HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Web of Science (8) Citing Articles via Google Scholar Google Scholar Articles by Stadler, D. Articles by Flühler, H. Search for Related Content PubMed Articles by Stadler, D. Articles by Flühler, H. GeoRef GeoRef Citation Agricola Articles by Stadler, D. Articles by Flühler, H.

DIVISION S-1-SOIL PHYSICS

Dye Tracing and Image Analysis for Quantifying Water Infiltration into Frozen Soils

ETH Zürich, Institute of Terrestrial Ecology, Grabenstrasse 3, CH-8952 Schlieren, Switzerland

staehli{at}ito.umnw.ethz.ch

	ABSTRACT

TOP ABSTRACT INTRODUCTION Materials and method Results Discussion and conclusions REFERENCES

 
New methods are needed to quantify infiltration into frozen soil, an important issue for agricultural management in northern latitude regions. A dye tracer method that uses digital image analysis and fluorescence imaging is presented for visualizing and quantifying flow pathways in frozen soils. The method was applied to three soil columns in a cold chamber. Two of them were packed with sand and the third was an undisturbed soil monolith. After complete freezing to -5°C, the columns were irrigated at above-freezing temperatures before they were vertically and horizontally sectioned to analyze pictures of the stained profiles and cross-sections. Image analysis was done for either the visual or fluorescent spectral ranges of three different tracers. Small samples were taken from the profiles to calibrate the dye tracer concentration. This was achieved by means of second-order polynomials of the R, G, and B values from the corresponding areas of the pictures with coefficients of determination of 0.92 to 0.99. This method results in concentration maps with a high spatial resolution reflecting the infiltration pattern. The experiment confirmed current hypotheses of infiltration mechanisms into frozen soil in that the infiltrability of the initially wet sand was restricted, whereas in the undisturbed soil monolith, the dye solution infiltrated through preferential pathways which were air filled at the time of freezing.

Abbreviations: DS, dry sand • UM, undisturbed soil monolith • WS, wet sand

	INTRODUCTION

TOP ABSTRACT INTRODUCTION Materials and method Results Discussion and conclusions REFERENCES

 
IN COLD CLIMATES, impeded water infiltration into frozen soil is a crucial problem for agricultural management. The amount of water and solutes transported under such conditions may be substantial during the main snow melt events, posing a potential risk for surface erosion or leaching of pesticides or nitrate. Therefore, it is important to know the governing mechanisms that control the partitioning of water between infiltration or freezing at the soil surface and lateral surface runoff. It is more difficult to predict infiltration into frozen than into unfrozen soils for the following reasons.

1. At below-freezing temperatures, heat and water fluxes are strongly coupled. Hence, not only the hydraulic but also the thermal soil properties need to be known.
   
2. In addition to the soil matrix, gas, and liquid phases, we also need to consider ice as the fourth phase that varies considerably in time and space. The amount and spatial arrangement of ice in the pore space is important for the infiltration capacity.
   
3. Some of the infiltrating water may refreeze when soil temperatures are below 0°C and thereby change the hydraulic properties.
   

Kane and Stein (1983), Baker and Spaans (1997), Stadler et al. (1997), and Stähli et al. (1999) have among others reported studies on this topic. They demonstrated that ice content (and hence the total water content prior to freezing), soil structure, the thermal regime at the soil surface, and the number of freezing and thawing cycles are the governing factors affecting the infiltration capacity of frozen soil. One-dimensional numerical models for simulating infiltration into frozen soil using combined heat and mass balance equations in layered profiles have been presented by Flerchinger and Saxton (1989), Stähli et al. (1996), and Zhao et al. (1997). These models assume similarity between freezing–thawing and drying–wetting with regard to the dependence of liquid water content upon temperature and matrix potential. Spaans and Baker (1996) demonstrated the validity of this assumption. However, with regard to water flow in frozen soil, it is not sufficient to consider the residual liquid water, which separates ice and solids. Major infiltration can occur through the larger, initially air-filled pores that are separated by the ice phase from the solid particles and the liquid water (Lundin and Johnsson, 1990). It is, thus, important to know the hydraulic conductivity of this fast flow domain (Stadler et al., 1997; Stähli et al., 1999). Seyfried and Murdock (1997) measured air permeability to estimate the hydraulic conductivity of the initially air-filled pores. Another method, though indirect, is to quantify infiltration with dye tracers, and then to estimate the hydraulic conductivity with inverse modeling. In recent years, dye tracer techniques have become an important tool for investigating water infiltration in unfrozen soil. There is ample experimental evidence that the water flow field in unsaturated soils is most often heterogeneous, exhibiting pronounced preferential pathways. An abundance of local measurements are needed to adequately describe the pertinent physical processes of such flow fields. Image analysis allows quantification of flow patterns stained by dye tracers with high spatial resolution. During recent years, this technique has been refined (Schincariol et al., 1993; Aeby et al., 1997) and applied to field soils under unfrozen conditions (Ewing and Horton, 1999; Forrer et al., 2000).

The aim of the present study was to develop a dye tracer technique for visualizing and quantifying water infiltration into frozen soils.

	Materials and method

TOP ABSTRACT INTRODUCTION Materials and method Results Discussion and conclusions REFERENCES

 
Infiltration Experiments
Two experiments were conducted in a cold chamber where the actual temperature fluctuated by ±1°C around the set temperature. Two columns packed with sand, one with a low (DS) and the other with a high initial water content (WS) were used in the first experiment; an undisturbed soil monolith (UM) was used in the second. Bulk density, initial water content, rate, and timing of tracer applications are summarized in Table 1 . The experimental set-up is shown in Fig. 1 .

View larger version (70K): [in this window] [in a new window]   Fig. 1 Set-up of the cold chamber experiment with the packed sand columns. In the case of the undisturbed soil monolith, the polyvinylchloride column walls were replaced by 2.5-cm-thick paraffin layers and 2-cm-thick wooden plates. All distances are given in centimeters

For the second experiment, an undisturbed soil monolith was taken from a forest site in the vicinity of Zürich (Bundt et al., 1998). The soil is a loamy sand with a small stone content. The front and two side walls of the monolith block were contained in a wooden U-shaped frame leaving 2 to 3 cm of free space between frame and soil. The soil was frozen with liquid nitrogen and, subsequently, the space was filled in increments with paraffin. The paraffin stabilized the cut surfaces and prevented water movement along the walls. After the final preparation in the laboratory the monolith measured 45 by 15 by 41 cm. The organic material on the surface and the uppermost 3 cm of the mineral soil were removed. After inserting the thermistors from the side, the back wall was also covered by a 2.5-cm-thick paraffin layer and a wooden plate.

The columns were brought to thermal equilibrium at , where only a small portion of water remained unfrozen. The subsequent infiltration experiment at positive chamber temperature was intended to simulate a short-term snow melt or rain event during winter. The solution applied in the course of the experiment consisted of deionized water, dye tracer and, in the case of the monolith experiment CaBr₂-solution, all at chamber temperature. The dye tracers were (i) Brilliant Blue FCF (color index (C.I.) 42090, 792.85 g mol^-1, absorption maximum: 630 nm), (ii) Brilliant Sulfaflavine (no listed C.I., 418.41 g mol^-1, absorption maximum: 422 nm), and (iii) Azophloxine (= Acid Red 1, C.I. 18050, 509.43 g mol^-1, absorption maximum: 532 nm). A moving sprinkler bar, consisting of 50 syringe needles, irrigated for two 5-h periods at a rate of 5 mm h^-1. The cold chamber was set to above freezing temperatures 2 h before the start of each irrigation, (Table 1), and reset to -5°C 2 h after the irrigation. At the end of the experiment, the uppermost 1.3 cm of the soil monolith consisting of brittle organic and only weakly consolidated material was removed prior to slicing. The frozen columns were sectioned with either a band saw or a radial arm saw equipped with small diamond pieces. Three horizontal cross sections were prepared from the monolith at depths of 1.3, 16 and 31 cm to detect flow pathways stained by Brilliant Blue over the cross-section. From these sections, the stained area at a depth of 1.3 cm was estimated to cover 20% of the entire section, at 16 cm about 7%, while no stained flow pathways were detected at a depth of 31 cm. Hence, vertical profiles were subsequently prepared only from the two blocks between 1.3 to 15 cm and 16 to 30 cm, respectively. The material between a depth of 15 and 16 cm was lost during the sectioning. The vertical profiles were cut 10, 8, and 6 cm away from the instrumentation wall for the monolith and 5, 7.5, and 10 cm from the instrumented wall for the sand columns. Because of possible boundary effects, only the inner 12 cm of the 15-cm-wide columns was used for image analysis. Small soil samples of 0.3 to 1 g (DS, WS; Fig. 2) and 0.02 to 0.1 g (UM; Fig. 3) , respectively, were scratched with a knife from one out of three exposed profiles from each column for image calibration. Then the material between the sectioned vertical planes and that between the column wall and the first section were horizontally cut to determine the final volume-averaged total water content and tracer concentrations. This yielded nine horizontal slices from each of the sand columns and six slices from the monolith.

View larger version (112K): [in this window] [in a new window]   Fig. 2 Geometrically and color-normalized images of the packed sand profiles at 5, 7.5, and 10 cm from the wall: dry sand (DS, top), and wet sand (WS, bottom). Small samples were taken from the locations indicated by squares

View larger version (88K): [in this window] [in a new window]   Fig. 3 Geometrically and color-normalized images (left), and Brilliant Blue concentration maps determined by image analysis (right) of the profiles 10, 8, and 6 cm from the wall of the soil monolith. Pixel size is 1 by 1 mm. Small samples were taken from the locations indicated by squares

Determination of Tracer Concentration from Soil Samples
The small soil samples scratched from the vertical profiles, as well as those from the horizontal sections were dried for 24 h at 80°C. For the sand columns, an aliquot of approximately 0.5 g of each sample was weighed into a disposable extraction column. The extraction columns were placed on a vacuum vessel. As an extraction solvent, 10 mL of a 4 to 1 volume ratio of water and acetone was added and the vessel evacuated at -600 hPa. The extracted solvent was passed through a 0.45-µm filter. In the case of the soil monolith, 0.5 g of soil was mixed with 10 mL of the same extraction solvent and agitated in a horizontal shaker for 10 h. Then, the solution was centrifuged and passed through a 0.45-µm filter.

The dye concentrations were measured spectrophotometrically (Philips, PU 8620 UV/VIS/NIR, Cambridge, UK): Brilliant Blue at a wavelength of 630 nm, Brilliant Sulfaflavine at 430 nm, and Azophloxine at 470 nm (not at its absorption maximum of 532 nm because of interference with Brilliant Blue). The fluorescence of Brilliant Sulfaflavine was measured with a fluorescence spectrometer, and bromide concentrations with HPLC.

The extraction method seemed to be accurate in the case of the packed-sand columns (see recovery rates in Table 2) . In the case of the soil monolith, on the other hand, the mass recovery was incomplete (Table 3) . A test with three replicates of soil material from the monolith yielded mean mass recoveries of 15% at an added Brilliant Blue concentration of 0.040 mg per g soil, 27% at 0.160 mg per g soil, 40% at 0.480 mg per g soil, and 46% at 1.920 mg per g soil. It is not clear whether the apparent sorption observed was actually an adsorption phenomenon, a dye-salt interaction, or a consequence of the chosen extraction procedure.

A gray frame with a ruler was attached to the vertical profile section to correct geometrical distortion of the photographs. A Kodak gray scale was placed onto the frame. The profiles were photographed with standard color slide films without any optical filters.

All analyses were done on a SPARC Ultra 1 (Sun Microsystems, CA) using the software packages PCI, Version 6.0.1 (PCI Geomatics, Richmond Hill, Ontario) and IDL, Version 4.0.1 (Interactive Data Language, Research Systems, Inc., Boulder, CO). The color slides were digitized and stored on a Kodak Photo CD. Because any arbitrary color can be represented as a composition of the three spectral densities in the red (R), green (G), and blue (B) range, each of these colors was saved in a separate image plane with 256 gray values. Images with 1536 by 1024 pixels were used for further processing.

The spectral composition of the daylight or chamber illumination varies in time. This means that a given object, repeatedly photographed at different times, appears in different colors. Therefore, the photographs had to be adjusted to a common standard. For this purpose, the Kodak gray scale was used as a standard. It consisted of 20 reference patches of different gray levels from white to black. Gray is by definition a composite of the same gray values for R, G, and B.

The color of each image is normalized in a two-step process on the basis of the median gray values for each of the 20 patches. First, the median R, G, and B values are stretched from 0 to 255 to calculate a corresponding normalized value. The relationship between the measured reflectance and the ideal values of the stretched patches is shown to be linear. Second, the normalized R, G, and B images are calculated by relating the original digitized values to the normalized value by the relationship established in Step 1.

At each location where a small soil sample was scratched out of the profile, the 20%-trimmed R, G, and B means were computed for an area of 100 to 400 pixels referred to as R_s, G_s and B_s. The logarithms of the dye concentration C of the samples, given in milligrams per kilogram soil, were fitted in a stepwise procedure by a second-order polynomial to the R_s, G_s and B_s values of the corresponding areas. The R_s, G_s and B_s values were normalized to reduce the colinearity problem. The number of terms in the full second-order regression model was reduced to enhance the predictive capability, because the variance of the parameter estimates decreases. On the other hand, the coefficient of determination R² decreases when reducing the number of terms.

For the calibration of the dye tracer concentration in the columns 15 (DS and WS) or 12 (UM) small soil samples were used. The locations of these calibration points are indicated in Fig. 2 and 3. The second-order polynomial expressions for the logarithmic concentrations of Brilliant Blue, logC_BB (DS: < 406 mg Brilliant Blue per kg soil; WS: < 214 mg Brilliant Blue per kg soil; UM: 114–681 mg Brilliant Blue per kg soil) were determined as follows:

Dry sand (DS):

(1)

Wet sand (WS):

(2)

Undisturbed soil monolith (UM):

(3)

where logC_BB is given in milligrams Brilliant Blue per kilogram soil (assuming logC_BB = 0 as a lower limit). The coefficients of determination are 0.984 (Eq. [1]), 0.997 (Eq. [2]) and 0.927 (Eq. [3]).

The following second-order polynomial expression for the logarithmic concentrations of Azophloxine, logC_Azo, (DS: < 152 mg Azophloxine per kg soil) was determined:

Dry sand (DS):

(4)

with an R² value of 0.962.

Interpretation of the resulting variables and coefficients of Eq. [1] to [4] is not simple because we are using a regression model with mixed terms instead of a more physical model. Nevertheless, there appears to be a certain similarity between Eq. [1], [2], and [3] (Brilliant Blue) in the constant and the high weight of R_s (all having negative values). Equation [4] (Azophloxine) is also dominated by R_s, but in this case with a positive value.

Equation [1] to [4] were then applied to each pixel of the corresponding digitized section images. At a later stage, horizontally and vertically averaged Brilliant Blue and Azophloxine concentrations were calculated for each pixel row and column. The concentrations [mg per kg soil] resulting from image analysis were first converted to volumetric concentrations [mg cm^-3] assuming a depth invariant bulk density (Table 1), and then to relative concentrations [cm^-1] calculated as the ratio of the concentration in the soil [mg cm^-3] to the applied amount of solute per surface area [mg cm^-2]. Relative concentrations allow a direct comparison between different tracers applied at different concentrations and, in addition, they are a conclusive measure of mass recovery, i.e., they represent the ratio of found and applied tracer mass. The mass recovery is based on the relative concentrations integrated over depth.

Fluorescence Imaging
For a detailed description of the multitracing detection device (MDD) we refer to Aeby (1998). The light emitted by a high-power xenon lamp is filtered according to the excitation spectra of the fluorescent dyes. This is accomplished by the use of different sets of optical interference filters. The hot mirror removes most of the infrared part of the light, whereas the dichroic filters preselect the wavelengths in the visible region before the light enters the liquid light guide. The transmission spectrum of the excitation filter specifically chosen for each dye closely matches the dye's excitation spectrum. In the same way, the emission filter is chosen according to the dye's emission spectrum. The proper selection of excitation light and emission filter results in a well-defined fluorescent signal, which is detected by the CCD (charge coupled device) camera. The signal-to-noise ratio can be further improved using correction images for inhomogeneous lighting and background.

	Results

TOP ABSTRACT INTRODUCTION Materials and method Results Discussion and conclusions REFERENCES

 
Packed Sand Columns
The initial soil water content controlled the infiltration in the packed sand columns. In the initially dry column (DS), all applied water infiltrated, i.e., 25 mm during the first and 25 mm during the second irrigation event. In the initially wet column (WS), only the water applied during the first irrigation event actually infiltrated, whereas during the second irrigation period infiltration was negligible. In the latter case, the water ponded on the surface and refroze when the chamber temperature was reset to -5°C at the end of the irrigation. On the WS column, water ponding was observed already during the first irrigation cycle after about 2 h of irrigation. However this water slowly infiltrated. In the DS column, some surface ponding was also noticed, but only during the second irrigation.

The temperature development in the sand columns and the chamber is shown in Fig. 4 . The DS column reached thermal steady-state conditions after 4 d of freezing and WS after 7 d. During both irrigation events, the temperature increased to freezing point. However, it did not exceed 0°C in any of the columns except for some short periods because of the warming effect of infiltrating water. From this, we conclude that the bulk of the soil water remained frozen and controlled the thermal status. After resetting the cold chamber temperature to -5°C at the end of each irrigation cycle, the soil temperatures at depths of 5 and 15 cm (not shown in Fig. 4) stayed close to 0°C for about 12 to 24 h, except in WS after the second irrigation. Here, the temperatures decreased more rapidly indicating that no additional water had infiltrated.

View larger version (35K): [in this window] [in a new window]   Fig. 4 Temperatures (°C) measured in the chamber, in the dry and the wet packed sand column

The two-dimensional concentration maps (Fig. 5) corresponded well with the stained pattern of Brilliant Blue. For DS, however, there are some intensively stained areas, which the calibration assigned to regions of intermediate concentrations. These misinterpretations can be explained as extrapolation errors because no calibration samples were taken in this elevated concentration range.

View larger version (113K): [in this window] [in a new window]   Fig. 5 Brilliant Blue concentration maps determined by image analysis for three profile-sections of the dry sand (DS, top), and wet sand (WS, bottom) at 5 cm (left), 7.5 cm (middle), and 10 cm (right) from the wall. Pixel size is 0.5 by 0.5 mm. The arrows indicate misinterpreted areas that probably had the highest concentrations but were classified as medium concentrations

View larger version (57K): [in this window] [in a new window]   Fig. 6 Vertically (top) and horizontally averaged (bottom) relative Brilliant Blue and Azophloxine concentration distributions of the profiles at 5, 7.5, and 10 cm distance from the wall in the dry sand (DS) and wet sand (WS) columns obtained from the concentration maps (Fig. 5 for Brilliant Blue). Vertical concentration distributions obtained from horizontal sections taken at 0 to 5 cm from the wall are also shown

The Azophloxine concentrations, integrated from the image-analyzed concentration maps, are homogeneously distributed, both vertically and horizontally. This pattern was similar to the Brilliant Blue distribution in WS. The reason for this similarity might be that the total water content of DS prior to the second irrigation was comparable with that of WS at the start of the first irrigation.

Mass recovery of Brilliant Blue and Azophloxine are summarized in Table 2. Mass recovery from the horizontal sections was between 0.68 and 1.28. The small value of 0.14 for the Azophloxine in WS is due to the fact that virtually no infiltration occurred. The values resulting from image analysis are between 0.47 and 1.24 for the two dye tracers and appear to be reasonable estimates. The higher variance for the Brilliant Blue in DS, as compared to WS might be an indication that infiltration in a relatively dry, frozen soil is more heterogeneous than in a relatively wet, frozen soil.

Soil Monolith
Soil and chamber temperature measurements during this second experiment with the undisturbed soil monolith are presented in Fig. 7 . When the soil had reached steady-state thermal conditions at +2°C, the chamber temperature was set to -5°C. A constant temperature profile of -5°C was reached after 10 d.

View larger version (27K): [in this window] [in a new window]   Fig. 7 Temperatures (°C) measured in the chamber and the soil monolith

Figure 3 shows color-normalized images and concentration maps of Brilliant Blue infiltration into the UM sample. At some locations, paraffin penetrated into the macropores and blocked them for the percolating solution. The applied solution infiltrated preferentially along cracks and rotted roots in the direction away from the instrumentation wall. Obviously, this preferential flow was initiated very closely to the surface. A large portion of the solution left the monolith at the back and side walls from where it moved to greater depths or even to the bottom. The bluish stained regions suggest that the Brilliant Blue diffused into the matrix after being preferentially channeled.

Assuming a detection limit for Brilliant Blue of 0.065 mg cm^-3, the two-dimensional concentration maps (Fig. 3) correspond well with the visible pattern of the Brilliant Blue. One exception is the few spots with an apparently high Brilliant Blue concentration, not detected by image analysis probably because of a lack of calibration samples from these relatively dark locations.

The horizontal and vertical relative Brilliant Blue concentration distributions from the concentration maps of the three profiles are shown in Fig. 8 . The mass recovery of Brilliant Blue is 0.46 for the profile at 10 cm, 0.37 at 8 cm, and 0.19 at 6 cm. The horizontal distributions show that the Brilliant Blue is concentrated on the left-hand side with a second peak on the opposite side. Also, the vertical distribution is bimodal. The accumulation between 9 and 13 cm is part of the main plume on the left-hand side, whereas the accumulation between 3 and 8 cm mainly originates from the smaller plume on the right hand side. Below a depth of 15 cm, Brilliant Blue was detected only along one or two preferential flow paths.

View larger version (55K): [in this window] [in a new window]   Fig. 8 Vertically (top) and horizontally averaged (bottom) relative Brilliant Blue concentration distributions of the profiles at 10, 8, and 6 cm from the wall of the soil monolith obtained from the concentration maps of Fig. 3. The vertical concentration distributions of Brilliant Blue and bromide obtained from horizontal sections taken at 6 to 10 cm from the wall are also shown

View larger version (47K): [in this window] [in a new window]   Fig. 9 Concentration map of Brilliant Sulfaflavine determined by fluorescence imaging (left) and vertical relative Brilliant Sulfaflavine concentration distribution in the soil monolith horizontally averaged over the profile-section taken at 10 cm from the wall (right). The vertical concentration distributions of Brilliant Sulfaflavine and bromide obtained from the horizontal sections taken at 6 to 10 cm from the wall are also given

The mass balance of applied water and tracers was not complete for the monolith because of the solution losses at the back and side walls. But it was possible to calculate the mass balance for the two horizontal sections between 0 and 6 cm and between 6 and 10 cm away from the instrumented wall (Table 3). The mean mass recovery of Brilliant Blue determined by image analysis of the three profiles is also listed in Table 3. The water balance supports the visual interpretation of the Brilliant Blue profiles prepared at different distances from the instrumentation wall. Only 21% of the applied water was found in the horizontal sections between 0 and 6 cm, and 54% between 6 and 10 cm. Relative to the recovered water, 83 to 88% of the bromide was found, but only 32 to 49% of Brilliant Blue, and 24 to 29% of Brilliant Sulfaflavine. While the extraction method used recovered most of the dissolved bromide, it was less effective for the dye tracers. The rates for Brilliant Blue determined by extraction and image analysis, which is also based on extraction of the dye from the calibration samples, were at least similar. This suggests that image analysis is a useful tool for producing two-dimensional concentration maps provided that the extraction method is accurate.

	Discussion and conclusions

TOP ABSTRACT INTRODUCTION Materials and method Results Discussion and conclusions REFERENCES

 
These experiments showed that digital image analysis is well suited for visualizing and quantifying solute infiltration into frozen soil. Because of different optical features of unfrozen and frozen soil surfaces, separate calibration samples need to be taken when the profile is only partly frozen. The following points have to be carefully considered: (i) the number of calibration samples from stained, as well as unstained areas, (ii) the method used for extracting the dye from the calibration samples, and (iii) the homogeneous illumination of the photographed profile.

The fluorescence imaging technique was also successfully applied to frozen conditions. Although this technique makes use of a specifically designed device, it has various advantages compared with the digital image analysis: (i) inhomogeneous illumination of the profile can be easily corrected for, (ii) the background reflection does not interfere with the determination of the dye concentration, and (iii) only a small number of calibration samples have to be taken because the solute concentration is linearly related to the pixel values.

The concentration profiles resulting from this experiment confirmed the current hypotheses concerning infiltration into frozen soil: (i) the water content at the time of freezing determined the infiltration capacity of the packed frozen sand, and (ii) in the structured natural soil monolith, infiltration almost exclusively occurred through a network of preferential flow pathways which were air-filled at the time of freezing.

In addition, the method allowed us for the first time to quantify frozen soil infiltration with high spatial resolution and sufficient accuracy. For the calculation of the mean vertical concentration distribution, the number of vertical profiles needed depends on the heterogeneity of the stained flow pathways. In the case of homogeneous infiltration, the distances between the sections can be larger than for an infiltration with preferential pathways. Although the mass recoveries of the dye tracers vary according to their sorption behavior, it was possible to scale them to the mass recoveries of water. One of the main disadvantages of the single-time spatial analysis is the fact that it misses the temporal development of the infiltration. This may be overcome by using several non-interacting tracers applied at different times, so called multiple tracing. Alternatively, using a larger number of experimental columns would produce a sufficient number of profile sections during different infiltration stages.

The choice of dye tracer and analyzing method (digital image analysis vs. fluorescence imaging) certainly influences the results. It can be concluded from the mass recoveries in this experiment that Brilliant Blue is the most suitable of the three tracers for quantifying infiltration in terms of absolute values. This experiment also showed a general disadvantage of column experiments, which was that the boundary effects of the side walls obviously influenced the resulting concentration profiles, although efforts were undertaken to reduce these disturbances as much as possible. Consequently, the next step will be to conduct similar dye tracer experiments in the field. Such experiments have already successfully been carried out during unfrozen conditions, both in arable field soils (Forrer et al., 1999) and forested soils (Bundt et al., 1998).

	ACKNOWLEDGMENTS

 
This study was conducted within the 4th EU framework as a sub-project of the Programme WINTEX: Land-surface-atmosphere interactions in a wintertime boreal forest. Financial support was received from the Swiss Federal Office for Education and Science (BBW, Bern). We would like to acknowledge the contributions, technical advice and help of the following members of the Soil Physics Group, ETH Zurich: Irene Forrer for her most supportive introduction to digital image analysis, Jörg Leuenberger for preparation of the soil monolith, Hanspeter Läser for construction of the laboratory columns, and Hans Wunderli for the electronic installation. Shelagh Green corrected the language.

Received for publication May 4, 1999.

	REFERENCES

TOP ABSTRACT INTRODUCTION Materials and method Results Discussion and conclusions REFERENCES

 

• Aeby P. Quantitative fluorescence imaging of tracer distributions in soil profiles. Switzerland: ETH Zürich. Schlieren, 1998 Ph.D. thesis, No. 12951..
• Aeby P., Forrer J., Steinmeier C., Flühler H. Image analysis for determination of dye tracer concentrations in sand columns. Soil Sci. Soc. Am. J. 1997;61:33-35.[Abstract/Free Full Text]
• Baker, J.M., and E.J.A. Spaans. 1997. Mechanics of meltwater movement in seasonally frozen soil. p. 31–36. In I.K. Iskandar, et al. (ed.) International Symposium on Physics, Chemistry, and Ecology of Seasonally Frozen Soils, Fairbanks, AK. 10–12 June 1997. U.S. Army Cold Regions Research and Engineering Lab., Special Report 97-10. CRREL, Hanover, NH.
• Bundt M., Zimmermann S., Blaser P., Flühler H. Chemische Heterogenitäten auf der Profilebene—Unterschiede zwischen präferentiellen Fließwegen und Matrix. Mitteilungen der Dt. Bodenkundl. Gesellsch. 1998;87:385-388.
• Ewing R.P., Horton R. Discriminating dyes in soil with color image analysis. Soil Sci. Soc. Am. J. 1999;63:18-24.[Abstract/Free Full Text]
• Flerchinger G.N., Saxton K.E. Simultaneous heat and water model of a freezing snow-residue-soil system, I. Theory and development. Trans. ASAE. 1989;32(2):565-571.
• Forrer I. Solute transport in an unsaturated field soil: Visualization and quantification of flow patterns using image analysis. Switzerland: ETH Zürich. Schlieren, 1997 Ph.D. thesis, No. 12476..
• Forrer I., Kasteel R., Flury M., Flühler H. Longitudinal and lateral dispersion in an unsaturated field soil. Water Resour. Res. 1999;35:3049-3060.
• Forrer I., Papritz A., Kasteel R., Flühler H., Luca D. Quantifying dye tracers in soil profiles by image processing. submitted to: Eur. J. Soil Sci. 2000;in press.
• Kane D.L., Stein J. Water movement into seasonally frozen soils. Water Resour. Res. 1983;19:1547-1557.
• Lundin, L.-C., and H. Johnsson. 1990. Modelling infiltration of snow melt water into frozen soils. p. 53–62. In G. Sigurdsson (ed.) Nordisk Hydrologisk konferens, NHP-rapport nr 26, Kalmar, Sweden. 29 July–1 Aug. 1990. Nordic Hydrological Programme, Norrköping, Sweden.
• Schincariol R.A., Herderick E.E., Schwartz F.W. On the application of image analysis to determine concentration distributions in laboratory experiments. J. Contam. Hydrol. 1993;12:197-215.
• Seyfried M.S., Murdock M.D. Use of air permeability to estimate infiltrability of frozen soil. J. Hydrol. 1997;202:95-107.
• Spaans E.J.A., Baker J.M. The soil freezing characteristic: Its measurement and similarity to the soil moisture characteristic. Soil Sci. Soc. Am. J. 1996;60:13-19.[Abstract/Free Full Text]
• Stadler D., Flühler H., Jansson P.-E. Modelling vertical and lateral water flow in frozen and sloped forest soil plots. Cold Reg. Sci. Tech. 1997;26:181-194.
• Stähli M., Jansson P.-E., Lundin L.-C. Preferential flow in a frozen soil — a two-domain model approach. Hydrol. Proc. 1996;10:1305-1316.
• Stähli M., Jansson P.-E., Lundin L.-C. Soil moisture redistribution and infiltration in frozen sandy soils. Water Resour. Res. 1999;35(1):95-103.
• Zhao L., Gray D.M., Male D.H. Numerical analysis of simultaneous heat and mass transfer during infiltration into frozen ground. J. Hydrol. 1997;200:345-363.

This article has been cited by other articles:

				C. J. Kelln, S. L. Barbour, and C. Qualizza Fracture-Dominated Subsurface Flow and Transport in a Sloping Reclamation Cover Vadose Zone J., February 10, 2009; 8(1): 96 - 107. [Abstract] [Full Text] [PDF]

				M. Janssen and B. Lennartz Characterization of Preferential Flow Pathways through Paddy Bunds with Dye Tracer Tests Soil Sci. Soc. Am. J., October 30, 2008; 72(6): 1756 - 1766. [Abstract] [Full Text] [PDF]

				S. E. Oswald, M. Menon, A. Carminati, P. Vontobel, E. Lehmann, and R. Schulin Quantitative Imaging of Infiltration, Root Growth, and Root Water Uptake via Neutron Radiography Vadose Zone J., August 13, 2008; 7(3): 1035 - 1047. [Abstract] [Full Text] [PDF]

				M. Persson Accurate Dye Tracer Concentration Estimations Using Image Analysis Soil Sci. Soc. Am. J., June 2, 2005; 69(4): 967 - 975. [Abstract] [Full Text] [PDF]

				A. P. Mortensen, K. H. Jensen, B. Nilsson, and R. K. Juhler Multiple Tracing Experiments in Unsaturated Fractured Clayey Till Vadose Zone J., May 1, 2004; 3(2): 634 - 644. [Abstract] [Full Text] [PDF]

				A. Samouelian, I. Cousin, G. Richard, A. Tabbagh, and A. Bruand Electrical Resistivity Imaging for Detecting Soil Cracking at the Centimetric Scale Soil Sci. Soc. Am. J., September 1, 2003; 67(5): 1319 - 1326. [Abstract] [Full Text] [PDF]

				J. Vanderborght, P. Gahwiller, H. Wydler, U. Schultze, and H. Fluhler Imaging Fluorescent Dye Concentrations on Soil Surfaces: Uncertainty of Concentration Estimates Soil Sci. Soc. Am. J., May 1, 2002; 66(3): 760 - 773. [Abstract] [Full Text] [PDF]

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Web of Science (8) Citing Articles via Google Scholar Google Scholar Articles by Stadler, D. Articles by Flühler, H. Search for Related Content PubMed Articles by Stadler, D. Articles by Flühler, H. GeoRef GeoRef Citation Agricola Articles by Stadler, D. Articles by Flühler, H.