探地雷达瞬时属性融合的土壤分层识别技术研究OA

Soil stratification structure is a critical factor governing water-salt transport and crop growth in farm-land,and its accurate identification is essential for assessing tillage quality and optimizing water-salt manage-ment.Addressing the challenges of rapid electromagnetic wave attenuation and the difficulty of identifying deep weak interfaces in arid saline-alkali soils,this study proposes a soil stratification identification method based on the fusion of ground-penetrating radar(GPR)instantaneous attributes.The method was validated using typical cotton fields with drip irrigation durations of 16 a and 26 a in Xinjiang.Using a 1000 MHz high-frequency GPR,shallow soil structure data were acquired,and instantaneous attributes were extracted via the Hilbert transform.On this basis,a time-gradient enhanced layer boundary fusion index was constructed to achieve weighted integra-tion of envelope energy and time-gradient information,thereby suppressing noise and enhancing weak reflection signals.The results indicate that the proposed fusion method effectively overcomes the limitations of single-attri-bute analysis,clearly identifying the topsoil,plow layer,plow pan,and subsoil within the 0-60 cm depth range.Compared with measured profiles,the absolute error of layer boundary identification ranged from 0.55 to 4.83 cm(mean:2.04 cm),and the relative error ranged from 1.78%to 16.60%(mean:9.09%),demonstrating high detec-tion precision.Further application of this method revealed that long-term drip irrigation(26 a)caused a marked downward migration of the boundaries between the plow layer and plow pan,quantitatively characterizing the re-modeling of the soil profile by long-term tillage.The attribute fusion approach proposed in this study provides a high-precision,non-destructive detection means for identifying farmland soil stratification under complex geolog-ical conditions.