Massive MIMO systems are one of the key driving technologies for the next 5G communication systems. Using a large number of antennas at the base station to serve the end users may, in many cases, render the MIMO channels sparse due to limited scatterers at the Base Station (BS). In order to exploit this fact, compressed sensing channel estimation has been proposed in the literature. In this paper, we propose an adaptive Split Bregman massive MIMO channel estimation technique based on compressive sensing. The objective function of the proposed algorithm is a Signal to Noise Ratio (SNR) based weighted sum of the low rank property and the sparsity of the angular domain representation of the channels, whereas the regularization and tuning parameters of the Split Bregman algorithm are chosen adaptively to suit the channel estimation problem. Simulation results demonstrate the improved performance of the proposed algorithm.