Landscape characteristics influence the spatial pattern of soil water storage: Similarity over times and at depths

Catena 116 (2014) 68–77

Contents lists available at ScienceDirect

Catena journal homepage: www.elsevier.com/locate/catena

Landscape characteristics influence the spatial pattern of soil water storage: Similarity over times and at depths Asim Biswas ⁎ Department of Natural Resource Sciences, McGill University, 21111 Lakeshore Road, Ste-Anne-de-Bellevue, QC H9X 3V9, Canada

a r t i c l e

i n f o

Article history: Received 23 June 2013 Received in revised form 10 December 2013 Accepted 11 December 2013 Keywords: Landscape Landform units Scale Time stability Wavelet coherency

a b s t r a c t Similarity in the spatial patterns of soil water storage (SWS) over time and at depths at multiple scales and locations reflects the similarity in the underlying hydrological processes. The objective of this study was to examine the similarity in the spatial patterns of SWS and its characteristic landscape positions for variable soil depths and over time at a field scale. Soil water content (further converted to SWS by multiplying with depth) was measured for five years (2007–2011) along a transect of 128 points at a study site that has representative hummocky landscape of the North American Prairie Pothole region. Surface (0–20 cm) and subsurface (20–140 cm at 20 cm interval) soil water contents were measured using time domain reflectometry and a neutron probe, respectively. High rank correlation coefficient between the measurements over time and at any depth layers (surface = 0–20 cm, root zone = 0–60 cm and total active soil profile = 0–120 cm) indicated strong similarity of the spatial patterns of SWS and thus the underlying hydrological processes. The spatial patterns at large scales (N 72 m) were contributed by alternating knolls and depressions (dominant macro-topographical variations in this type of landscape) and were very similar between any measurement times and depth layers. Similarity over time was changed at medium scales (18–72 m) due to the changes in the landform elements. However, changes in the small-scale (b18 m) spatial patterns were not associated with any landscape characteristics. Similarity was increased at different scales with increase in soil depth owing to strong buffering capacity. Information on the similarity of the spatial patterns at different scales and locations can be used to identify change in sampling domain as controlled by hydrological processes operating at different scales and locations and thus can deliver maximum information with minimum sampling efforts. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Soil water is a primary limiting factor for semi-arid ecosystem functioning (Porporato et al., 2004) and a key determinant factor in environmental health (Sun, 1986). It is involved in a wide variety of natural processes (ecological, climatic and geomorphological) that act at different spatio-temporal scales (Entin et al., 2000; Goovaerts, 1998). Knowledge on the behavior of soil water storage (SWS) and its spatio-temporal distribution provides essential information on various hydrologic, climatic, and general circulation models, weather prediction, evapotranspiration and runoff (Famiglietti and Wood, 1995), precipitation and atmospheric variability (Koster et al., 2004). A number of highly heterogeneous factors and processes acting in different intensities over a variety of scales make the distribution of SWS highly variable in space and time and pose a challenge in hydrology and climate studies (Quinn, 2004). Due to high spatio-temporal variability in SWS, a large number of observations or samples are necessary to characterize a field. Fortunately, the factors and processes controlling SWS exhibit non-random ⁎ Corresponding author. Tel.: +1 514 398 7620; fax: +1 514 398 7990. E-mail address: [email protected]. 0341-8162/$ – see front matter © 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.catena.2013.12.004

spatial patterns consisting of a regular or systematic variation within a catchment or a field and are termed as the ‘spatial organization’ (Western et al., 1999). For example, factors like topography, lithology, parent material, climate and vegetation (Van Wambeke and Dulal, 1978) may result in a distinct and consistent pattern in the distribution of SWS within a catchment or a field (Grayson and Western, 1998; Kachanoski and Dejong, 1988). Moreover the scaling heterogeneity of factors makes the spatial pattern of SWS highly scale-dependent. For example, at small catchment and hill slope scale factors like water routing processes (Anderson and Burt, 1978), differential radiation (Western et al., 1999), heterogeneity in soil (Famiglietti et al., 1998) and vegetation (Hupet and Vanclooster, 2002) control SWS spatial pattern on and within the landscape. The saturation excess water at a particular location is important for runoff producing processes in many catchments (Anderson and Burt, 1978) resulting in a systematic organization in SWS associated with topographic convergence (Barling et al., 1994). Whereas, the atmospheric, geologic, and climatic variability determines the organization of SWS over a large area (Entin et al., 2000; Schneider et al., 2008). The contribution of infiltration, runoff, and lateral redistribution makes the small-scale variability in the spatial pattern of SWS of utmost interest in hydrological studies at small and medium catchments or watersheds.

A. Biswas / Catena 116 (2014) 68–77

Spatial pattern of SWS was found to be similar from a repeated measurement within a catchment or a field. This is because some locations are found to be consistently wetter or drier than the field-averaged SWS yielding a similar spatial pattern over time. Vachaud et al. (1985) used Spearman's rank correlation to examine the similarity in the overall spatial pattern and cumulative probability function of relative mean difference to examine the similarity in the rank of individual locations over time. The concept of similarity in the spatial pattern of SWS was termed as time stability (Vachaud et al., 1985) and has been investigated over a range of study area, sampling scheme, sampling depth, study period, and land use (Hu et al., 2009; Kachanoski and Dejong, 1988; Martinez-Fernandez and Ceballos, 2003; Pachepsky et al., 2005). Time stability was defined as a time invariant association between spatial location and classical statistical measures of SWS most often the mean (Grayson and Western, 1998). One of the important applications of this concept has been the identification of time stable locations, which can considerably reduce the number of sampling locations in obtaining mean SWS for an area of interest. Scale dependence in the time-stable spatial patterns of SWS was also investigated. Kachanoski and Dejong (1988) used the spatial coherency analysis to examine the similarity of the spatial patterns as a function of spatial scale. However, information on the characteristic landscapes of the scale-dependent time-stable spatial patterns was out of reach of this study. In extracting scale information, spectral analysis looses the spatial information. In a previous study, Biswas and Si (2011e) examined the scale-dependent time-stability of the spatial pattern of SWS within a season (intra-season), between different seasons (inter-season) and between same season from different years (inter-annual) using the wavelet coherency. The intra-season time stability was found to be stronger than inter-annual time stability, which was stronger than inter-season time stability (Biswas and Si, 2011e). However, information on how the similarity in the spatial pattern persists with the increase of the time difference between measurements is missing. The change in the similarity of the scale-dependent spatial patterns over time and its characteristic landscape positions can provide a complete picture of the hydrological dynamics in the field and is an important area to explore. Similarity in the overall spatial pattern of SWS over time at a deep soil profile was also examined (Hu et al., 2009; Pachepsky et al., 2005). There are fewer studies that have explored the changes in the time stability of the overall spatial pattern with depth (Cassel et al., 2000). Biswas and Si (2011b) examined the scale-specific similarity of the spatial patterns of SWS of the surface and various subsurface layers measured at a point of time. However, there is no information on the scale-specific time stability of the spatial pattern of SWS at classified surface layer, root zone and total active soil profile. The surface layer is the part of the soil zone that is subject to climate forcing and the root zone is where the majority of the plant roots are located and exhibits strong variability in SWS over time. The total active soil profile is the zone below which seasonal changes in SWS are suppressed. Various processes operating at soil layers of different depths determine the hydrology of soil layers. As the depth of soil layer changes, the dominant processes controlling the spatial pattern of SWS within the layer may also change. Therefore, information on the similarity of the scalespecific spatial pattern of SWS at different layers of soil with variable depths and over time would help better understand soil water dynamics from the surface to the whole soil profile and its temporal evolution. The objective of this study was to examine a) the similarity in the spatial patterns of SWS at variable soil depths over time at a field scale and b) the landscape characteristics of these similar spatial patterns over time and at depths. Information on the spatial variability of SWS over time and at depths would provide a complete picture of the hydrological dynamics within the soil profile. The similarity of the spatial patterns of SWS measured over five years was examined at multiple scales using the wavelet coherency along a transect from the prairie pothole region of North America.

69

2. Materials and methods 2.1. Study area and site selection A field study was conducted at St. Denis National Wildlife Area (52°12′ N latitude, 106°50′ W longitude), Saskatchewan, Canada (Fig. 1). Detailed description of study site, measurement of soil water and controlling factors and instrument calibration can be found in Biswas and Si (2011b), Biswas and Si (2011d, 2011e) and Biswas et al. (2012). Briefly, the landscape of the study site is hummocky, typical of the North American Prairie Pothole region (Hogan and Conly, 2002), the largest wetland landscape in North America encompassing an approximately 780,000 km2 area from the north-central United States to south-central Canada (Fig. 1) (National Wetlands Working Group, 1997). The hummocky landscape is formed during the last glacier retreat and is characterized by a complex sequence of slopes extending from different-sized rounded depressions to irregular complex knolls and knobs (Huel, 2000). A 571.5 m long transect was established with 128 sampling points extending in the north–south direction over several knoll–depression cycles ensuring repeatability (Fig. 1). Sampling points were selected at a regular interval (4.5 m) to catch the systematic variability in the landscape. Geographic locations and the elevation of the sample points were noted using a Trimble Pro XRS Global positioning system (Trimble Navigation, Sunnyvale, CA). Topographic survey of the study site was completed using an aerial light detection and ranging (LiDAR) survey at a 5 m ground resolution. A digital elevation map was prepared at the same ground resolution using SURFER (Golden Software Inc., Golden, CO) (Fig. 1). The landform elements along the transect were characterized as convex (CX), concave (CV), uncultivated wetlands (UW) and cultivated wetlands (CW) (Fig. 1) following Pennock and Corre (2001). Convex elements are topographically high positions with a positive profile curvature. Concave elements are positions with a negative profile curvature. Cultivated wetlands are depressions, roughly circular in shape, which temporally collect rain and snowmelt water and also known as seasonal depressions (Woo and Rowsell, 1993). Non-agricultural positions of the transect were classified as uncultivated wetlands. Soil of the study area is classified as Dark Brown Chernozem according to the Canadian System of Soil Classification (Mollisol in USDA soil classification system) and developed from moderately fine to fine textured, moderately calcareous, unsorted glacio-lacustrine deposits and modified glacial till (Saskatchewan Centre for Soil Research, 1989). The climate of the study area is mainly semiarid with the mean annual air temperature (at the Saskatoon airport, 40 km west of the study site) of 2 °C with the monthly mean of − 19 °C in January and 18 °C in July. The vegetation of the study site was mixed grass seeded in 2004 and allowed to grow each year. Before the grass was seeded, the area was under cultivation and the history of cultivation was used in classifying landform elements (Fig. 1). 2.2. Data collection Soil water content (SWC) at each sampling point was measured down to 140 cm. Surface 0–20 cm SWC was measured using vertically installed time domain reflectometry (TDR) probe and a metallic cable tester (Model 1502BpTektronix, Beaverton, OR). A standard calibration ffiffiffiffiffi equation, θ ¼ 0:115 ka −0:176 following Topp and Reynolds (1998) was used to calculate the SWC for the surface layer. Where the dielectric constant ka = (L2/L)2, L2 is the distance between curves and L is the length of the TDR probe. SWC from 20 to 140 cm was measured using a neutron probe (Model-CPN 501 DR Depthprobe, CPN International, Martinez, CA). A site specific calibration was completed for the neutron probe comparing volumetric water content and neutron count ratio at different topographic locations along the transect over a three year time period (2007–2009) with different initial soil water conditions. The final calibration equation is θv = 0.8523 R + 0.0612 with n = 101 and r2 = 0.86, where R is the ratio of the neutron count to

70

A. Biswas / Catena 116 (2014) 68–77

Fig. 1. Geographic location of study site at St. Denis National Wildlife Area within the Prairie Pothole Region in North America with a three-dimensional and a cross sectional view of the study transect with different landform elements.

the standard neutron count. SWC measurement at each sampling location and depth was converted to SWS by multiplying with the depth of soil layer (here 20 cm for each measurement). Water storage at various soil layers (20 cm each) was added together to get the SWS for the surface layer (0–20 cm), the root zone (0–60 cm) and the total active soil profile (0–120 cm). The surface soil layer was classified irrespective of soil horizons present at different landscape positions and the average presumed depth of roots was considered as the root zone soil layer. The soil profile in the study area undergoes freeze–thaw cycle and the most active part of the soil profile is generally just over a meter depth. Therefore, the water storage at up to 120 cm depth was considered as the total active soil profile to catch the variations in SWS. SWS was measured for 25 times over a 5-year period (2007–2011) at various environmental situations including snowmelt, rainfall or prolonged dry period to get a representative of different soil water conditions. 2.3. Data analysis Similarity in the overall spatial pattern of SWS was examined using the Spearman's rank correlation analysis following Vachaud et al. (1985). Scale specific similarity of the spatial patterns of SWS at different depth layers was examined using the wavelet coherency. Wavelet

coherency required the calculation of wavelet coefficients for each of the two data series and cross wavelet coefficients between those data series. Wavelet coefficients at different scales and locations were calculated using the wavelet transform (Graps, 1995). The square of the wavelet coefficients (wavelet spectra) represented the variance of the spatial series at a particular scale and location, therefore dividing the overall spatial variance (Mallat, 1999). While the wavelet transform can divide the overall variance in a spatial series into multiple scales and locations, the cross wavelet transform provides the covariance between two spatial series at multiple scales and locations (Biswas et al., 2008; Grinsted et al., 2004; Si and Zeleke, 2005). Cross wavelet spectrum or the covariance at a scale and location is the product of two variables at that scale and location. Therefore, it depends on the individual variance of those variables at that scale and location (Si and Zeleke, 2005). Wavelet coherency is the cross spectrum normalized by the product of the variance of two variables. Therefore, the wavelet coherency calculates the scale-dependent correlation coefficients at different locations (Biswas and Si, 2011a; Biswas et al., 2008; Si and Zeleke, 2005) and can be plotted as a color contour graph with locations along the Xaxis, scales along the Y-axis and the coherency values along the Z-axis. The scale at the Z-axis (color scale) indicated the degree or the strength of the correlation (red is the strongest), black solid line indicated 95%

A. Biswas / Catena 116 (2014) 68–77

significant correlation and the direction of the arrow indicated the type of correlation with right being positive and left being negative. Based on the dominant coherencies, the scales were classified into three groups; small scales (b18 m), medium scales (18–72 m) and large scales (N72 m). Wavelet coherencies were calculated at scales between 9 and 200 m at all locations along the transect. There were a total of 54 scales at each of the 128 locations. Therefore, the total number of wavelet coherency was 6912. In the contour plot, each significant coherency occupied a pixel representing the correlation at a particular scale and location. The total area of significant relationship in each coherency spectra can be calculated by counting the number of pixels representing significant coherencies over all scales and locations (Biswas and Si, 2011d). Wavelet coherency between two spatial series of SWS measured up to certain depth in two different occasions would provide the information on the similarity in the spatio-temporal pattern of SWS at multiple scales and locations. The locations of the similarity in the spatial patterns at different scales provide information on the landscape characteristics. Wavelet coherency spectra were calculated between any two measurements. Due to space restriction, I only present the spectra between the first and the remaining measurements within the same year. Theory of the wavelet transform and the wavelet coherency is well established in the literature. A detailed discussion on the methodology or the theory of the wavelet transform can be found in Farge (1992) and Kumar and Foufoula-Georgiou (1997) and the wavelet coherency can be found in Grinsted et al. (2004), Si and Zeleke (2005), Biswas et al. (2008), Biswas and Si (2011a) and others and is beyond the scope of this manuscript. The wavelet coherency analysis was completed using the MATLAB code (The MathWorks Inc., Natrick, MA) written by Grinsted et al. (2004) and is freely available at URL: http://noc.ac.

71

uk/using-science/crosswavelet-wavelet-coherence (Verified on 22 June 2013). Minor modifications of the code were performed to suit the data analysis required for this manuscript. 3. Results and discussion The average SWC varied over the measurement period and with depth (Fig. 2). A general decreasing trend in SWC was observed from the beginning to the later part of the growing season due to the geographic location of the study area and its environmental settings. The study area received about 30% of the total precipitation (30 year annual average is 350 mm) as snow during the winter months (Pomeroy et al., 2007). The spring and early summer rainfall contributed to high SWC in addition to the snowmelt that occurred within a very short period (Flerchinger and Cooley, 2000; Gray et al., 1985; Winter and Rosenberry, 1995). During this time, the only source of water loss was evaporation and to a lesser extent, the ground water interaction (Hayashi et al., 1998). However, the transpirative demand of the growing vegetation reduced the water content at the later part of the year and showed a decreasing trend in SWC (Fig. 2). Excessive spring and summer precipitation in 2010 (645 mm during April to October) (Environment Canada, 2011) made the growing season extremely wet almost without any decreasing trend. The surplus water from 2010 was stored in soil until 2011 and yielded a weak decreasing trend. Above average precipitation in 2009 (average precipitation of 2007, 2008, and 2009 was 366 mm, 331 mm, and 402 mm, respectively) also made the trend weak (Fig. 2). The trend was a little different with depth. In wetter situation (e.g., early part of every year), the surface layer had the highest SWC which gradually decreased with the increase in the thickness of the soil layer such as the root zone

0

30

0-20 cm 0-60 cm 0-120 cm

45

4 3 2 1

Skewness

Coefficinet of variation (%)

15

0

0.5

0.3

Sept 29, 2011

Jun 6, 2011

Jun 29, 2011

May 13, 2011

Jun 14, 2010

Sept 28, 2010

Apr 6, 2010

May 19, 2010

Oct 27, 2009

Aug 27, 2009

Jul 21, 2009

May 7, 2009

May 27, 2009

Oct 22, 2008

Apr 21, 2009

Aug 23, 2008

Sept 17, 2008

July 16, 2008

Jun 21, 2008

May 31, 2008

May 2, 2008

Oct 12, 2007

Sept 1, 2007

0.1

Jul 17, 2007

0.2

Aug 7, 2007

Average soil water content (cm3/cm3)

-1 0.4

Measurement date Fig. 2. Average soil water content (cm3/cm3), coefficient of variation (%) and skewness in soil water content measured over a five-year period at the surface layer (0–20 cm), the root zone (0–60 cm) and the total active soil profile (0–120 cm).

72

A. Biswas / Catena 116 (2014) 68–77

or total active soil profile (Fig. 2). However, the surface layer water content decreased at an elevated rate and became drier at the later part of the year compared to the root zone and total active soil profile (Fig. 2). There was a positive correlation between SWC and standard deviation (SD) (correlation coefficient, r = 0.77, 0.75, and 0.67 for the surface layer, root zone and total active soil profile, respectively). However, the SD decreased as depth increased (Fig. 2). The wet surface layer in spring (wet period) exhibited a large SD compared to the drier deep layers. Similarly, the dry surface layer in fall (dry period) exhibited a small SD compared to the wetter deep layers (Fig. 2). High coefficient of variation (CV) during the wet period gradually decreased in the later part of the growing season or in the dry period (Fig. 2). The trend in the CV with depth and time was very similar to that in the SD (Fig. 2). A similar positive correlation was also observed between the SWC and the CV (r = 0.50, 0.38, and 0.17 for the surface layer, root zone and total active soil profile, respectively). While some studies reported a positive correlation (Famiglietti et al., 1998; Martinez-Fernandez and Ceballos, 2003), others reported a negative (Famiglietti et al., 1999; Hupet and Vanclooster, 2002) correlation between SWC and CV. A negative exponential relationship was also reported as the average SWC decreased with the increase of sampling extent (Choi et al., 2007; RodriguezIturbe et al., 1995). However a positive correlation was observed in this study with a fixed sampling extent. High organic matter in the surface soil layer of a grassed field resulted in low bulk density and high porosity and thus the high SWC during the wet period. At the same time, a prolonged dry period removed more water from the surface layer compared to the whole soil profile and made the surface layer highly dynamic. The OC content along the transect was highly variable (CV = 41%) and contributed to the variability in the surface SWC. Variability in the SWC was also reflected in the statistical distribution. During wet period, SWC at few locations with standing water was calculated from the bulk density assuming complete saturation. Very high SWC at those locations made the distribution positively skewed with large skewness value compared to the near normal distribution in the dry period with small skewness value (Fig. 2). Skewed nature of the distribution was more prominent in the surface layer compared to the deep layers. This may be due to the dynamic nature of the surface layer resulting from its exposure to the environmental forcing such as solar radiation and wind (Hu et al., 2009, 2010). Moreover, plants generally take up more than 70% of the water they need from the top 50% of the root zone (Feddes et al., 1978). Increased root activity in the surface layer depleted water more quickly compared to the whole root zone and the total active soil profile. This resulted in high SD and CV in the surface layer. In contrast, less response to meteorological conditions (Hu et al., 2010), less root activity (Cassel et al., 2000) and less disturbed soil structure (Pachepsky et al., 2005) in the deep soil profile resulted in less variations in SWC. Variability in SWC was also affected by its input. In the hummocky landscape, generally depressions receive snow from the surrounding uplands or knolls (effect of strong prairie wind) and create an uneven spatial distribution (Fang and Pomeroy, 2009; Woo and Rowsell, 1993). Moreover, the frozen soil surface during snowmelt did not allow complete infiltration of the snowmelt water. The divergent characteristics of the rounded knolls redistributed water in the landscape and created a wide range in SWC between knolls and depressions (Flerchinger and Cooley, 2000; Gray et al., 1985; Winter and Rosenberry, 1995). High SWC in the depressions and low on the knolls created a spatial pattern that is almost a mirror image of the spatial pattern of the relative elevation (RE) (Biswas and Si, 2011e). Biswas and Si (2011e) examined the similarity of the spatial patterns of SWS (SWC × depth) and reported the degree of similarity in the order of intra-season N inter-annual N inter-season. The change of similarity with time and depth is important in understanding the soil water dynamics at the surface layer, root zone and the total active soil profile and is studied in this manuscript.

There was a strong correlation (r2 = 0.57 to 0.95 for 0–20 cm, r2 = 0.70 to 0.98 for 0–60 cm and r2 = 0.80 to 0.99 for 0–120 cm) between any two measurements (within a year or between years) (Fig. 3). All correlation coefficients were significant at P b 0.0001. Very strong correlation was observed between any two consecutive measurements and gradually decreased with the increase of time difference between measurements (Fig. 3). For example, the value r2 between 2 May 2008 and 31 May 2008 was 0.91 and 21 April 2009 and 7 May 2009 was 0.94 (between first and second day of measurement), which gradually decreased to 0.63 and 0.64, respectively (between first and last day of measurement of respective year). The decreasing trend was observed at all depths with the strongest at the surface layer due to its dynamic nature. At any point of time, the strongest correlation was observed at the deepest layer due to high buffering capacity (Biswas and Si, 2011b). The pattern repeated over years (Fig. 3) irrespective of the total precipitation. Similarity in the overall spatial patterns of SWS over time and depth clearly indicated the similarity in the underlying hydrological processes. The large-scale (N 72 m) coherencies between 17 July and 7 August in 2007 were significant at all locations and all depth layers (Fig. 4) indicating a similarity in the large-scale spatial patterns. The mediumscale (18–72 m) coherencies were only significant from the beginning of the transect to around 150 m and from 360 m to the end of the transect at the root zone and the total active soil profile (Fig. 4) and was slight different from the surface layer. At the surface layer, significant coherencies were only observed from the beginning of the transect to about 200 m at the scales from 36 to 72 m (Fig. 4). There were few significant coherencies at small scales (b 18 m) mainly at the beginning and end of the transect (Fig. 4). These small-scale variations were contributed by the localized variations in SWS and the measurement errors and were not particularly associated with any landscape characteristics. The large-scale variability was evolved from the alternating knoll– depression cycles repeating at the scale of about 100 m (Biswas and Si, 2011a). The medium-scale variations were attributed to the landform elements which varied at a scale of around 20–70 m. More variations in landform elements at the middle part of the transect have resulted in the loss of consistent spatial patterns due to the differential exposure of landform elements to the environmental forcing and meteorological conditions. Significant coherency at a scale and location indicated the similarity in the spatial pattern of SWS at that scale and location and might have evolved from similar underlying hydrological processes. Moreover, persistence of the relationship indicated a consistent underlying control. Non-significant coherencies at small and medium scales at few locations indicate dissimilar spatial pattern at those scales and locations. Total area of significant coherencies was increased with depth (Fig. 4) indicating the increase in the degree of similarity in the spatial pattern and thus the underlying hydrological processes. The change in the area of significant relationship with depth was larger at small scales compared to the medium and large scales (Fig. 4) due to the increased buffering capacity with depth. The coherency between 17 July and 1 September and 17 July and 12 October in 2007 were very similar to that of between 17 July and 7 August in 2007 at all depths (Fig. 4). However, the total area of significant coherency gradual decreased with the increase in time difference between measurements but increased with the depth of soil layers. Similar hydrological processes tend to yield similar spatial patterns within a short time interval. However, gradual change in environmental conditions and vegetation development over time changed the spatial pattern. The dynamic behavior of the surface layer resulted more change in the coherency pattern at small- and medium-scales. However, strong effect from alternating knolls and depressions (macro-topography) was persistent at large scales at different depths and over time (Biswas and Si, 2011c). Wavelet coherency spectra between 2 May 2008 and the remaining measurements (31 May, 21 June, 16 July, 23 August, 17 September, and

A. Biswas / Catena 116 (2014) 68–77

73

0-20 cm 0-60 cm 0-120 cm 1.0

Rank Correlation Coefficient

0.8 0.6 0.4 0.2 1.0 0.8 0.6 0.4

Sept 29, 2011

Jun 6, 2011

Jun 29, 2011

Sept 28, 2010

May 13, 2011

Jun 14, 2010

Apr 6, 2010

May 19, 2010

Oct 27, 2009

Aug 27, 2009

Jul 21, 2009

May 7, 2009

May 27, 2009

Oct 22, 2008

Apr 21, 2009

Sept 17, 2008

July 16, 2008

Aug 23, 2008

Jun 21, 2008

May 31, 2008

May 2, 2008

Oct 12, 2007

Sept 1, 2007

Jul 17, 2007

0.0

Aug 7, 2007

0.2

Measurement Date Fig. 3. Rank correlation coefficients between the first day and the remaining days of measurements in each year and the first day (17 July 2007) and the remaining days of measurement over five year's time at the surface layer (0–20 cm), the root zone (0–60 cm) and the total active soil profile (0–120 cm).

Fig. 4. Wavelet coherency between 17 July 2007 and 7 August, 1 September and 12 October of 2007 at the surface layer (0–20 cm), root zone (0–60 cm) and the total active soil profile (0–120 cm). The X axis indicates location along the transect (m), the Y-axis indicates scale (m), the color scale indicates the strength of correlation, the black solid line indicates 95% significance level and the direction of the arrow indicates the type of correlation.

74

A. Biswas / Catena 116 (2014) 68–77

Fig. 5. Wavelet coherency between 2 May 2008 and 31 May, 21 June, 16 July, 23 August, 17 September and 22 October of 2008 at the surface layer (0–20 cm), root zone (0–60 cm) and the total active soil profile (0–120 cm). The X axis indicates location along the transect (m), the Y-axis indicates scale (m), the color scale indicates the strength of correlation, the black solid line indicates 95% significance level and the direction of the arrow indicates the type of correlation.

22 October) in 2008 are shown in Fig. 5. Total area of significant coherency gradually decreased at all depths with the increase of time difference between measurements (Fig. 5). This indicated a gradual change in underlying hydrological processes and thus the effect from various controlling factors (Biswas and Si, 2011c). Total area of significant coherency increased with the increase in depth of soil layers at any point of time (Fig. 5) due to the increased buffering capacity. The environmental conditions and vegetation development created a change in the spatial patterns at small- and medium-scales with increasing time difference between measurements (Fig. 5). There were two trends in the wavelet coherency in 2008. The relationship between 2 May and 31 May, 21 June and 16 July were very

similar. During this part of the year, the snowmelt runoff and rain water redistribution contributed to high SWS and created a spatial pattern at large scales controlled by alternating knolls and depressions (Biswas and Si, 2011c; Kachanoski and Dejong, 1988). However, during the later part of the growing season with developed vegetation, the evapotranspiration demand created a different spatial pattern. The relationship between 2 May and 23 August, 17 September and 22 October was very similar but different from the relationship between the measurements at the early part of the growing season. These two situations were classified as the recharge and the discharge period (Biswas and Si, 2011b; Kachanoski and Dejong, 1988). There are no quantitative criteria to identify these periods. However, the change in the wavelet coherency

Fig. 6. Wavelet coherency between 21 April 2009 and 7 May, 27 May, 21 July, 27 August and 27 October of 2009 at the surface layer (0–20 cm), root zone (0–60 cm) and the total active soil profile (0–120 cm). The X axis indicates location along the transect (m), the Y-axis indicates scale (m), the color scale indicates the strength of correlation, the black solid line indicates 95% significance level and the direction of the arrow indicates the type of correlation.

A. Biswas / Catena 116 (2014) 68–77

indicated the change in the spatial patterns and thus the underlying hydrological processes. This separation was not possible for the year 2007 (Fig. 4) as the measurements were completed within the discharge period (July onwards). A change in the coherency pattern was also observed in 2009 (Fig. 6) and 2011 (Fig. 8). However, the change was not distinct in the year of 2010 (Fig. 7) due to high amount of precipitation throughout the growing season. While the change in the spatial patterns at small-scales was not associated with any landscape positions, the change at medium scales was associated with landform elements along the transect. More variations in landform elements at the middle part of the transect yielded dissimilar spatial pattern over time (non-significant coherencies) while less variations at the later part of the transect yielded similar spatial pattern. Fewer variations in landform elements resulted in similar hydrological processes and thus the spatial pattern. Alternating knolls and depressions or macro-topography controlled hydrological processes yielded very similar large-scale spatial patterns over time (within a year and between years). The depressions in the hummocky landscape often hydrologically disconnect from nearby depressions and act as an independent micro-watershed. The possible ways of water loss from these places were either evaporation or evapotranspiration and deep drainage. However, deep drainage through fine till material in the hummocky landscape could be as low as 2 to 40 mm per year (Hayashi et al., 1998; van der Kamp et al., 2003) mainly occurring through fractures and preferential flow paths. Therefore, the main vector of water loss from this landscape was the evapotranspiration by growing vegetation. The localized vegetation change created small scale variations while landform elements created medium-scale variations in SWS. However, alternating knolls and depressions or the macro-topographic variations controlled the redistribution of water in the landscape and determined the spatial pattern at large scales. Large amount of water allowed lush growth of grasses in the depressions, while small amount of water on

75

knolls restricted plant growth and evapotranspiration. Therefore, the relative change in the water storage maintained the spatial patterns at large scales. The snowmelt and spring precipitation determined the spatial pattern of SWS at the beginning of the growing season. The phenomenon repeated over years yielding similar spatial patterns in SWS over the years and at depths. 4. Conclusions This paper examined the similarity in the spatial patterns of SWS measured at different depths over five years from a hummocky landscape of the Prairie Pothole Region of North America. Visually, the spatial distribution of SWS was almost a mirror image of the spatial distribution of relative elevation. High rank correlation coefficients indicated overall similarity in the spatial patterns. The similarity was the highest between measurements completed closest in time and gradually decreased with the increase in time difference between the measurements. The trend in the correlation repeated over years. Change in environmental conditions with time controlled underlying hydrological processes and thus determined the spatial pattern in SWS. Similarity at multiple scales was examined using the wavelet coherency. Large scale (N72 m) spatial patterns were very consistent over time and at depths and were contributed by the alternating knolls and depressions. Similarity in the spatial patterns of SWS at medium scales (18–72 m) changed over time. Variations in landform elements were the major driving factor controlling the spatial pattern at this scale range. Small scale (b18 m) similarities were mainly random to the landscape positions and were contributed by the localized variations in SWS and possible measurement errors. These similarities were stronger with the increase in depth of soil layers due to the increased buffering and less exposure to the variable environmental forcing. Higher root activity and exposure to variable meteorological conditions made the surface

Fig. 7. Wavelet coherency between 6 April 2010 and 19 May, 14 June, and 28 September of 2010 at the surface layer (0–20 cm), root zone (0–60 cm) and the total active soil profile (0–120 cm). The X axis indicates location along the transect (m), the Y-axis indicates scale (m), the color scale indicates the strength of correlation, the black solid line indicates 95% significance level and the direction of the arrow indicates the type of correlation.

76

A. Biswas / Catena 116 (2014) 68–77

Fig. 8. Wavelet coherency between 13 May 2011 and 6 June, 29 June and 29 September of 2011 at the surface layer (0–20 cm), root zone (0–60 cm) and the total active soil profile (0–120 cm). The X axis indicates location along the transect (m), the Y-axis indicates scale (m), the color scale indicates the strength of correlation, the black solid line indicates 95% significance level and the direction of the arrow indicates the type of correlation.

layer more dynamic and thus the SWS spatial patterns at this layer. Change in the coherency pattern within a year helped in identifying the change in moisture states and its temporal evolution. The temporal scales of the change in spatial patterns were not identifiable with this data structure and required continuous monitoring of SWS in the landscape and are the future direction of study. Similarity and dissimilarity in the spatial patterns of SWS over time and at depths indicated the similarity or dissimilarity in the underlying hydrological processes. Dominant scales and locations of these processes can be determined from the scale–location specific relationships of the spatial patterns occurring between temporal measurements. Therefore, the change in sampling domain as controlled by hydrological processes operating at different scales and locations can deliver maximum information with minimum sampling efforts. Acknowledgments The funding for this project from the Natural Science and Engineering Research Council (NSERC) of Canada and the University of Saskatchewan is highly appreciated. A special thanks to Professor Bing Si for the guidance in completing the project. Help in field data collection from the members of soil physics team in the Department of Soil Science at the University of Saskatchewan is also highly appreciated. References Anderson, M.G., Burt, T.P., 1978. Role of topography in controlling throughflow generation. Earth Surf. Process. Landf. 3, 331–344. Barling, R.D., Moore, I.D., Grayson, R.B., 1994. A quasi-dynamic wetness index for characterizing the spatial-distribution of zones of surface saturation and soil-water content. Water Resour. Res. 30, 1029–1044. Biswas, A., Chau, H.W., Bedard-Haughn, A.K., Si, B.C., 2012. Factors controlling soil water storage in the hummocky landscape of the Prairie Pothole Region of North America. Can. J. Soil Sci. 92, 649–663.

Biswas, A., Si, B.C., 2011a. Application of continuous wavelet transform in examining soil spatial variation: a review. Math. Geosci. 43, 379–396. Biswas, A., Si, B.C., 2011b. Depth persistence of the spatial pattern of soil water storage in a hummocky landscape. Soil Sci. Soc. Am. J. 75, 1099–1109. Biswas, A., Si, B.C., 2011c. Identifying scale specific controls of soil water storage in a hummocky landscape using wavelet coherency. Geoderma 165, 50–59. Biswas, A., Si, B.C., 2011d. Revealing the controls of soil water storage at different scales in a hummocky landscape. Soil Sci. Soc. Am. J. 75, 1295–1306. Biswas, A., Si, B.C., 2011e. Scales and locations of time stability of soil water storage in a hummocky landscape. J. Hydrol. 408, 100–112. Biswas, A., Si, B.C., Walley, F.L., 2008. Spatial relationship between delta(15)N and elevation in agricultural landscapes. Nonlinear Process. Geophys. 15, 397–407. Cassel, D.K., Wendroth, O., Nielsen, D.R., 2000. Assessing spatial variability in an agricultural experiment station field: opportunities arising from spatial dependence. Agron. J. 92, 706–714. Choi, M., Jacobs, J.M., Cosh, M.H., 2007. Scaled spatial variability of soil moisture fields. Geophys. Res. Lett. 34. Entin, J.K., Robock, A., Vinnikov, K.Y., Hollinger, S.E., Liu, S.X., Namkhai, A., 2000. Temporal and spatial scales of observed soil moisture variations in the extratropics. J. Geophys. Res.-Atmos. 105, 11865–11877. Environment Canada, 2011. Canada's Top Ten Weather Stories for 2010. (Available at http://www.ec.gc.ca/meteo-weather/default.asp?lang=En&n=7E58ECA3-1 (Verified on 23/06/2013)). Famiglietti, J.S., Devereaux, J.A., Laymon, C.A., Tsegaye, T., Houser, P.R., Jackson, T.J., Graham, S.T., Rodell, M., van Oevelen, P.J., 1999. Ground-based investigation of soil moisture variability within remote sensing footprints during the Southern Great Plains 1997 (SGP97) Hydrology Experiment. Water Resour. Res. 35, 1839–1851. Famiglietti, J.S., Rudnicki, J.W., Rodell, M., 1998. Variability in surface moisture content along a hillslope transect: Rattlesnake Hill, Texas. J. Hydrol. 210, 259–281. Famiglietti, J.S., Wood, E.F., 1995. Effects of spatial variability and scale on areally averaged evapotranspiration. Water Resour. Res. 31, 699–712. Fang, X., Pomeroy, J.W., 2009. Modelling blowing snow redistribution to prairie wetlands. Hydrol. Process. 23, 2557–2569. Farge, M., 1992. Wavelet transforms and their applications to turbulence. Annu. Rev. Fluid Mech. 24, 395–457. Feddes, R.A., Kowalik, P.J., Zaradny, H., 1978. Simulation of Field Water Use and Crop Yield. Halsted Press, John Wiley and Sons Inc., New York, NY. Flerchinger, G.N., Cooley, K.R., 2000. A ten-year water balance of a mountainous semi-arid watershed. J. Hydrol. 237, 86–99. Goovaerts, P., 1998. Geostatistical tools for characterizing the spatial variability of microbiological and physico-chemical soil properties. Biol. Fertil. Soils 27, 315–334. Graps, A., 1995. An introduction to wavelets. IEEE Comput. Sci. Eng. 2, 50–61.

A. Biswas / Catena 116 (2014) 68–77 Gray, D.M., Landine, P.G., Granger, R.J., 1985. Simulating infiltration into frozen prairie soils in streamflow models. Can. J. Earth Sci. 22, 464–472. Grayson, R.B., Western, A.W., 1998. Towards areal estimation of soil water content from point measurements: time and space stability of mean response. J. Hydrol. 207, 68–82. Grinsted, A., Moore, J.C., Jevrejeva, S., 2004. Application of the cross wavelet transform and wavelet coherence to geophysical time series. Nonlinear Process. Geophys. 11, 561–566. Hayashi, M., van der Kamp, G., Rudolph, D.L., 1998. Water and solute transfer between a prairie wetland and adjacent uplands, 2. Chloride cycle. J. Hydrol. 207, 56–67. Hogan, J.M., Conly, F.F., 2002. St. Denis national wildlife area land cover classification: 1997. Technical Report Series Number 384. Canadian Wildlife Service, Prairie and Northern Region. Prairie Northern Wildlife Research Centre, Saskatoon, SK. Hu, W., Shao, M., Han, F., Reichardt, K., Tan, J., 2010. Watershed scale temporal stability of soil water content. Geoderma 158, 181–198. Hu, W., Shao, M.A., Wang, Q.J., Reichardt, K., 2009. Time stability of soil water storage measured by neutron probe and the effects of calibration procedures in a small watershed. Catena 79, 72–82. Huel, D., 2000. Managing Saskatchewan Wetlands: A Landowner's Guide. Saskatchewan Wetland Conservation Corporation, Regina, SK 69. Hupet, F., Vanclooster, M., 2002. Intraseasonal dynamics of soil moisture variability within a small agricultural maize cropped field. J. Hydrol. 261, 86–101. Kachanoski, R.G., Dejong, E., 1988. Scale dependence and the temporal persistence of spatial patterns of soil-water storage. Water Resour. Res. 24, 85–91. Koster, R.D., Dirmeyer, P.A., Guo, Z.C., Bonan, G., Chan, E., Cox, P., Gordon, C.T., Kanae, S., Kowalczyk, E., Lawrence, D., Liu, P., Lu, C.H., Malyshev, S., McAvaney, B., Mitchell, K., Mocko, D., Oki, T., Oleson, K., Pitman, A., Sud, Y.C., Taylor, C.M., Verseghy, D., Vasic, R., Xue, Y.K., Yamada, T., Team, G., 2004. Regions of strong coupling between soil moisture and precipitation. Science 305, 1138–1140. Kumar, P., Foufoula-Georgiou, E., 1997. Wavelet analysis for geophysical applications. Rev. Geophys. 35, 385–412. Mallat, S., 1999. A Wavelet Tour of Signal Processing. Academic Press, New York, NY. Martinez-Fernandez, J., Ceballos, A., 2003. Temporal stability of soil moisture in a largefield experiment in Spain. Soil Sci. Soc. Am. J. 67, 1647–1656. National Wetlands Working Group, 1997. The Canadian Wetland Classification System. In: Warner, B.G., Rubec, C.D.A. (Eds.), Wetlands Research Centre. University of Waterloo, ON.

77

Pachepsky, Y.A., Guber, A.K., Jacques, D., 2005. Temporal persistence in vertical distributions of soil moisture contents. Soil Sci. Soc. Am. J. 69, 347–352. Pennock, D.J., Corre, M.D., 2001. Development and application of landform segmentation procedures. Soil Tillage Res. 58, 151–162. Pomeroy, J.W., de Boer, D., Martz, L.W., 2007. Hydrology and water resources. In: Thraves, B. (Ed.), Saskatchewan: Geographic Perspectives. CRRC, Regina, SK, Canada. Porporato, A., Daly, E., Rodriguez-Iturbe, I., 2004. Soil water balance and ecosystem response to climate change. Am. Nat. 164, 625–632. Quinn, P., 2004. Scale appropriate modelling: representing cause-and-effect relationships in nitrate pollution at the catchment scale for the purpose of catchment scale planning. J. Hydrol. 291, 197–217. Rodriguez-Iturbe, I., Vogel, G.K., Rigon, R., Entekhabi, D., Castelli, F., Rinaldo, A., 1995. On the spatial-organization of soil-moisture fields. Geophys. Res. Lett. 22, 2757–2760. Saskatchewan Centre for Soil Research, 1989. Rural Municipality of Grant, Number 372: Preliminary Soil Map and Report. Publication# SK372. Saskatchewan Centre Soil Research, University of Saskatchewan, Saskatoon, SK. Schneider, K., Huisman, J.A., Breuer, L., Zhao, Y., Frede, H.G., 2008. Temporal stability of soil moisture in various semi-arid steppe ecosystems and its application in remote sensing. J. Hydrol. 359, 16–29. Si, B.C., Zeleke, T.B., 2005. Wavelet coherency analysis to relate saturated hydraulic properties to soil physical properties. Water Resour. Res. 41, W11424. Sun, M., 1986. Groundwater ills—many diagnoses, few remedies. Science 232, 1490–1492. Topp, G.C., Reynolds, W.D., 1998. Time domain reflectometry: a seminal technique for measuring mass and energy in soil. Soil Tillage Res. 47, 125–132. Vachaud, G., Desilans, A.P., Balabanis, P., Vauclin, M., 1985. Temporal stability of spatially measured soil-water probability density-function. Soil Sci. Soc. Am. J. 49, 822–828. van der Kamp, G., Hayashi, M., Gallen, D., 2003. Comparing the hydrology of grassed and cultivated catchments in the semi-arid Canadian prairies. Hydrol. Process. 17, 559–575. Van Wambeke, A., Dulal, R., 1978. Diversity of soils in the Tropics. American Society of Agronomy Special Publication 34, 13–28. Western, A.W., Grayson, R.B., Bloschl, G., Willgoose, G.R., McMahon, T.A., 1999. Observed spatial organization of soil moisture and its relation to terrain indices. Water Resour. Res. 35, 797–810. Winter, T.C., Rosenberry, D.O., 1995. The interaction of ground-water with prairie pothole wetlands in the cottonwood lake area, east-central north-dakota, 1979–1990. Wetlands 15, 193–211. Woo, M.K., Rowsell, R.D., 1993. Hydrology of a prairie slough. J. Hydrol. 146, 175–207.