Liu, Y.L., Li, Z.C., Yang, G.Q. and Liu, Q., 2018. An improved RGB attribute for fluid or gas identification based on MP. Jounal of Seismic Exploration, 27: 319-330. One single technology cannot satisfy the requirements of reservoir identification and prediction. More technologies are being used jointly to improve the predicted accuracy of reservoir parameters. For the purpose of recognizing fluid or gas, this paper gives a method of improving red, green, blue (RGB) attributes based on the matching pursuit (MP) algorithm. This is a new attempt to use the two methods jointly in order to find oil and gas. The role of the MP algorithm in our method is different from conventional usage. An iterative process is used to select the region that has large amplitude. The approach has three steps. First, the data is divided into three frequency bands after spectral analysis. Second, the divided data is processed by a MP algorithm, which selects the region with large amplitude. This approach is based on the idea that a favourable reservoir is often a sandstone with strong reflection, which is similar to the principle of the “bright spot” technique. This analysis is the most important step and determines the final result. Third, a RGB attribute is calculated that define frequency variation by means of colour mixing or transition. The characteristic of both large amplitude and frequency variation is considered to be the signal of oil or gas. The model and real data demonstrate that this is an effective method of reservoir prediction. Compared with the common RGB attributes, this improved RGB attribute could give a more accurate prediction.