Römer-Stange, Nikolas; Merl, Maximilian; Baltz, Anna Katharina; Meyer, Maximilian; Albers, Aike; Asakura, Yukari; Feddersen, Greta; Schwenk, Tilmann; Spieß, Volkhard: 2D multichannel seismic reflection processed data (working area dataset) of RV SONNE during cruise SO304, Bengal Fan, Indian Ocean [dataset]. PANGAEA, https://doi.pangaea.de/10.1594/PANGAEA.995405 (dataset in review)

High‑resolution, multichannel reflection‑seismic data were acquired during the RV Sonne cruise SO304 (22 February – 07 April 2024) to the Bay of Bengal. The archived profile SO304‑GeoB24‑019‑MCS‑td96‑uGI was recorded on 08 March from 06:25 UTC to 09:25 UTC to image a lobe complex on the Bengal Fan. Acquisition employed a Sercel Micro Generator‑Injector air‑gun and a 96‑channel Teledyne streamer; it started at 14°58.732′ N 087°12.296′ E and ended at 14°58.542′ N 087°26.036′ E. Five Digicourse Model 7010 bird units were used to steer the streamer at a target depth of 0.7 m. Seismic signals were digitised with the MaMuCS system (Dr. Keil, University of Bremen). The processed seismic sections were generated from the raw records by first establishing the geometry of each shot‑receiver pair from the ship's GPS position and lay‑back geometry. The data were then binned on a rectangular grid that follows the smoothed ship track, using a nominal bin spacing of 10 m and a bin size of 20 m with the overlapping bins improving signal‑to‑noise ratio. After editing, the effective fold is ≈ 100 traces per bin (the original fold ranged from 120 to 140, with ≈ 18 % of traces rejected). Pre‑stack processing began with a debias step, followed by a minimum‑phase Butterworth band‑pass filter (corner frequencies 25 Hz, 50 Hz, 700 Hz and 1 400 Hz). To suppress noise generated by the depth‑control birds, an FK‑filter was applied after randomising the order of traces within each bin. Water‑column noise was removed by muting all samples that lie 20 ms above the picked seafloor. Stacking employed normal‑move‑out (NMO) correction using an approximated velocity model and an 80 % stretch‑mute to discard excessively stretched traces. A horizontal despike filter (600‑ms window, scaling factor 1.2) was applied, after which median stacking was performed with a maximum fold of 100. The stacked volume was further cleaned with a post‑stack FK filter. Migration was carried out with a finite‑difference algorithm that uses an interval‑velocity model derived from the NMO analysis. Migration was implicitly limited to dip angles of 45°, although the algorithm can handle dips up to 60° without instability. The final seismic image has a vertical resolution of 0.7–1 ms (≈ 0.5–1 m in depth, according to Ricker's criterion) and a horizontal resolution of 50–70 m, estimated from the Fresnel‑zone radius at a water depth of 3 000 m where incidence angles are nearly vertical. Coordinates are given in WGS84/UTM45N. SEG-Y Rev1.0 (2017), header locations are: 21 CMP Number (4-byte Integer); 71 Coordinate Scalar (2-byte Integer); 73&181 X-Position CMP Center (4-Byte Integer); 77&185 Y-Position CMP Center (4-byte Integer); 109 Recoding delay in ms (2-byte Integer).