Preliminary results of the study of using ground-penetrating radar (GPR) as a tool to locate artificial burrows similar to souslik burrows and future directions of mapping burrow systems of sousliks or other burrowing mammals alike

Endangered burrowing mammals are good indicators of ecosystem quality as they frequently play a crucial role in the functioning of grassland ecosystems, maintaining their diversity, functions, or services. However, the non-destructive estimation of their population size, spatial and temporal population dynamics remains a challenge. The number of burrow openings is a good proxy for estimating actual population sizes if one individual occupies one burrow system and the ratio of openings per burrow system is known. Remote, semi-automated counting of animals’ surface burrows has been successful, and we now focus on detecting subsurface animal burrows. For this purpose, we investigate the applicability of GPR surveys to non-destructively identify and locate artificial burrows of the same size dimensions as burrows of protected ground squirrels.  Based on the results we present an approach to non-invasively map ground squirrel burrows.

A Mala system with 160 and 750MHz antennas was used for the GPR surveys. Artificial burrows (ABs) (5-7cm wide, 1m long) were drilled in the wall of a ditch (depth of 2m, length of 20m). Each burrow location was known and placed between 5 and 160cm depth perpendicular to the direction of the GPR survey. Burrow locations were marked both in the field and radargram. The survey area was a grassland (similar to natural ground squirrel habitats) with short vegetation and even ground surface.

A standard processing of the raw GPR data was used in Reflexw2D, including: compressing original data (deleting every 2^nd trace), bandpass filtering, time-zero correction using the automatic correct max phase option, and move-starttime. Processed radargrams were also (fk) migrated and gain adjusted for better display of burrows on images. The last step was the time-depth conversion with constant velocity of 0.1 m/ns. The processing sequence was saved and applied to each raw data file with the same data acquisition parameters.

Preliminary results indicate that although many of ABs can be found through the use of GPR, this method has some drawbacks. Penetration depth was limited to less than 150cms. Since, sousliks dig deeper in the soil, that depth could be one of the limiting factors in mapping entire burrow systems. A general difficulty of locating ABs was that ABs’ reflections were often indistinguishable from unknown subsurface objects despite the prior knowledge of their exact location in the soil. Although reverse polarity of the reflected wave was expected due to the air-filled burrows in the soil, the data did not show this phenomenon clearly. ABs in the upper ~30cm, opposite to ABs deeper, were identifiable more with less plotscale colour intensity.

In summary, while some ABs were detected by GPR, many were not, even though their exact location was known. This experience has indicated a different approach for mapping animal burrows may be necessary. Multiple-point geostatisitcs (MPS) could be a good approach for modelling non-linear burrows. Information about burrows can be obtained from burrow maps used as training images could be combined with GPR data to enable modelling of multiple-point relations and complex zig-zag patterns.