The exponential growth of dynamic monitoring data for reservoir development poses critical challenges, including transmission bandwidth, storage pressure, and latency. To solve these problems, a novel adaptive compression algorithm based on temporal and spatial correlation is introduced. A modified ARMA model based on time-series analysis is uniquely combined with the improved Kriging interpolation method to describe spatial correlation. A hybrid coding strategy based on Huffman-arithmetic has been proposed to reduce redundancy while preserving data fidelity. Superior performance was demonstrated by experimental validation using 6-month pressure, temperature, and flow data (1-min sampling interval). Compared with Huffman encoding (18.2), LZW coding (15.8), and wavelet transform (25.3), respectively, the average compression ratio was 32.6%. Notably, it maintains an average absolute error of only 0.023, representing a 35.3–62.3% reduction compared to benchmark methods. Processing of edge nodes (0.8 ms compression time) and data transfer (12 ms delay) meet real-time operation requirements. It provides a transformative solution to efficiently transmit and analyze reservoir monitoring data, which will have an impact on enhanced decision-making in the field.