The gully source point is the most active part on the shoulder line. The spatial length of streamlines from the gully source point to the upstream watershed boundary line and the downstream gully line is an important identifications for spatial structure of the loess shoulder line, watershed boundary line and gully line. Proximity degree of gully source point to watershed boundary line spatially is a key point for quantifying the geomorphological development process in loess basin. To explore the loess landform areas watershed gully source point to approach watershed boundary line, reveal the loess watershed landscape development process and the method of main erosion processes. In this paper, A core factor of quantifying the three line spatial structural: Proximity Index (PI) was established from horizontal and vertical dimensions, including Horizontal Proximity Index (HPI) and Vertical Proximity Index(VPI). 42 study sites of 16 geomorphic types in the Loess Plateau of northern Shaanxi based on the digital elevation model data with 5 m resolution were selected and the spatial differentiation of their mean values (MHPI, MVPI) in the Loess Plateau of northern Shaanxi was discussed by using of the digital terrain analysis method. Besides, 4 typical watersheds including 5 whole gully levels were selected from 42 sites as the key study areas in Chunhua, Yijun, Ganquan, and Suide which located in loess tableland, loess residual tableland, loess ridge, and loess hill respectively in the north-south sequence. The Mean Proximity Index Variability (MPIV) of key study sites was calculated in the watershed scale. Results show that: ① There is a strong spatial autocorrelation of the Mean Proximity Index(MPI) in the Loess Plateau of northern Shaanxi. MHPI presents a trend of increasing firstly and then decreasing in the north-south sequence, and gradually decreasing in the east-west direction. MHPI reaches the maximum in the valley hilly area along the Yellow River. MHPI decreases firstly and then increases from southwest to northeast, and gradually decreases from northwest to southeast, and reaches the minimum in the Loess tableland area of Weibei region; ② The positive and negative MHPI values in watershed scale are sensitively related to loess tableland and hilly gully regions; ③ MHPI and MVPI of the 4 key sample areas are consistent with other terrain indexes in the north-south sequence. The MHPI of 104 external confluence region has a good correlation with the average slope (P=0.43, a<0.001),while the MVPI has a strong correlation with the hypsometric integral (P=0.75, a<0.001).The Mean Proximity Indexes (MPI) comprehensively take into consideration the spatial relationship of the three typical structural control lines in the Loess Plateau and have obvious indication significance for the development degree of the loess landform.