Photometric reverberation mapping employs a wide bandpass to measure the AGN continuum variations and a suitable band, usually a narrow band (NB), to trace the echo of an emission line in the broad line region (BLR). The narrow band catches both the
emission line and the underlying continuum, and one needs to extract the pure emission line light curve. We performed a test on two local AGNs, PG0003+199 (=Mrk335) and Ark120, observing well-sampled broad- (B, V) and narrow-band light curves with the robotic 15cm telescope VYSOS-6 on Cerro Armazones, Chile. In PG0003+199, H_alpha dominates the flux in the NB by 85%, allowing us to measure the time lag of H_alpha against B without the need to correct for the continuum contribution. In Ark120, H_beta contributes only 50% to the flux in the NB. The cross correlation of the B and NB light curves shows two distinct peaks of similar strength, one at lag zero from the autocorrelated continuum and one from the emission line at tau_cent = 47.5 +/- 3.4 days. We constructed a synthetic H_beta light curve, by subtracting a scaled V light curve, which traces the continuum, from the NB light curve. The cross correlation of this synthetic H_beta light curve with the B light curve shows only one major peak at tau_cent = 48.0 +/- 3.3 days, while the peak from the autocorrelated continuum at lag zero is absent. We conclude that, as long as the emission line contributes at least 50% to the bandpass, the pure emission line light curve can be reconstructed from photometric monitoring data so that the time lag can be measured. For both objects the lags we find are consistent with spectroscopic reverberation results. While the dense sampling (median 2 days) enables us to determine tau_cent with small (10%) formal errors, we caution that gaps in the light curves may lead to much larger systematic uncertainties. (Abstract shortened, see the manuscript.)
