Presentation #100.74 in the session AGN.
Reverberation mapping is a powerful tool for mapping out the accretion disk on scales that are too small to be directly imaged. Recent reverberation mapping campaigns in the X-ray, UV, and optical, however, have found consistent discrepancies from standard disk reprocessing. In all cases, the measured time delays are longer than expected by at least a factor of 2-3, with especially long lags in the U-band. A leading theory is that the lags are contaminated by the broad line region (BLR), located beyond the disk and thus leading to lags longer than if produced by the disk alone. A frequency-resolved approach to computing the lags is thus essential for distinguishing the lags from the BLR versus the disk by isolating the variability produced on different timescales.
We present the frequency-resolved time lags of Mrk 335, which we compute by first modeling the light curves of a recent 100-day campaign using machine learning (Gaussian processes) to interpolate the data gaps. The lags have the largest discrepancies found to-date: the lags are too long by a factor of 7, and even longer in the U-band, with the X-rays surprisingly lagging the UV. By filtering out the longer timescales on which the BLR contamination would occur, we find the “too-long” lags are consistent with the expected disk reprocessing lags, with the U-band excess resolved as well. By modeling the frequency-resolved lags, we find that the long low-frequency lags require additional contribution from a distant reprocessor located at a radius consistent with the BLR based on previous measurements. These results support strong contamination of the lags by the BLR, and could explain the discrepancies seen by recent reverberation mapping campaigns.