In this study, we discuss possible origins of the D reflector beneath the North Atlantic region based on a combined analysis of P and S wave data. We use over 700 USArray station recordings of the M w 6.3 earthquake that occurred in April 2010 in Spain. In order to investigate the D layer we look for waves reflected off the top of it, namely PdP and SdS waves, and compare them to the core-mantle boundary (CMB) reflections used as reference phases. The differences in traveltimes and amplitudes are sensitive to D properties. Because the USArray installation generates a dense array, we are able to provide an almost continuous map of the detection or absence of PdP and SdS waves in the North Atlantic region. We use a Bayesian inversion for traveltimes, together with synthetic seismogram calculations, to find the best-fitting D properties, (V p , V s) jumps across the D interface and D thickness. We find that the best-fitting models are for a D layer of about 300 km thick, with or without a velocity gradient of about 30 km at the top of it. Regardless of the model type, positive and similar velocity increases in both P and S velocities at the D interface, ranging from 2.7 to 3.8 per cent, are required to fit the data well. Our data rule out velocity decreases in P and S waves at the D interface as well as no velocity reduction above the CMB. There are also regions where we do not observe PdP and SdS waves. Collectively, these observations suggest lateral variations in both chemistry and temperature, combined with phase transitions. For instance, ancient oceanic basalt debris from the Farallon slab could be modulating the detection of the D reflector in this region.