Estimation of time delays from unresolved photometry

Context. Longtime monitoring of gravitational lens systems is often done using telescopes and recording equipment with modest resolution. Still, it would be interesting to get as much information as possible from the measured lightcurves. From high resolution images we know that the recorded quasar images are often blends and that the corresponding time series are not pure shifted replicas of the source variability. Aims. In this paper we will develop an algorithm to unscramble this kind of blended data. Methods. The proposed method is based on a simple idea. We use one of the photometric curves, which is supposedly a simple shifted replica of the source curve, to build different artificial combined curves. Then we compare these artificial curves with the blended curves. Proper solutions for a full set of time delays are then obtained by varying free input parameters and estimating statistical distances between the artificial and blended curves. Results. We performed a check of feasibility and applicability of the new algorithm. For numerically generated data sets the time delay systems were recovered for a wide range of setups. Application of the new algorithm to the classical double quasar QSO 0957+561 A, B lightcurves shows a clear splitting of one of the images. This is an unexpected result and extremely interesting, especially in the context of the recent controversy about the exact time delay value for the system. Conclusions. The proposed method allows us to properly analyse the data from low resolution observations that have long time coverages. There are a number of gravitational lens monitoring programmes that can make use of the new algorithm.