Available via license: CC BY 4.0
Content may be subject to copyright.
Vadose Zone Journal | Advancing Critical Zone Science
TopCap: A Tool to Quantify Soil Surface
Topology and Subsurface Structure
Amin Garbout, Craig J. Sturrock, Elena Armenise, Sujung
Ahn, Robert W. Simmons, Stefan Doerr, Karl Ritz,
and Sacha J. Mooney*
The sur face of a material such as soil, as characterized by its topology and
roughness, typically has a profound effect on its functional behavior. While
nondestructive imaging techniques such as X-ray computed tomography
(CT) have been used extensively in recent years to characterize the inter-
nal architecture of soil, less attention has been paid to the morphology
of the soil surface, possibly because other techniques such as scanning
electron microscopy and atomic force microscopy are viewed as more
appropriate. However, X-ray CT exploration of the surface of a soil also
permits analyses immediately below its surface and beyond into the sam-
ple, contingent on its thickness. This provides important information such
as how a connected structure might permit solute infiltration or gaseous
diffusion through the surface and beyond into the subsurface matrix. A
previous limitation to this approach had been the inability to segment and
quantify the actual three-dimensional structural complexity at the sur face,
rather than a predefined geometrically simplistic volume immediately
below it. To overcome this, we formulated TopCap, a novel algorithm that
operates with ImageJ as a plugin and automatically captures the actual
three-dimensional surface morphology, segments the pore structure within
the acquired volume, and provides a series of incisive morphological mea-
surements of the associated porous architecture. TopCap provides rapid,
automated analysis of the immediate surface of materials and beyond,
and while developed in the context of soil, is applicable to any three-
dimensional image volume.
Abbreviations: 2D, two-dimensional; 3D, three-dimensional; CT, computed tomogra-
phy; GUI, graphical user inter face; ROI, region of interest; SIOX, Simple Interactive Ob-
ject Extraction.
The texture and roughness of a surface are important factors controlling
the functional behavior of any material, such as its mechanical properties (Pyka et al.,
2013; Mohamad et al., 2014), frictional behavior (Sahin et al., 2007), or fluid dynamics
across the surface (Taylor et al., 2006). This is especially the case for a biomaterial such as
soil, where the outer surface typically acts as a dynamic interface between the above- and
belowground compartments. The surface structure of soil has profound functional con-
sequences via its control of the movement and flow of gases, liquids, and solutes and can
impact the emergence of germinating seedlings. The complexity of soil structure across a
range of functional scales and depths is well known (Haygarth and Ritz, 2009); however,
the physical structure of the immediate surface has remained remarkably understudied.
The impact of rainfall on the soil surface can lead to a reduction in porosity and the for-
mation of a soil crust or seal through particle detachment and rearrangement (Assouline
and Mualem, 1997; Fohrer et al., 1999). An alternative source of crust development is
microbial activity and subsequent enmeshment of materials at the soil surface, forming
bio-crusts, which have been previously well researched under both arid and semiarid (e.g.,
Chamizo et al., 2012) and temperate conditions (Jeffery et al., 2007). In both cases, it is
probable that a surface crust leads to a reduction in hydraulic conductivity and an increase
in surface runoff and f looding, in addition to a poorer environment for plant emergence
and establishment (Sandin et al., 2017).
Core Ideas
•A new ImageJ plugin, TopCap,
automatically captures soil surface
complexity from CT images.
•TopCap can quantify the immediate
subsurface structure, highlighting soil
crusting and sealing.
•Crust thickness varies under different
soil textures following similar rainfall.
A. Garbout, Imaging and Analysis
Centre (IAC), Natu ral Histor y Museum,
London SW7 5BD, UK; A. Garbout, C.J.
Sturrock, K. Ritz, and S.J. Mooney, Divi -
sion of Agriculture & Environmental
Sciences, School of Biosciences, Univ.
of Nottingham, Loughborough LE12
5RD, UK; E. A rmenise, R.W. Simmons,
and K. Ritz, Sch ool of Water Energy a nd
Environment, Cranfield Univ., Cranfield
MK43 0AL, UK; S. Ahn, Korea Forest Ser-
vice, Seoul, Korea; S. Doerr, College
of Science, Swansea Univ., Swansea
SA2 8PP, UK. *Cor responding autho r
(sacha.mooney@nottingham.ac.uk).
Received 2 May 2017.
Accepted 2 Oct. 2017.
Supplemental material online.
Citation: Garbout, A., C.J. Sturrock,
E. Armenise, S. Ahn, R.W. Simmons,
S. Doer r, K. Ritz, and S.J. Mooney.
2018. TopCap: A tool to quantify soil
surface topology and subsurface
structure. Vadose Zone J. 17:170091.
doi:10.2136/vzj2017.05.0091
Special Section: Noninvasive
Imaging of Processes in
Natural Porous Media
© Soil Science Society of Amer ica.
This i s an open access a rti cle di str ibute d
under the CC BY-NC- ND license
(http://creativecommons.org/licenses/
by-nc-nd/4.0/).
Published online March 29, 2018
VZJ | Advancing Critical Zone Science p. 2 of 10
Within the material sciences and engineering sector, surface
metrology plays a key role in quality control for manufacturing,
especially with the increasing development of advanced compo-
nents. The inspection and assessment of surface roughness can
be performed by many different techniques (e.g., interferometry,
microscopy, contact or laser profilometry) that have contrasting
inherent capabilities but also limitations concerning the size of the
samples (typically only at the millimeter scale) or the interaction
of the instrument with the surface (Gao et al., 2008). For example,
laser profilometry can quickly capture the surface of a material,
but inaccuracies can occur due to atypical points captured when
scanning due to the optical properties of the material (Rousseau
et al., 2012). Furthermore, a surface with cracks or concavities can
be difficult to measure due to limitations in access by the instru-
ment probe.
Continuing developments in three-dimensional (3D) imaging and
particularly tomographic techniques enable the internal visualiza-
tion of materials at increasingly high resolutions (nanometers to
micrometers). The nondestructive nature of X-ray micro computed
tomography (CT) has been applied to the analysis of the inner struc-
ture of a wide range of biomaterials including plants (Garbout et al.,
2011; Pajor et al., 2013), rocks (Zakirov et al., 2016), soils (Helliwell
et al., 2013), and animals (Badea et al., 2008) and has also been used
in many areas of industr y for the internal inspection of components
(Bull et al., 2013). X-ray CT has a lso been applied to the 3D measure-
ment of surfaces of additive manufactured porous alloy materials
(Kerckhofs et al., 2012; Pyka et al., 2014). Depending on the dimen-
sions of the surface roughness, CT can be applied accurately and in
a robust manner for surface roughness quantification of 3D porous
materials. Chinga-Carrasco et al. (2008) demonstrated the suit-
ability of the technique as a method to assess coatings and surface
properties of paper, with a performance comparable to atomic force
microscopy or laser profilometry.
Unlike techniques limited to surface examination, there is great
potential, especially in the biomaterial sciences, in the application
of CT imaging to contribute to our understanding of the relation-
ships between the surface topology and the functional properties
of a material. One key goal would be the ability to quantify the
structure at the immediate surface of a soil and link this with the
behavior of the subsurface matrix. However, accurate quantifica-
tion of soil surfaces with irregular, non-planar topologies have
represented a major impediment to adoption of the technology
because the uniformly sized region of interest (ROI) approach,
commonly used in image analysis operations (Fig. 1), does not
capture the complexity of the soil surface structure.
Alternative approaches to overcome this limitation include manual
two-dimensional measurements of the structure near the surface
using the intensity values of CT image pixels along a specified tran-
sect (i.e., mapping changes in the X-ray attenuation of the material
in a cross-section image). Riley et al. (2014) utilized a similar
approach to visualize eggshell surface morphology. However, two-
dimensional (2D) measurements of 3D media give an incomplete
description of the true surface topography and then subsurface
pore structure. Removing the non-planar surface from the analy-
sis by “cropping” a new ROI in an image has been typically used
as a solution to deal with the complex 3D
surface geometry in many previous stud-
ies (e.g., Helliwell et al., 2014; Tracy et al.,
2015). In this approach, the image datasets
are trimmed to remove regions based on the
roughness, edge effects, or complexity. A key
issue is that these regions are, in many cases,
crucial interfaces or pathways, and removing
them leads to a loss of important data that
explain how a surface and internal structure
are interconnected.
To overcome this, some approaches have
been developed based on 3D object extrac
-
tion tools such as Simple Interactive Object
Extraction (SIOX), a plugin developed
and implemented under FIJI, a freeware
image processing tool with a graphical user
interface (GUI) (http://imagej.net/SIOX:_
Simple_Interactive_Object_Extraction).
SIOX permits simple extraction of fore-
ground objects from 2D images. This plugin
is semiautomated and requires a modest
interactive effort on the part of the user to
Fig. 1. Example of a thresholded soil micro-CT image: A typical region of interest (ROI) cannot
capture the complexity of the soil surface even when located very near the surface.
VZJ | Advancing Critical Zone Science p. 3 of 10
avoid misclassification of the material surface. However, it cannot
currently be applied to a stack of images, as is typically the require-
ment for CT imagery, hence the procedure needs to be repeated
for each slice or image, which is not preferable when the number
of images per CT scan typically exceeds 2000.
In metrology, different image filtration techniques have been
used that aim to separate the desirable and undesirable fea-
tures in a surface topography dataset (De Chiffre et al., 2000).
Application of mathematical morphology theory supported
the emergence of morphological filters and so-called “envelope”
filtering systems (also known as E-systems) (Srinivasan, 1998;
Scott, 2000). The E-system approach simulates the process
of rolling a ball with a selected radius over the surface (von
Weingraber, 1954). The envelope filter is relative to the geo-
metrical properties of the surface and thus gives better results
on the functional prediction of surfaces. Several algorithms
have been developed using these approaches (Shunmugam
and Radhakrishnan, 1974; Lou et al., 2011), each with some
drawbacks. Primarily, these methods are either time consum-
ing, especially for large datasets and structuring elements, or
hard to extend to areal data. Furthermore, the maximum ball
radius is typically limited in practice due to the associated huge
computational requirement, while for many real applications,
ball or disk radii much larger than the signal length are required
(Jiang et al., 2011). Consequently, there is a need to develop
tools able to extract surfaces automatically without significant
user interaction to accurately segment the surface of a material
from the background via a robust and repeatable methodology.
Here we introduce and describe TopCap (Topography Capture),
a novel method formulated as a plugin tool for ImageJ, designed
primarily to quantify the structure of the immediate, heteroge-
neous surface of any material. We developed the method in the
context of soil, generally considered to be the most biologically
active and structurally complex component of the terrestrial bio-
sphere (Ritz and Young, 2011). The plugin was designed primarily
to accurately quantif y the heterogeneity of pore morphology at the
near and immediate surface of a material, accounting for the exact
geometrical complexity as revealed by imager y of the actual surface
rather than a predefined rectangular ROI. Using the concept of
morphological operations, TopCap automatically:
ʶ
detects and captures the material surface by iterative morpho-
logica l operations of dilatation and erosion i n three dimensions;
ʶ
defines sections parallel to the surface and measures the micro-
structural characteristics;
ʶ
collates the defined sections and pore geometry within each
separated section; and
ʶcalculates the total pore volume, porosity, and other pore mor-
phological measurements within each of the defined sections.
6Materials and Methods
X-ray Computed Tomography
The soil samples used in this work were scanned using a Phoenix
Nanotom 180NF system (GE Sensing & Inspection Technologies).
The scanner consisted of a 180-kV nanofocus X-ray tube fitted
with a diamond transmission target and a five-megapixel flat panel
detector (Hamamatsu Photonics). A maximum X-ray energy of
130 kV and 100 mA current was used to scan each soil core. A total
of 1440 projection images were acquired throughout a 360° rota-
tion. Typical scan time was around 50 min per core. Resolution
varied between scans and is stated where relevant. Reconstruction
of the projection images was used to produce 3D volumetric
datasets using the software datos|rec (GE Sensing & Inspection
Technologies). The reconstructed CT volumes were visualized
and resized in VGStudio Max 2.1 prior to the assessment of the
soil surfaces.
Mathematical Description
Detection of the Boundaries of the Surface
To extract the boundary pixels of a foreground structure such as
the extreme soil surface (Fig. 2), a morphological filtering approach
was adopted (Lou et al., 2013). Morphological filtering consists
of a defined series of operators that transform an image (I) by
probing it with a predefined shape or structuring element (H).
In TopCap, a 3 by 3 structuring element was used. Two of the
most fundamental operations within a morphological filtering
approach are dilation and erosion (Burger and Burge, 2016), and
essentially all other morphological operations are generally built
from a combination of these two. Conventionally, A Å B corre-
sponds to the dilation of A by B and A $ B to the erosion of A by
B. In binar y images, dilation is an operation that increases the size
of foreground objects. It is defined as
( )
{ }
| for some and I H pq pI qHź + Î Î
[1]
where p is an element (pixel) from the image space I to which is
added an element q from H (the structural element). One of the basic
applications of dilation is to bridge gaps and connect objects. In con-
trast, erosion is an operation that increases the size of background
objects (and shrinks the foreground objects) in binary images:
( )
{ }
2
| , for every I H p pq I qHº Î +Π΢!
[2]
The consequence of Eq. [2] is that the background regions grow in
size and foreground features tend to disconnect or further sepa-
rate. Further erosion results in more growth of the background or
shrinking of the foreground.
TopCap was based on the use of both erosion and dilatation fil-
ters. Due to their semi-duality, dilation and erosion are often used
VZJ | Advancing Critical Zone Science p. 4 of 10
together in composite operations referred to as opening and closing.
TopCap offers the possibility of setting the desired sequence of
dilations and erosions.
We composed our filter by repeated application of the structuring
element H. This removes holes and fissures in the foreground
structure that are smaller than H multiplied by n (the number of
iterations of H). This permits the extraction of the boundary of
the foreground object surface. The background pixels enclosed in
the foreground structure are removed after the dilation sequence
with a particle sizing filter (Abràmoff et al., 2004). This method
is more accurate than a convex hull filter to obtain the boundary
of an object surface. The detection of surfaces permits labeling
the background image black. The foreground pixels are labeled
white, as well as the background pixels included in the fore-
ground image.
Interval Sections Parallel to the Surface
The extraction of the boundary of the soil surface is achieved by
the application of successive dilation operations (as explained
above) followed by filtration of the pores in the soil and then
iterative erosion. Sections parallel to the surface are obtained by
iterative erosions of the foreground image. The numbers of ero-
sions are specified by the user in the GUI (Fig. 2). The obtained
masks for each section, and the pores within each section are then
saved in a specified folder (Fig. 2d and 2e), allowing examination
of the pore space at specific separate soil depths.
Theoretical Validation
To validate TopCap, our first approach was to generate a series of
artificial images, which comprised 128 stacked images of 10 by 500
by 10 pixels representing a single white rectangle (with a width of
10 pixels) on a black background (Fig. 3). The position (x, y) of each
rectangle in each of the 128 stacked images was set as
( )
sin 10
500 ; with 0 128
11
u
Lu
éù
+
êú
= <<
êú
ëû
[3]
0 and 500x yL= =-
where u is the slice number and L is the distance from the bottom
of the image to the rectangle.
A hole (in black) with a rectangular shape representing the pore
space was situated 20 pixels from the top of the white rectangle.
The 128-image stack was then combined to create one stack of 10
slices of 600 by 1280 pixels. Three separate stacks of images—W1,
W2, and W3—were produced, each with 10-pixel holes but with
different widths, i.e., one, two, and six pixels, respectively, to simu-
late contrasting surface irregularities. TopCap was applied to all
the images and the theoretical measurements compared with those
obtained from our plugin (Table 1). For comparative purposes, the
same calculations were undertaken on the artificia l surfaces using
SIOX. The theoretical surface area (S) of the artificial images was
determined by measuring the 2D surface profile across 10 image
Fig. 2. Presentation of TopCap: different outputs after selecting options from the graphical user interface (GUI) (on the left side screenshot of the GUI):
(a) original grayscale CT image of soil; (b) segmented image by thresholding using the Otsu algorithm; (c) mask obtained with a closure coefficient of 10;
(d) mask set at 100 voxels depth surface obtained by setting DEPTH = 100; (e) segmented pores within the mask capture in g; (f ) two-dimensional soil
surface at the surface obtained by setting DEPTH = 1; and (g) gaps in the soil surface associated with surface-connected pores using INTERVALS = 1.
VZJ | Advancing Critical Zone Science p. 5 of 10
slices. The volume of the section (M) located from the surface to
the 30-voxel depth was calculated as M = 30S. The output of the
plugin and the predicted properties were very similar, even when
the holes were one-pixel width.
6Results and Discussion
Comparison of TopCap
with Other Approaches
The output from TopCap was compared with two other methods
routinely applied to similar studies, namely selection of a rectan-
gular ROI and SIOX. Each tool was used to identify or segment
the soil surface and calculate the porosity associated with differ-
ent depth intervals from the surface to a maximum of 6.7 mm or
50 voxels (Fig. 4). This particular soil surface was used due to its
complexity, with a significant number of visible surface cracks and
numerous biopores. Fifty image slices with a spatial resolution of
134 mm were used for the test.
After segmentation, the ROI method, as shown in Fig. 1, con-
sisted of selecting a first rectangular ROI as near as possible
to the surface and measuring the porosity within the selec-
tion. Subsequent rectangles were then selected below the first
selection to assess porosity with depth. The rectangular selec
-
tion does not offer the possibility of capturing the soil surface
without selecting background pixels and overestimating the
porosity nearest the surface; it also needs to be repeated for
each slice. TopCap was tested on the same soil, following the
method described in the Technical Manual (see below), where
the surface was captured automatically, as was the porosity in
three dimensions, for the different intervals or depths within
seconds (Fig. 4d and 4h). Minimal user input was required, and
the whole process took approximately 10 min (for comparison,
Fig. 3. Validation images with holes of (a) one-, (b) two-, and (c) six-pixel width used to test TopCap; (d) porosity measured within depth with different
intervals, showing three peaks corresponding to the distance of pores to the surface; and (e) image with holes of six-pixel width showing the surface in
white and detected pores in different colors according to their distance from the surface.
Table 1. Comparison of measurements by TopCap and Simple Interactive Object Extraction (SIOX): three samples with rectangular holes W1, W2,
and W6 with one-, two-, and six-pixel widths and volumes V1, V2, and V3, respectively, and selecting Interval = 1 and Depth = 40.
Sample
Calculated para meters Measured p arameters
Total pore volume from 0- to 3 0-voxel
depth Surf ace area
Pore total volume f rom 0- to 30-voxel depth
Surface areaTop C a p SIOX
pixel3pixel2————————— p i x e l 3 ————————— pixel2
W1 Pv1 = V1 ´ 128 = 12,800 S1 = 5080 ´ 10 = 50,8 00 12,800 12,548 49,530
W2 Pv2 = V2 ´ 128 = 25,60 0 S2 = 50,80 0 25,6 00 25,091 49,530
W6 Pv6 = V6 ´ 128 = 76,800 S6 = 50,800 76,800 75,262 49,530
VZJ | Advancing Critical Zone Science p. 6 of 10
assessment of the artificial structure took about 2 min) using
the computer configuration described below. SIOX was also able
to automatically extract the soil surface, however, and impor-
tantly, only in two dimensions, and thus it had to be repeated
for each image or slice unless a macro can be created to auto-
mate the process. Also with SIOX, there is an additional step
that requires manual input to remove the black pixels within
the mask (Fig. 4f and 4g), whereas this step is automatic with
TopCap. The operation via SIOX took approximately 1 h (for
comparison, assessment of the artificial structure took about10
min). Qualitative analysis shows that the final surface extracted
by SIOX was not as accurate as TopCap, as it was unable to
handle the irregularity of the soil surface adequately (Fig. 4d
and 4g). This has important implications for the subsequent
porosity analysis, which visually appears to be overestimated by
SIOX compared with TopCap (Fig. 4h and 4i). In addition, the
opportunity to quantify porosity at specified depth intervals
is not available automatically within SIOX. For comparative
purposes, we undertook a transect-style analysis of porosity at
approximately 130-mm depth intervals to a depth of 0.67 mm
for each method (Fig. 5), where the overestimation of the poros-
ity within the crust area seen in Fig. 4 is clearly demonstrated.
For comparative purposes, a similar analysis to that shown in Fig.
5 was undertaken on the artificial structures and presented in
Supplemental Fig. 1 and Table 1, showing greater complementar-
ity than with the soil surfaces examined.
Fig. 4. Evaluation of soil surface detection and measurement of porosity by TopCap, the region of interest (ROI) approach, and Simple Interactive
Object Extraction (SIOX): (a,b,c) the original grayscale images are converted to a binary mask by (d) TopCap, (e) the ROI method, and (f ) SIOX.
Note that the ROI method does not capture the complexity of the soil surface; successive ROI images are needed to measure porosity at different
depths; (g ) the SIOX approach requires a further step to manually remove the black pixels within the soil before obtaining the final surface mask as pre-
viously shown by TopCap in (d). The final extracted porosity of the immediate soil surface is shown for (h) TopCap and (i) SIOX. One voxel = 74 mm.
Fig. 5. Differences in quantification of
porosity from the surface to the 0.67-
mm depth by TopCap, the region of
interest approach (ROI), and Simple
Interactive Object Extraction (SIOX).
VZJ | Advancing Critical Zone Science p. 7 of 10
Demonstration of TopCap
on Different Soil Types
TopCap was tested on two texturally contrasting, crust-susceptible
soils. A silty clay loam from Butterwick, Lincolnshire (52°59¢12² N,
0°3¢33² E) classified as the Wisbech series (Soil Survey of England
and Wales) and a sandy loam from the Eardiston series sampled
from Coughton, Ross-on-Wye, Herefordshire (51°53¢43² N,
2°33¢50² W). The qualitative and quantitative results (Fig. 6 and 7,
respectively) obtained from TopCap revealed a clear soil crust at
the surface and an initial decrease in porosity at the immediate
surface corresponding to the variable surface roughness between
the soil types. Crucially, this demonstrated the presence of a physi-
cal crust at the immediate surface not measurable by alternative
approaches. The soil crust was clearly thicker in the silty clay loam
soil, and a visual inspection of the CT images (Fig. 6a, 6b, and
6c) showed the considerable difference in porosity between the
Fig. 6. Test of TopCap on a silty clay loam (left) and a sandy loam (right): (a,e) X-ray CT image of the soil (the pore space is in black); (b,f ) thresholding
using the Otsu method to separate the pores from the soil matrix to obtain a binary image, with white being the soil matrix (the pore space is in black);
(c,g) two-dimensional view showing in green the soil surface detected using TopCap with closure coefficient = 10, depth = 100, and interval = 10, and
pores within the setup depth in white; and (d,h) three-dimensional representations of the soil surface.
VZJ | Advancing Critical Zone Science p. 8 of 10
near-surface and below-surface material, as low as 4% of the total
volume (Fig. 7). The sandy loam soil had a thinner crust with a
greater porosity at the immediate surface compared with the silty
clay loam soil (Fig. 6e, 6f, and 6g), with the lowest porosity ?9%
of the total volume (Fig. 7). Further morphological data shown in
Table 2 illustrates that the total pore volume and surface area were
lower in the silty clay loam, highlighting development of the crust,
whereas the surface area index (the surface area of the measured
surface compared with that of a perfectly f lat and smooth surface)
was smaller for the sandy soil loam than for the silty clay loam, sug-
gesting greater surface roughness for the sandy soil. These support
results from a parallel experiment on the same soil types, sampled
at the same time, where crusts formed more readily in the silty
clay loam soils when subjected to variations in rainfall intensity,
leading to significant reductions in unsaturated hydraulic con-
ductivity compared with the sandy loam soil (Armenise et al.,
2018; Supplemental Fig. S1). A limitation here is that the porosity
derived by imagery cannot be validated by physical measurements
due to the high resolution and small volumes considered. However,
this approach will ultimately support efforts to improve our under-
standing of the physical processes affected by surface crusts and
permits the characterization of the dynamics of soil crust devel-
opment through repeated CT scanning experiments. While the
differences between the two soils explored here are considerable,
it is worth noting that these soil surfaces were generated under
experimental conditions on sieved soils repacked into columns at
the same bulk density. One would expect the soil surfaces under
field conditions to vary more than in the example we have provided
here. However, we have confirmed that TopCap is applicable to all
forms of soil surface by testing it on field-structured soils such as
a sandy loam soil that had been tilled 6 mo before sampling and
where earthworm activity was observed (Supplemental Fig. S3).
Advantages and Limitations
TopCap permits exploration of both 2D and 3D data with mini-
mal user input. There is f lexibility to vary the ball radius, which
enhances the applicability of the morphological filtering, allowing
it to be used across a range of nonuniform surfaces. By developing
TopCap as an ImageJ macro, the processing of a high number of
samples or image slices (?1000–2000 per scan) in a fully auto-
mated way with minimal user input is possible. As an example,
one image stack of 400 images in a 400- by 130-pixel array was
processed in 100 s (via a personal computer with the following
specifications: processor Intel Xeon CPU E5-2630, with 128 GB
of RAM and a Quadro K5000 graphics card).
Despite the possibility offered to increase the radius of the ball by
adjusting the coefficient of closure to filter the surface and extract
Fig. 7. Pores volumes measured for 100-voxel depth with a binning of 10 voxels for two soil samples. Plots are from the results obtained after selecting
the options DEPTH = 100 and INTERVALS = 10: (A) silty clay loam soil surface; (B) sandy loam soil surface.
Table 2. Example of TopCap morphological measurements for the silty clay loam (Soil 1) and sandy loam (Soil 2) soils. The mask volume and the total
pore volume were calculated for a 100-voxel depth. For comparative purposes, the physically derived total porosity of both soils packed to 1.2 g cm−3
was 0.55 mm3 mm−3.
Soil sample
Measured p arameters
Total pore volume M ask volume Porosit y Projecte d surface Sur face area Surfa ce of cavities Cavities rat io SAI†
——————— m m 3 ——————— mm3 mm−3 ———————————— m m 2 ———————————— mm2 mm−2
Soil 1 240 2338 0.103 352 1007 138 0.138 0.349
Soil 2 326 2455 0.133 352 1230 472 0.383 0.286
† Surface area index.
VZJ | Advancing Critical Zone Science p. 9 of 10
it, caution must be used to minimize user bias because choosing a
high coefficient will reduce the precision of the surface detection
and thus the measure of roughness. Furthermore, in cases where
the images have a low contrast, the Otsu thresholding algorithm
as used for the soil samples here may be suboptimal, so a further
preprocessing step may be invoked before using the plugin. We
would recommend that TopCap users explore other thresholding
algorithms available in ImageJ in these circumstances. The selection
of the closure coefficient is important because it conditions how the
plugin will accurately capture the surface. The coefficient should
be selected by measuring the diameter of the largest gap or pore at
the surface on a 2D slice image, and the optimal coefficient should
not exceed more than half of the measured distance. A high closure
coefficient will tend to overestimate the porosity near the surface
and smooth the surface, losing details about the soil surface rough-
ness (Supplemental Fig. S4). The porosity for the first 10 voxels for
a closure coefficient of 16 was 48%, dropping to 40% in the interval
10 to 20 voxels below the surface. This can be explained by the fact
that background pixels were included in the pore space. However, a
low coefficient will not allow detection of the surface because gaps
and pores at the surface would not be closed; therefore, the coef-
ficient should be chosen carefully and with parsimony. For other
user-defined parameters such as DEPTH and INTERVALS, it is
not necessary to consider minimizing user bias.
6Conclusions
We have developed TopCap, a new algorithm to automatically cap-
ture, segment, and measure the 3D morphological properties of the
immediate soil surface, providing several benefits over preexisting
methods for soil surface segmentation for a range of different soil
types. The data obtained have the potential to provide an unprec-
edented insight into the biophysical properties and functioning of
the soil surface. The surface of a soil remains a crucial but often
ignored interface, a lthough it is usually the initiation site of impor-
tant processes concerning the transport and exchange of gases and
liquids. Tools such as TopCap offer the potential to examine the
mechanisms behind such behavior in ways generally not considered
possible until now. In addition, TopCap could be applied to a wider
range of materials where the complexity of the 3D surface and its
immediate below-surface porosity has implications for behavior
and function, such as plant leaves, where the position of the sto-
mata and guard cells in relation to the overall lower epidermis are
important for gaseous transport.
Supplemental Material
Supplement 1 contains four supplemental figures. Supplemental Fig. S1
shows a comparison of the output from (A) SIOX and (B) TopCap on
porosity for three depth intervals (0–10, 10–20, and 20–30) and three
pixel hole sizes (1, 2, and 6); Fig. S2 shows the effect of rainfall duration
and soil type on the unsaturated hydraulic conductivity (K
un
); Fig. S3
shows an example of the output from TopCap from an und isturbed field
soil; and Fig. S4 shows the effect of different closure coefficients (8–16)
on the soil porosity as a function of depth. Supplement 2 contains the
TopCap technical manual, and Supplement 3 contains the source code
for the TopCap Plugin.
Acknowledgments
This work was supported by UK Biotechnology and Biological Sciences Research Council
Grants (BB/J006092/1 and BB/J006793/1), both of which were also partly supported by the
UK Department of Environment, Food and Rural Affairs under the Government Partnership
Award scheme. Sacha J. Mooney and Craig J. Sturrock are supported by the ERC Futureroots
project.
References
Abràmoff, M.D., P.J. Magalhães, and S.J. Ram. 2004. Image processing
with ImageJ. Biophotonics Int. 11:36–41.
Armenise, E., R.W. Simmons, S. Ahn, A. Garbout, S.H. Doerr, S.J. Mooney,
et al. 2018. Soil seal development under simulated rainfall: Structural,
physical and hydrological dynamics. J. Hydrol. 556:211–219.
doi:10.1016/j.jhydrol.2017.10.073
Assouline, S., and Y. Mualem. 1997. Modeling the dynamics of seal
formation and its effect on infiltration as related to soil and rainfall
characteristics. Water Resour. Res. 33:1527–1536.
Badea, C.T., M. Drangova, D.W. Holdsworth, and G.A. Johnson.
2008. In vivo small-animal imaging using micro-CT and digital
subtraction angiography. Phys. Med. Biol. 53:R319–R350.
doi:10.1088/0031-9155/53/19/R01
Bull, D.J., L. Helfen, I. Sinclair, S. Spearing, and T. Baumbach. 2013. A
comparison of multi-scale 3D X-ray tomographic inspection techniques
for assessing carbon fibre composite impact damage. Compos. Sci.
Technol. 75:55–61. doi:10.1016/j.compscitech.2012.12.006
Burger, W., and M.J. Burge. 2016. Digital image processing: An algorithmic
introduction using Java. 2nd ed. Springer, London. doi:10.1007/978-1-
4471-6684-9
Chamizo, S., Y. Canton, I. Miralles, and F. Domingo. 2012. Biological
soil crust development affects physicochemical characteristics
of soil surface in semiarid ecosystems. Soil Biol. Biochem. 49:96–105.
doi:10.1016/j.soilbio.2012.02.017
Chinga-Carrasco, G., H. Kauko, M. Myllys, J. Timonen, B. Wang, M. Zhou,
and J.O. Fossum. 2008. New advances in the 3D characterization
of mineral coating layers on paper. J. Microsc. 232:212–224.
doi:10.1111/j.1365-2818.2008.02092.x
De Chiffre, L., P. Lonardo, H. Trumpold, D.A. Lucca, G. Goch, C.A. Brown,
et al. 2000. Quantitative characterisation of surface texture. CIRP Ann.
49:635–642, 644–652. doi:10.1016/S0007-8506(07)63458-1
Fohrer, N., J. Berkenhagen, J.M. Hecker, and A. Rudolph. 1999. Changing
soil and surface conditions during rainfall: Single rainstorm/subsequent
rainstorms. Catena 37:355–375. doi:10.1016/S0341-8162(99)00026-0
Gao, G., R.K. Leach, J. Petzing, and J.M. Coupland. 2008.
Surface measurement errors using commercial scanning
white light interferometers. Meas. Sci. Technol. 19:015303.
doi:10.1088/0957-0233/19/1/015303
Garbout, A., L.J. Munkholm, S.B. Hansen, B.M. Petersen, O.L. Munk, and R.
Pajor. 2011. The use of PET/CT scanning technique for 3D visualization
and quantification of real-time soil/plant interactions. Plant Soil
352:113–127. doi:10.1007/s11104-011-0983-8
Haygarth, P.M., and K. Ritz. 2009. The future of soils and land use in the UK:
Soil systems for the provision of land-based ecosystem services. Land
Use Policy 26:S187–S197. doi:10.1016/j.landusepol.2009.09.016
Helliwell, J.R., A.J. Miller, W.R. Whalley, S.J. Mooney, and C.J. Sturrock.
2014. Quantifying the impact of microbes on soil structural
development and behaviour in wet soils. Soil Biol. Biochem. 74:138–
147. doi:10.1016/j.soilbio.2014.03.009
Helliwell, J.R., C.J. Sturrock, K.M. Grayling, S.R. Tracy, R.J. Flavel, I.M.
Young, et al. 2013. Applications of X-ray computed tomography for
examining biophysical interactions and structural development in soil
systems: A review. Eur. J. Soil Sci. 64:279–297. doi:10.1111/ejss.12028
Jeffery, S., J.A. Harris, R.J. Rickson, and K. Ritz. 2007. Microbial community
phenotypic profiles change markedly with depth within the first
centimetre of the arable soil surface. Soil Biol. Biochem. 39:1226–1229.
doi:10.1016/j.soilbio.2006.12.023
Jiang, X., S. Lou, and P.J. Scott. 2011. Morphological method for surface
metrology and dimensional metrology based on the alpha shape.
Meas. Sci. Technol. 23:015003. doi:10.1088/0957-0233/23/1/015003
Kerckhofs, G., G. Pyka, M. Moesen, S.V. Bael, J. Schrooten, and M. Wevers.
VZJ | Advancing Critical Zone Science p. 10 of 10
2012. High-resolution microfocus X-ray computed tomography for 3D
surface roughness measurements of additive manufactured porous
materials. Adv. Eng. Mater. 15:153–158. doi:10.1002/adem.201200156
Lou, S., X. Jiang, and P.J. Scott. 2011. Fast algorithm for morphological filters.
J. Phys. Conf. Ser. 311:012001. doi:10.1088/1742-6596/311/1/012001
Lou, S., X. Jiang, and P.J. Scott. 2013. Geometric computation theory
for morphological filtering on freeform surfaces. Proc. R. Soc. A
469:20130150. doi:10.1098/rspa.2013.0150
Mohamad, M., H.F.A. Marzuki, E.A.E. Ubaidillah, M.F.Z. Abidin, S. Omar,
and I.M. Rozi. 2014. Effect of surface roughness on mechanical
properties of aluminium–carbon laminates composites. Adv. Mat. Res.
879:51–57.
Pajor, R., A. Fleming, C.P. Osborne, S.A. Rolfe, C.J. Sturrock, and S.J.
Mooney. 2013. Seeing space: Visualization and quantification of plant
leaf structure using X-ray micro-computed tomography: Viewpoint. J.
Exp. Bot. 64:385–390. doi:10.1093/jxb/ers392
Pyka, G., G. Kerckhofs, I. Papantoniou, M. Speirs, J. Schrooten, and M.
Wevers. 2013. Surface roughness and morphology customization of
additive manufactured open porous Ti6Al4V structures. Materials
6:4737–4757. doi:10.3390/ma6104737
Pyka, G., G. Kerckhofs, J. Schrooten, and M. Wevers. 2014. The effect
of spatial micro-CT image resolution and surface complexity on the
morphological 3D analysis of open porous structures. Mater. Charact.
87:104–115. doi:10.1016/j.matchar.2013.11.004
Riley, A., C.J. Sturrock, S.J. Mooney, and M.R. Luck. 2014. Quantification of
eggshell microstructure using X-ray micro computed tomography. Br.
Poult. Sci. 55:311–320. doi:10.1080/00071668.2014.924093
Ritz, K., and I.M. Young, editors. 2011. Architecture and biology of soils.
CAB Int., Wallingford, UK.
Rousseau, B., P. Rivard, A. Marache, G. Ballivy, and J. Riss. 2012.
Limitations of laser profilometry in measuring surface topography
of polycrystalline rocks. Int. J. Rock Mech. Min. Sci. 52:56–60.
doi:10.1016/j.ijrmms.2012.03.003
Sahin, M., C.S. Çetinarslan, and H.E. Akata. 2007. Effect of surface roughness
on friction coefficients during upsetting processes for different
materials. Mater. Des. 28:633–640. doi:10.1016/j.matdes.2005.07.019
Sandin, M., J. Koestel, N. Jarvis, and M. Larsbo. 2017. Post-tillage
evolution of structural pore space and saturated and near-saturated
hydraulic conductivity in a clay loam soil. Soil Tillage Res. 165:161–168.
doi:10.1016/j.still.2016.08.004
Scott, P.J. 2000. Scale-space techniques. In: Proceedings of the 10th
International Colloquium on Surfaces, Chemnitz, Germany. 31 Jan.–1
Feb. 2000. Chemnitz Univ. of Technology, Chemnitz. p. 153–161.
Shunmugam, M.S., and V. Radhakrishnan. 1974. Computation of the
three-dimensional envelope for roughness measurement. Int. J. Mach.
Tool Des. Res. 14:211–216. doi:10.1016/0020-7357(74)90028-6
Srinivasan, V. 1998. Discrete morphological filters for metrology. In: P.H.
Osanna et al., editors, Proceedings of the 6th ISMQC Imeko Symposium
on Metrology for Quality Control in Production, Vienna. 8–10 Sept.
1998. Abteilung Austauschbau und Messtechnik, Technische Univ.
Wien, Vienna. p. 623–628.
Taylor, J.B., A.L. Carrano, and S.G. Kandlikar. 2006. Characterization
of the effect of surface roughness and texture on fluid
flow: Past, present, and future. Int. J. Therm. Sci. 45:962–968.
doi:10.1016/j.ijthermalsci.2006.01.004
Tracy, S.R., K.R. Daly, C.J. Sturrock, N.M.J. Crout, S.J. Mooney, and T. Roose.
2015. Three-dimensional quantification of soil hydraulic properties
using X-ray computed tomography and image-based modeling.
Water Resour. Res. 51:1006–1022. doi:10.1002/2014WR016020
von Weingraber, H. 1954. Zur Definition der Oberflächenrauheit. In:
Werkstattstechnik Maschinenbau. Springer, Berlin.
Zakirov, T.R., A.A. Galeev, E.A. Korolev, and E.O. Statsenko. 2016. Flow properties
of sandstone and carbonate rocks by X-ray computed tomography.
Curr. Sci. 110:2142–2147. doi:10.18520/cs/v110/i11/2142-2148
