The need to deal with non-homogeneous clutter has driven much of the recent research in space-time adaptive processing (STAP). An extension of the low-complexity, sigma-delta (Sigma Delta) algorithm incorporating the direct data domain (D(3)) processing is presented. The new algorithm is practical and improves target detection in non-homogeneous clutter environments. The algorithm employs a hybrid approach, combining D(3) processing with the more traditional statistical approach, thereby obtaining advantages of both. First, a modified D(3) algorithm, which maximises signal-to-interference-plus-noise ratio (SINR), is presented. Then this D(3) algorithm is used as an adaptive transformer to create sum (Sigma) and difference (Delta) beams. The residual interference after the D(3) processing is further cancelled by Sigma Delta STAP. The proposed hybrid algorithm using D(3)-Sigma Delta STAP is tested in non-homogeneous clutter modelled using spherically invariant random variables (SIRV) and artificially injected discrete interferers. Performance of the proposed methods is compared with those of traditional statistical approaches, illustrating significant benefits of hybrid processing in non-homogeneous scenarios.