Abstract: In order to meet the needs of efficient management and application of massive, multi-sourc and heterogeneous data in the process of intelligent development of coal mines, this paper proposes an intelligent lakehouse system for coal mine big earth data based on multiple big data technologies. In response to the needs in efficient indexing of multi-source heterogeneous coal mine geological data, this paper proposes a spatiotemporal classification method based on Hilbert curves and GeoHash coding, which uniformly encodes multidimensional attributes such as time, space and classification, reduces the data index dimension, and improves retrieval efficiency. The construction of the intelligent lake house system not only optimizes the integration and utilization efficiency of coal mine geological data, but also provides important practical experience for the digital transformation and transparent development of coal mining enterprises, and promotes the coal mining industry to move towards intelligence, transparency, and high efficiency.