Combined soil-terrain stratification for characterizing catchment-scale soil moisture variation

Abstract

1Soil properties and terrain characteristics influence spatiotemporal patterns of soil 2 moisture across a watershed. To improve the predictive power of landscape hydrology models, it 3 is essential to consider both soil and terrain attributes when stratifying a catchment into similar 4 hydrologic functional units. In this study, we developed and validated a new catchment-scale 5 stratification scheme for the Shale Hills watershed by combining soil and terrain attributes in an 6 attempt to delineate soil-landscape units with similar soil moisture dynamics. Terrain was 7 combined with soils information by first using a Random Forest supervised classification 8 algorithm to predict a detailed soil map using 47 field soil samples and terrain variables derived 9 from 1-m LiDAR. A slope class map generated from the LiDAR-derived digital elevation model 10 (DEM) was overlaid on the predicted soil map to delineate areas of similar slope value across the 11 hillslope. We compared the performance of this new stratification scheme with two classical 12 stratification schemes, a soil map developed from detailed field survey and a landform unit map 13 based on the DEM, for estimating soil moisture time-series across the forested watershed. The 14 combined soil-terrain method outperformed classical stratification schemes in estimating soil 15 moisture time-series over a 4-year period. Our results demonstrate that combining soil and 16 terrain attributes can help improve the stratification of a catchment into similar soil hydrologic 17 functional units, which is valuable to distributed hydrology modeling and other applications. 18 19Soil type 21 Understanding the link between soil moisture patterns and landscape features is critical to 23 improving landscape hydrologic modeling (Band et al., 1993; Pauwels et al., 2001; Yu et al., 24 2014). A common assumption in catchment hydrology is that terrain places a dominant control 25 on hydrologic functions (Beven and Kirkby, 1979; Winter, 2001). This assumption leads many 26 researchers to parameterize hydrologic models based on landforms or sub-catchment units using 27 terrain alone. Since topographic information in the form of digital elevation models (DEM) has 28 been increasingly available, stratifying catchments into similar hydrologic functioning units with 29 terrain has been widespread (Moore et al., 1991; Bloschl and Sivapalan, 1995; Winter, 2001). 30However, field-based soil properties are often not directly included in these stratification 31 schemes, and terrain is assumed to be a proxy for inferring soil properties. These assumptions 32 remain largely unchallenged, since many catchment hydrologic studies do not validate terrain-33 based sub-catchment units using in situ collected soil moisture data or compare model 34 performance with actual soil distributions. 35Topographically-based stratification approaches have been continuously improved over 36 time with advancements in GIS and remote sensing technologies. Following the conceptual 37 work by Beven and Kirkby ...

Nested Scales of Spatial and Temporal Variability of Soil Water Content Across a Semiarid Montane Catchment

“…Differences in results across systems and measurement approaches suggest that assessment of which terrain metrics best represent individual hydrologic processes in various topographic and climatic environments should be explored further. For example, shallow soil moisture (10–40 cm) could be related to topographic wetness indices (UAA and TWI), whereas deeper soil moisture could be more related to elevation and vertical distance to stream, as was observed at Shale Hills (Baldwin et al, ). We suggest that the strong seasonality of catchment wetness at TCEF provides an ideal laboratory for examining when and to what degree individual topographic metrics might reflect hydrologic processes.…”

Simulating high‐resolution soil moisture patterns in the Shale Hills watershed using a land surface hydrologic model

“…The relatively small variation in vegetation distribution tested has negligible effects on the soil moisture pattern. Baldwin et al (submitted) calculated the temporal autocorrelation of soil moisture (2007–2010) from the 61 sites across Shale Hills. They found that the combination of topography and a digital soil map was significantly more accurate in characterizing watershed‐scale soil moisture variation than five other prevailing stratification systems, which indicates that the soil moisture pattern at the Shale Hills watershed is controlled by topography and soil types; our model results agree with their findings.…”

Observing and Simulating Spatial Variations of Forest Carbon Stocks in Complex Terrain

“…Most locations estimate volumetric water content between 5‐ and 40‐cm depth, with valley floor and swale sites having extra measurements down to 1.3‐m depth (Lin et al, ). The soil moisture measurements at different depths are grouped into surface (top layer, usually represented by 5‐cm depth) and root zone (the average of all other layers) for simplicity (Baldwin et al, ). Here we used observed root zone soil moisture and observed 8‐cm soil temperature to drive the model.…”

“…Previous studies in Shale Hills watershed have investigated hydrologic and biogeochemical processes in complex terrain. Shi et al (, ) and Baldwin et al () have shown predictable patterns of soil moisture, which are closely related to soil physical properties and topographic features. Weitzman and Kaye () calculated the N budget in the watershed and indicated that the watershed is N limited even given the relatively high N deposition that exists at present.…”