Data supporting "Disentangling operational and geological controls on induced seismicity with distributed acoustic sensing"

# Data Files and Descriptions This dataset contains the files used to generate the figures and analyses in the accompanying manuscript. ## 1. File: `Figure2a.dat` & `Figure2b~i.dat`* **Description:** This binary data file contains the raw Distributed Acoustic Sensing (DAS) record acquired in January.* **Usage:** This dataset was used to generate the visualizations presented in Figures 2a through 2i. ## 2. File: `Figure3.xlsx`* **Description:** This spreadsheet contains the processed spectral-attribute data, including "peak frequency" and "peak amplitude," calculated for each effective channel over time.* **Usage:** These data were used to construct the heatmaps shown in Figure 3.* **Note on Data Processing:** The attributes in this file were derived from the raw DAS data following a specific processing workflow. A summary of this workflow is provided in the "Processing Methodology" section below. ## 3. File: `Figure6.xlsx`* **Description:** This spreadsheet contains the spectral attributes (peak frequency and peak amplitude) partitioned by shift intervals for each stratigraphic unit. The values represent calculations from 10 designated DAS channels within each geological layer.* **Usage:** This dataset was used to create the plots in Figure 6. ## 4. File: `Microseismic_time_series_STALTA.csv`* **Description:** This CSV file contains time-series segments from the raw DAS record that were identified as microseismic events using a Short-Term Average/Long-Term Average (STA/LTA) detection algorithm.* **Usage:** The activity metrics presented in Figures 4 and 5 were derived from this event-detected time series. ## 5. File: `Work_information_statistics.xlsx`* **Description:** This spreadsheet contains daily operational statistics and work information provided by the coal mine.* **Usage:** The data were used to plot the operational context in Figure 5b. --- ## Processing Methodology for `Figure3.xlsx` Data To characterize the spectral response of the borehole DAS array to microseismic waves, the following frequency-domain processing and statistical analysis workflow was applied: 1. **Filtering:** A band-pass filter was first applied to the raw data to suppress low-frequency drift and high-frequency random noise. Subsequently, notch filters were used to eliminate specific narrowband interferences, such as higher-order harmonics. 2. **Power Spectral Density (PSD) Calculation:** The Hilbert envelope of the filtered signal was computed with light smoothing to highlight the primary energy ridge. Power spectral densities (PSDs) were then calculated from the time-sequenced channel records using the Welch method. 3. **Peak Identification:** Within the dominant seismic band (below 60 Hz), spectral peaks were identified. The peak with the maximum amplitude in this band was defined as the fundamental frequency, and its corresponding amplitude was recorded. These two values constitute the "peak frequency" and "peak amplitude" attributes. 4. **Unsupervised Segmentation:** To extract spatially coherent zones from the attribute heatmaps, a tile-quantile unsupervised segmentation algorithm was employed: * **Tiling:** The data matrix was aggregated into tiles of 20 channels by 50 time indices. * **Value Calculation:** A representative value was calculated for each tile. Robust location estimators (e.g., median) were used for peak frequency to mitigate outlier bias, while the arithmetic mean was used for PSD amplitude to preserve energy scaling. * **Clustering:** Tile values were binned by global quantiles and then back-filled to the original pixel resolution. This process groups spectrally similar regions into contiguous blocks, effectively suppressing random noise while preserving patterns related to geological structures. * **Boundary Delineation:** Irregular boundaries were drawn at the transitions between adjacent classes to finalize the segmentation.