U.S. flag

An official website of the United States government

Other Access

The information on this page (the dataset metadata) is also available in these formats:


via the DKAN API

Data Extent

Data from: Threshold Behavior of Catchments with Duplex Hillslope Soils Feeding Soil Pipe Networks

Location of Goodwin Creek Experimental Watershed in Mississippi, USA and Catchments within GCEW of the study site.

Soil pipes are large soil pores that are generally parallel to the hillslope surface and are created by internal erosion of a pore due by rapid water flow. The internal erosion of the soil pipe can result in collapse of the soil pipe which produces surface expressions such as flute holes (small vertical opening connected to the soil pipe), sinkholes (closed depressions), and gullies. Soil pipes tend to form in soil with a water-restricting sublayer, i.e. duplex soils, that foster the development of perched water tables on hillslopes and thus lateral flows. This dataset corresponds with two published studies conducted on loess covered catchments in northern Mississippi, USA within the Goodwin Creek Experimental Watershed that contain extensive networks of soil pipes and corresponding collapse features. These loess soils contain fragipan layers that were found to perch water, thereby initiating the piping processes. The dataset contains data from two papers, specifically these include: (i) the spatial distribution of soil pipe collapses and their size measurements from the Wilson et al. (2015) paper, and (ii) hydrologic measurements of perched water tables on hillslopes, water levels of selected soil pipe locations, and precipitation from the Wilson et al. (2017) paper.

The 2,132 ha Goodwin Creek Experimental Watershed (GCEW) is located in Panola County, Mississippi. The mean annual precipitation is 1,358 mm with high mean monthly precipitation of 130 mm for December through May and lower (monthly mean of 93 mm) but potentially more intense convective thunderstorms during July through October. Based upon long-term records of streamflow in GCEW, the following four 3-month cycles were delineated: high flow (HF) season (January through March) during which rainfall averages 4.29 mm/d, streamflow averages 2.22 mm/d and the streamflow is increasing by 0.16 mm/d during this season; drying out (DO) season (April through May) with 4.03 mm/d rainfall and 1.29 mm/d of streamflow on average but with a rapid decrease in streamflow of 0.56 mm/d; low flow (LF) season (July through September) with 2.82 mm/d rainfall and just 0.35 mm/d streamflow that is stable (only 0.05 mm/d decrease); and a wetting up (WU) season with 3.80 mm/d rainfall and 1.12 mm/d of streamflow that is noted for a rapid increase of 0.87 mm/d.

The watershed is in the Loess-Bluff hills physiographic province with an elevation range from 71 to 128 m above msl. The upland areas have a thin Pleistocene age loess surface that is highly erodible. Data are reported from two catchments, Main or C2 and the Back or C3, which are located on the north-western side of GCEW. The upper contributing area of the Main catchment has three branches (West, Middle, and East) with the thalweg below their confluence extending down to an edge-of-field gully that is connected to Goodwin Creek. The Back catchment was defined at a point close to the spring that serves as the soil pipe outlet. Both catchments are mapped as Loring silt loam soil (fine-silty, mixed, active thermic Oxyaquic Fragiudalf) or gullied (unidentified soil due to erosion) with percentage Loring by area of 98, and 55 % for the Main and Back, respectively.

Soil pipe collapses were first detected in December 2012 at which point an intensive field reconnaissance was made to determine the spatial distribution of soil pipe collapses. The location of each pipe collapse feature was determined by differential GPS (Topcon GR-3). The accuracy of this GPS was 1 cm in horizontal direction and 1.5 cm in vertical direction. Pipe collapse features were surveyed for their spatial location, their dimensions measured manually, and classified by type feature, e.g. sinkhole (closed depression), flute hole, gully window (linear coalescence of flute holes or linear collapse of a soil pipe). For small gully window features, the center location was surveyed and the depth, width, and length measured manually. For large gully windows, the differential GPS was used to survey along the upper edge and the bottom of the gully. The initial survey was completed in January 2013 with additional surveys made as new locations were discovered or features were enlarged by further erosion and a complete survey was repeated in the summer of 2014.

Onset Hobo U20L pressure transducers were installed in 10 pipe collapse features in the three Branches (West, Middle, and East) of the Main Catchment and six collapse features in the Back Catchment (Figure 2). For the West Branch, one transducer was installed in the upper most collapse feature (P87) while the Middle Branch had pressure transducers in three pipe collapses (P129, P49, P44), and the East Branch had a transducer at the head of the upper most gully (G2) and two downslope pipe collapses (P13, P8). In addition, three transducers were installed in pipe collapse features in the Swale area (P36, P27, and P2) downslope of the West Branch where the three branches converge. Hobo transducers were installed in the three branches of the Main Catchment on January 31, 2013 and installed in the Back Catchment on 13 April, 2014. The Main Catchment soil pipe transducers were removed 3 July, 2013 and reinstalled on 5 November, 2013. For both Catchments these record water pressures on five minute intervals. In addition, a soil pipe collapse feature at the upper extent of each Branch, designed WH, MH, EH, was equipped with a Siemens VS100 water flow velocity meter and a Keller American pressure transducer (Figure 2). Each of these three sites had a CSI datalogger for recording flow velocity and flow depth (pressure) and transmitting these data to the laboratory. Only pressure head data are presented and these values are expressed as a water level (Head, cm) above the bottom of the pipe collapse feature.

The spatial location, depths and sizes of soil pipes are presented in data files. The West Branch site (WH) was the upper most extent of gully G1, just below a 97 cm deep flute hole (56) and the soil pipe was 22 cm diameter where it entered G1. The Middle Branch site (MH) was in a 40 cm diameter flute hole (P80) that extended 43 cm down to a soil pipe of unknown diameter. The East Branch site (EH) was in a flute hole (P19) with an 8 to 14 cm wide oblong opening at the surface that extended down 39 cm to a soil pipe of unknown diameter. The flow velocity/pressure measurements (three Branches) started recording at two minute intervals on 22 February, 2013. The Back catchment was not equipped with flow velocity meters. The flow contributing area of the Main Catchment, as defined by surface topography at a surface runoff culvert, was 4.33 ha; the contributing area entering the soil pipe networks of the three Branches, as defined at WH, MH, and WH, was 0.93, 0.15, and 0.54 ha, respectively; the area of the Back Catchment was 1.36 ha as defined at the outlet spring (Figure 2).

Three shallow wells, with a 40 cm slotted section and 10 cm solid pipe bottom were installed into the fragipan to 82 cm depth, i.e. slotted section was 32-72 cm depth, on the hillslopes of each Branch in the Main Catchment (Figure 2) to monitor perched water tables. Wells were installed along a transect perpendicular to the slope from a soil pipe collapse pressure measurement location, 7.6 m linear distance apart to 22.9 m upslope from the soil pipe. Wells were installed by auguring a 63.5 mm diameter hole with a bucket auger to 82 cm depth. The annulus of the slotted section was filled with coarse filter sand while the solid well section above was filled with bentonite and excavated soil to seal the well from the surface. The solid well section extended 5 cm above the surface and was capped with a vented cap. Wells were aligned with the WH site on the West Branch, flute hole 129 on the Middle Branch and EH on the East Branch with the West, Middle and East well transects having 8.5%, 8.1%, and 13.6% slopes, respectively. Each well was equipped with a Hobo U20L water level recorder that recorded on five minute intervals starting 5 November, 2013. Well pressure heads are relative to the well bottom corrected to 72 cm depth.

Rainfall was measured with a Texas Electronics TR-525USW tipping bucket rain gauge with a 20 cm diameter funnel and a tipping bucket that records with 1% accuracy for rainfall rates up to 50 mm/h. The rain gauge used to represent precipitation for the catchments was about one km from the study site. Rainfall intensity and energy was determined using RIST 3.94 (ars.usda.gov/Research/docs.htm?docid=3251) from the time record of each tip of the rain gauge bucket. Individual storms were defined by a minimum six hour break without rainfall. RIST determined the total storm precipitation (mm), storm duration (hr), maximum intensity (mm/hr) for 5, 10, 15, 30 and 60 minute periods, kinetic energy of the storm and 30 minute erosion index (EI30). The kinetic energy of individual storms was determined according to the equation by Brown and Foster (1987). The time of first rain gauge tip was used as the start time for storm events.

The data provided begin with the installation of the instruments and continues to the end of WY 2015 (9/30/2015). Gaps in the timeline occur when instruments were either offline or not functioning properly in which case no data are provided. The start of each event defined as the time of the first rise in pressure. Soil pipe hydrographs commonly did not return to baseline (typically zero) pressure before a subsequent event started. This was even more common for well hydrographs. For such conditions, the ending time was defined either by the start of a subsequent event (pressure rise) or the time it reached baseline pressure. In addition to water pressure, the dataset also includes the “magnitude” or Hydrograph Area with time during each event which was calculated as the product of the average pressure (m) and the measurement interval (min), which was either 2 or 5 minute recording interval depending upon the instrument. Periods between events did not have a magnitude and therefore a dummy value of -9999 was used.

Release Date
Spatial / Geographical Coverage Area
POLYGON ((-89.879984665022 34.259590794517, -89.879984665022 34.267464619477, -89.866938400373 34.267464619477, -89.866938400373 34.259590794517))
Spatial / Geographical Coverage Location
Goodwin Creek Experimental Watershed is located in the bluff hills of the Yazoo River basin of northern Mississippi, with the outlet at longitude 89o 54' 50" W and latitude 34o 13' 55" N.
Ag Data Commons
Temporal Coverage
January 1, 2013 to December 31, 2015
Contact Name
Wilson, Glenn
Contact Email
Public Access Level
Program Code
005:040 - Department of Agriculture - National Research
Bureau Code
005:00 - Department of Agriculture