carefully checking the raw data from their electrical surveys, discarding any readings that were clearly flawed or inconsistent, much like a photographer reviewing a roll of film to remove any blurry or overexposed shots. They selected the cleanest data sets from two specific survey lines, ensuring that the equipment and methods used were identical for both locations to make a fair comparison.

For the radar data, the researchers faced a similar challenge of clarity. They processed the raw radio wave signals to remove background noise and instrument drift, then converted the time it took for the signals to return into actual depth measurements. To do this, they had to assume a specific speed at which the radio waves traveled through the ground, a necessary step to turn the radar's time-based picture into a depth-based map. Once both the electrical and radar data were cleaned and converted, the team did something crucial: they forced both maps onto the exact same grid. This meant that every single point on the electrical map had a direct partner on the radar map at the same horizontal position and depth. Only then could they begin to compare the two.

The core of their new method involved looking at the edges of the features in both maps. In any geological map, a boundary between two different types of soil—say, a layer of wet clay sitting on top of dry sand—creates a sharp change or "gradient" in the data. The researchers asked a simple geometric question: do the edges in the electrical map point in the same direction as the edges in the radar map? If the two methods are seeing the same underground wall or layer, the changes in their data should line up perfectly. If they are seeing different things, the changes will point in different directions. By calculating a score for every single point on the grid, they could determine how much the two maps agreed on the structure of the ground. They focused their final analysis on the top 6.4 meters of the ground, which is the depth range where the radar data is most reliable, ensuring they were not trying to compare the radar's shallow view with the electrical method's deeper, less certain view.

The results showed that the two methods were largely in agreement, but with a slight difference between the two survey lines. For the first line of investigation, the consistency score was 0.573, while the second line scored slightly higher at 0.607. These numbers, which range from zero to one, indicate a moderate-to-good level of agreement, suggesting that in many places, both tools were indeed detecting the same structural boundaries in the soil. The researchers found that about 81 percent of the points they analyzed were valid for this comparison, meaning the data was strong enough to make a judgment. The second line showed slightly stronger consistency, implying that the underground layers there were perhaps more uniform or easier for both tools to detect. However, the team was careful to note that this was not a perfect match; there were still areas where the two methods disagreed, likely because one tool was seeing a small feature the other missed, or because the soil properties were too complex for a single explanation.

This study does not claim to have solved the mystery of the subsurface or to have created a perfect, unified image of the ground. Instead, it offers a transparent and reproducible way to test how well two different geophysical methods agree with each other before scientists draw conclusions. The researchers explicitly ruled out the idea that simply looking at the pictures side by side is enough, showing that without a strict, mathematical check on the alignment of features, interpretations can be misleading. They also emphasized that their results are specific to the shallow depths where the radar works well, and that the numbers they produced are a measure of structural alignment, not a direct translation of what the soil is made of. By separating the initial data cleaning from the final comparison, the team provided a clear path for others to follow, ensuring that future studies can distinguish between what the instruments actually measured and what the scientists interpreted. The work confirms that while electrical and radar methods see the world differently, they can be made to speak to one another, provided the comparison is done with care, precision, and a clear understanding of the limits of each tool.