This paper proposes a novel method for abnormal state detection in low SNR environment by employing Maximum Eigenvalue of Sample Covariance Matrix (MESCM) for developing the theory and method of data-driven power grid situation awareness. Inspired by the random matrix theory, it firstly constructs a data source matrix, and obtains a moving window matrix and its standard matrix, then acquires the sample covariance matrix. In this way, the situation awareness and early warning for interconnected power systems could be achieved by MESCM calculation and its violation check. Utilizing PSS/E? software, the case studies have been carried on an IEEE 39-bus system and a planning system of China Southern Power Grid, involving two main working conditions such as abnormal load change and short circuit fault. The results show that the proposed methodology has the advantage of higher noise resistance and less computing time in comparison with the traditional mean spectral radius based method and preliminarily verifies that it would be robust under incomplete information. This work is supported by National Natural Science Foundation of China (No. 51567006), Program for Top Science & Technology Talents in Universities of Guizhou Province (No. 2018036) and Guizhou Province Science and Technology Fund (No. [2019]1100).