On the Greenland Ice Sheet, buried supraglacial lakes modulate surface mass balance through water storage and dynamic mass loss through drainage events. These buried lakes are typically mapped using C-band synthetic aperture radar (SAR) data, but to date, there has been no comprehensive validation of the radiometric signatures used for their detection. Here, we use ice-penetrating radar (IPR) to create a ground-truth catalog of buried lakes in spring 2017. We compare this catalog to coincident imagery from Sentinel-1 and investigate how lake geometry and environmental setting influence the radiometric signatures observed in the SAR imagery. We find that Sentinel-1 can directly sense buried water when the depth to the water table is less than 5 m thick. However, only 23-27% of buried lakes in our IPR catalog show clear negative anomalies in the HV polarization and those lakes are preferentially located in the lower accumulation zone. This suggests that current buried lake detection algorithms are spatially biased and significantly underestimate the total number of buried lakes. We show that incorporating multiple polarization metrics, particularly the difference between HH and HV backscatter, can reduce both spatial bias and missed detections. Our analysis provides a quantitative assessment of the strengths and limitations of C-band SAR for mapping buried lakes and offers a basis for improving detection algorithms and uncertainty quantification.