Trend-Constrained DTW for Horizontal Well In-Zone Geosteering with Azimuthal Gamma-Ray Data

In unconventional horizontal drilling, limited logging measurements pose challenges for accurate in-zone positioning and geosteering. This study proposes an azimuthal gamma-ray trend-constrained dynamic time warping (DTW) method with adaptive significance-threshold updating. The relative responses of upper and lower azimuthal gamma-ray measurements are used to identify stratigraphic movement trends and constrain the DTW endpoint search space. A fixed-start–sliding-end matching strategy estimates the endpoint position using an independently field-interpreted starting position. An XGBoost classifier predicts the significance threshold from window-level features to improve adaptability to varying gamma-ray responses. The predicted endpoint position is further combined with wellbore trajectory information to estimate the local apparent formation dip for auxiliary steering. Leave-one-well-out cross-validation on five model-development wells (247 windows) yielded a well-level macro-averaged mean absolute error (MAE) of 0.332 m relative to field-interpreted reference positions. Independent validation on two additional horizontal wells, H1 and H2, achieved MAEs of 0.314 and 0.338 m, respectively, representing reductions of 12.8% and 13.8% compared with fixed-threshold trend-constrained DTW. Compared with unconstrained DTW, the proposed method reduced the MAE by 55.9% and 44.8% on H1 and H2, respectively. These results demonstrate the potential of combining azimuthal trend constraints with adaptive threshold updating for gamma-ray-based in-zone positioning and support its applicability to previously unseen wells within the investigated geological block.