4.1 Rainfall control effects
The relationship between their cumulative rainfall P, frequency of light rainfall Nx, frequency of moderate rainfall Nz, frequency of heavy rainfall Nd, and the density of water damage hazard sites SH-ρp in the study area from June to September 2015 to 2021 is shown in Fig. 6. Their rainfall information was obtained from the Gautai National Meteorological Station.
In Fig. 6, the 2015–2021 flood season in the study area is dominated by light rainfall, with intermittent moderate rainfall. The flatness of the two curves indicates a small year-to-year variation in frequency of occurrence. The frequency of heavy rainfall is low, occurring only twice, on 1 September 2018 and 21 September 2020, and their daily rainfall amounts are 35.5 mm and 25.7 mm, respectively. The P-curve of cumulative rainfall during the season shows a significant trend of increasing and then decreasing, with a maximum difference of 57.6 mm. The following conclusions were obtained by analysing the relationship between SH-ρp and P、Nx、Nz, and Nd. The trend of SH-ρp with P, Nx and Nz curves did not show significant regularity, and the correlation coefficients between SH-ρp and P, Nx and Nz were all less than 0.3, with low or no correlation. Nd corresponded to two steep rises in the SH-ρp curve, with a correlation coefficient of about 0.6, and there was There is some positive correlation between the two.
Combining Fig. 3 and Fig. 6 yields the following conclusions, SH-ρx, SP-ρx, SP-ρp, and SH-ρp are generally consistent with the annual trend. Nd has a positive correlation with all of them. Nd has a weak effect on SG-ρp and SG-ρx, causing only a small increase in SG-ρp in 2018.
Therefore, the frequency of heavy rainfall and its intensity has a greater impact on the degree of development and spatial and temporal distribution of pipeline water damage hazards in the foothills, especially on slope water damage, while the impact of moderate and light rainfall is weaker. The climate of the Hexi Corridor is influenced by the warm and humid transition, and the intensity of rainfall generally increases in autumn [22], so the potential threat to pipelines from water damage caused by extreme weather cannot be ignored.
4.2 Slope control effects
With the pipe as the center and 100m on each side as the boundary, the mean value of longitudinal slope L per kilometer width was obtained by RTK section measurement combined with remote sensing image slope extraction. The point density and line density of water damage hazards in the study area with slope are shown in Fig. 7, and the point density of the slope and river channel with slope is shown in Fig. 8.
As can be seen in Fig. 7, in the section from K0 to K20, the pipeline crosses the fan skirt superimposed by the floodplain fan at the northern foot of the Yumu Mountain from west to east, with a gradual increase in slope along the line, varying from 0.039 to 0.064. In the section from K20 to K28, the in-service pipeline crosses the fan skirt diagonally, and the slope along the line gradually decreases, with a varied range of 0.064 to 0.033. The maximum difference in slope variation along the pipeline in the study area is only 0.031, which is a small fluctuation range.
The density curve of water-damaged disaster points does not show any obvious regularity or correlation with the change of slope for the time being. In the section from K0 to K14, the line density of water damage increases with the increase of slope. From K23 to K28, the line density of water-damaged hazards decreases with decreasing slope, and the trend is more significant, and the two are positively correlated, which is consistent with the findings of Zhang et al. [19]. From K14 to K23, the line density of water damage is not positively correlated with slope due to the influence of other factors such as steep hills in the landscape.
In Fig. 8, with the trend of the slope increasing and then decreasing, the overall density of water damage points on the slope surface and river and gully channels does not show a significant pattern of change. Combined with Fig. 5 and Fig. 8, the following conclusions can also be drawn: the river and gully channel line density curves do not correlate with the slope curve for the time being, while the trend of slope surface line density with slope is the same as that of water damage hazard line density with slope.
4.3 Flooded fan superimposition control effect
The micro-geomorphic morphology and hydrological features of the floodplain fan are obvious in the remote sensing images of the northern foot of the Yumu Mountain, where human engineering activities are weak. In this study, the lateral boundaries of typical floodplain fans in the study area were extracted by visual interpretation [27], and the extent of the floodplain fan overlay area was finally determined after field verification by combining the development of alluvial channels and water flow characteristics, as shown in Fig. 9. The influence of the floodplain fan overlay on the distribution of point and line densities of water damage in the study area is shown in Fig. 10, and its influence on the distribution of water damage on slopes and river gully channels is shown in Figs. 11 and 12.
In Fig. 10, the seven flood fan stacked areas are intermittently distributed. The geomorphology of the study area is controlled by fan skirts, with K5 to K6, K17, and K20 to K21 superimposed areas corresponding to four peaks of the point density curve for water damage, and K4, K8, K10 to K11, K13, and K24 to K25 superimposed areas corresponding to five peaks of the line density curve for water damage. The flooded fan-superimposed areas correspond significantly to the crests of the density curve. This indicates that this area has a high concentration of water damage hazards and is a priority for pipeline hazard control.
In Fig. 11, the stacked areas of K4 to K6, K17, and K20 to K21 correspond significantly to the two peaks of the point density curve for water damage in the river and gully channels. The point density of water damage on slopes within the stacked areas of K10 to K11, K13, and K24 to K25 is slightly higher than in the adjacent areas, but this does not cause the larger peaks in the curve to appear. In Fig. 12, the stacked areas of K4 to K6, K17, and K20 to K21 correspond significantly to the two peaks of the line density curve of water damage in the river and gully channels. The stacked areas of K8, K10 to K11, K13, and K24 to K25 correspond to the peaks of the lineal density curve for slope water damage. In summary, the floodplain fan stacking area has a concentration of slope and channel water damage.
4.4 Scarp control effect
The pipeline in the study area runs alongside the 'Camel City Steeple' and crosses it twice, as shown in Fig. 1. The straight-line distance d between the two is assumed to be positive if the pipeline is located upstream of the steeple and negative if it is not. The effect of scarp on the distribution of water damage hazards in the study area is shown in Fig. 13. The effect of the scarp on the distribution of water damage on the slope versus the channel is shown in Fig. 14.
In Fig. 13, the pipeline is located downstream of the scarp in sections K0 to K5 and K20 to K28. In sections K5 to K20, the pipeline is located upstream of the scarp. In sections K0 to K5, the straight-line distance between the pipeline and the scarp decreases in order, and there is no obvious pattern of change in the density of the corresponding water damage hazard points and the line density. From section K5 to K20, the distance between the pipeline and the scarp increases and then decreases, and the density of the corresponding water damage points first decreases and then increases, and the line density first increases and then decreases. From section K20 to K28, as the distance between the pipeline and the scarp l increases, the point density curve of water damage hazards shows a decreasing trend, and the line density curve shows an increasing trend. The trend of increasing and decreasing density tends to stabilize with the change of distance between the pipeline and the scarp.
In Fig. 14, for the section K0 to K5, the point density of slope water damage decreases as the distance between the pipeline and the scarp decreases, and the point density of river gully channel water damage increases. From K5 to K20, as the distance between the pipeline and the scarp increases and then decreases, the point density of water damage on the regional slope shows a steep increase and then a slow decrease. The point density of water damage in the river gully channel shows a trend of steep decrease to 0 places/km and then a steep increase. From K20 to K28, as the distance between the pipeline and the scarp increases, the density of water damage points on the slope surface increases slowly, and the density of water damage points in the river and gully channel decreases steeply to 0 points/km and then stabilizes. Combined with Fig. 5, Fig. 13 and Fig. 14, the following conclusions can be drawn: with the change of distance between the steep hills and the pipeline, the trend of the change of the slope surface line density is generally consistent with the change of the water damage line density, and the trend of the change of the river and gully channel line density is basically consistent with the change of its point density. In the area where the scarp and the pipeline are in close proximity to each other and intersect, the density curve of water damage on the slope surface shows two significant troughs, and the density curve of water damage on the corresponding river and gully channels shows two significant peaks.