1
Digi al image-based measu emen o
deg ee o sa u a ion on mo ing soil
Ge a do Mo ales PhD candida e
Depa men o Ci il and En i onmen al Enginee ing
Uni e si a Poli ècnica de Ca alunya, Ba celona, Spain
ORCID: 0000-0002-9548-3556
Nú ia M. Pinyol Associa e P o esso (co esponding au ho )
Depa men o Ci il and En i onmen al Enginee ing
Uni e si a Poli ècnica de Ca alunya, Ba celona, Spain
Cen e In e nacional de Mè odes Numè ics a l'Enginye ia
ORCID: 0000-0002-1878-1365
Add ess: Campus No d UPC. Edi icio D2. 08034 Ba celona
Phone: 34 93 401 18 20
e-mail: nu [email protected]
Mau icio Al a ado Pos -Doc o a e candida e
Depa men o Ci il and En i onmen al Enginee ing
Uni e si a Poli ècnica de Ca alunya, Ba celona, Spain
ORCID: 0000-0001-6033-8327
Edua do Alonso Pé ez de Ág eda Full P o esso
Cen e In e nacional de Mè odes Numè ics a l'Enginye ia
Uni e si a Poli ècnica de Ca alunya, Ba celona, Spain
ORCID: 0000-0003-2472-3951
2
ABSTRACT
This pape p esen s a no el me hodology ha combines Pa icle Image Velocime y - Nume ical
Pa icles (PIV-NP) and Sho -Wa e In a ed Spec al Imaging o he non-in asi e measu emen o
displacemen s and deg ee o sa u a ion on mo ing soils. This allows o con inuous moni o ing and
isualiza ion o soil beha io and p o ides aluable insigh in o he de o ma ion pa e ns and
mois u e e olu ion. The me hod was applied o a small-scale dam ailu e expe imen . The
me hodology o e s an in eg a ed, comp ehensi e, and non-in asi e app oach o in es iga ing soil
physical models by simul aneously measu ing displacemen s, s ains, eloci ies, and deg ee o
sa u a ion in soils in mo ion.
Keywo ds: Digi al image co ela ion, pa icle image elocime y, SWIR image analysis, geo echnical
physical modelling, unsa u a ed soils
INTRODUCTION
Image-based echniques a e widely applied in labo a o y es s. Fo ins ance, pa icle image
elocime y (PIV) echnique (Ad ian 1991) is used in he geo echnical ield (Take e al. 2004; Wang
e al. 2020). Pinyol and Al a ado (2017) p esen ed a nume ical ool (PIV-NP) o p ocessing Eule ian-
based PIV displacemen measu emen s o p o ide la ge de o ma ions and s ains.
Soil wa e con en can be also measu ed om images because o he dependence o he soil su ace
e lec ance wi h he wa e con en (Leu 1977; Sadeghi e al. 2015). Pa e a e al. (2021) p esen ed a
p ocedu e o measu e he deg ee o sa u a ion (S ) om sho -wa e in a ed (SWIR) digi al images
by co ela ing he ligh e lec ance and S . This me hod allows measu ing wa e con en o speci ic
con igu a ion expe imen s and speci ic imes.
This No e p oposes a me hodology o con inuously mapping he inc emen al and cumula i e
displacemen s, eloci ies, s ains, and S o e a soil domain in mo ion by means o combining he
PIV-NP echnique (Pinyol and Al a ado 2017) and SWIR image-based me hodology p esen ed in
Pa e a e al. (2021). The measu emen s a e plo ed wi h he p e- and pos -p ocessing so wa e GiD
3
(GiD 2020). An ins umen ed small-scale homogeneous dam ailu e is analysed compa ing image-
based and senso -based measu emen s.
PIV-NP-S TECHNIQUE
Figu e 1 p esen s a schema ic illus a ion o he me hod. Following Pinyol and Al a ado (2017),
compa ing successi e images aken a an elapsed ime in e al, he Eule ian-based PIV analysis
me hodology p o ides he eloci y o cen es o ma e ial domain- illed subse s ixed in space which
a e aken as nodes o a suppo mesh in he PIV-NP me hodology. The ma e ial domain is hen
disc e ized in o nume ical pa icles (NPs) ha a e ini ially dis ibu ed wi hin he suppo mesh. The
eloci y alues a he nodes a e ans e ed in o he NPs using mapping unc ions and hei posi ion
a e upda ed a he end o each ime s ep. The S is measu ed a each ime s ep a subse cen es
coinciding wi h he suppo mesh nodes o SWIR images a e applying Gaussian smoo hing o e an
a ea o in e es su ounding each poin o measu e. The S is also ans e ed o he NPs. In his
con ex , he necessa y inpu da a is: (i) ime in e al be ween consecu i e ames; (ii) coo dina es
o subse cen es; (iii) PIV Eule ian eloci y esul s (Figu e 1a; and (i ) SWIR-based measu emen s o
S (Figu e 1b).
When using di e en came as o PIV-NP analysis, in isual ange, and S measu emen , in SWIR
ange, he images ha e di e en pe spec i es o cap u e he same scene. Me ging he wo images
is equi ed. I was made h ough he homog aphic echnique (Ha ley and Zisse man 2003).
Mo e de ails o he p ocedu e a e a ailable in he Supplemen a y Ma e ial.
APPLICATION
The soil selec ed o he expe imen was he same one used in Pa e a e al. (2021), bu in his case,
he sand was dyed (15 % o he g ains). The isual ange images we e aken wi h a Canon EOS 600D
came a wi h a CMOS image senso and 1920x1080 pixel esolu ion a 30 ames pe second. Fo
SWIR images, he same came a desc ibed in Pa e a e al. (2021) was used.
Figu e 2 shows wo simila calib a ion cu es de i ed om SWIR images o samples p epa ed a wo
oid a ios (e) (0.9 and 1.0). To cap u e he e ec o he soil densi y, Pa e a e al. (2021) in es iga ed
wha mois u e- ela ed a iable, ei he wa e con en o S , de e mines he ligh e lec ance o
4
sa u a ed soil. This is a key aspec especially when dealing wi h expe imen s o de o mable soils.
The au ho s, a e es ing samples a di e en e ( anging om 0.7 o 1.08) and wa e con en ,
concluded ha S is, ins ead o he wa e con en , he a iable ha can be p ope ly co ela ed wi h
he a ia ion o he pixel in ensi y.
The expe imen in ol es he we ing-induced ailu e o a small-scale dam placed in a anspa en
Pe spex ank 400mm high, 1000mm long, and 200mm deep (Figu e 3a). The dam was p epa ed by
mois amping se en laye s a 5.2% o ini ial wa e and 1.39g/cm3 o d y densi y (e=0.93), in a e age.
I was ins umen ed wi h six con en ional soil mois u e senso s emplaced a he ea wall o he
ank. Two di e en s ages a e no iced: (a) An ini ial we ing collapse; and (b) a subsequen apid
mo ion a e ailu e. Since wo p ocesses ha e di e en ime scales, wo ame speeds (1 ps and
30 ps) we e selec ed.
Figu e 3 shows some isual ames o he model and he c acks de eloped. The expe imen is
in e p e ed wi h he p oposed echnique (Figu e 4). Displacemen s and de ia o ic s ains inc eased
ab up ly when he oe o he downs eam slope eached comple e sa u a ion, a a ound 723s, and
c ack#1 de elops exhibi ing posi i e olume ic s ain, clea ly a ec ed by he loca ion o he
mois u e senso E2. The uppe pa emains s able, which may ha e an e ec on he ailu e
mechanism. Fu he sa u a ion induces he second ailu e phase (773s). C ack#2 condi ioned by he
posi ion o senso E1, mobilizes he uppe unsa u a ed egion. The ailu e delimi ed by c ack#2 did
no exhibi a sliding su ace mechanism. The soil expe ienced a oppling o wa d ins ead o a sliding
shea ing.
These obse a ions unde line he in luence o senso s in he na u e and geome y o ailu e. This
limi a ion is no p esen in image-based moni o ing.
The in e p e a ion o con inuous PIV-NP-S maps enables he assessmen o p o ile a ia ions o S
a speci ic loca ions and imes wi hin he model (Figu e 5). Image-based measu emen s i well wi h
senso eadings.
Figu e 6 shows a compa ison be ween S alues in e p e ed h ough PIV-NP-S echnique and senso
da a. The co espondence is good below he ime o soil sa u a ion. The d op in he senso alues
a e maximum sa u a ion may be explained by a decay in he pe mi i i y associa ed wi h we ing
wi h a liquid wi h high elec ical conduc i i y (wa e om he gene al ne wo k ins ead o he ini ially
deionized wa e used o mois amping o he laye s) as obse ed by Schwa z e al. (2013).
5
Mo eo e , he e o s in senso ’s alues nea he ailu e o he dam model may be explained by
obse ed c acks and ai gaps o he model.
No e ha he image-based measu emen s a he nume ical pa icle, ini ially loca ed a he posi ion
o he senso , can only be consis en ly compa ed wi h he ixed senso -based measu emen s while
he nume ical pa icle emains a i s ini ial posi ion.
CONCLUSIONS
The pape p esen s he combina ion o he PIV-NP and SWIR-S image analysis echniques o
measu ing he su ace S o soils du ing mo ion. The me hodology was success ully applied o he
analysis o soil mo ion and sa u a ion a ia ions du ing an ins umen ed small-scale dam ailu e
expe imen . The esul s show he capabili ies o his non-in asi e app oach in p o iding
comp ehensi e isualiza ion and analysis o de o ma ion pa e ns, mois u e changes and ailu e
mechanisms in soils. Some ad an ages a e obse ed when compa ed wi h ixed physical senso s in
case o la ge displacemen s. The wa e con en measu ed a ound he senso becomes un eliable
and inaccu a e when he su ounding soil goes in o mo ion. The applica ion o he echnique o
o he ypes o soils emains o be done.
ACKNOWLEDGEMENTS
The i s au ho acknowledges he inancial suppo o SENACYT (Panama). The second au ho
hanks he inancial suppo ecei ed om Se a Hún e P og am (Depa men o En e p ise and
Knowledge o he Sec e a ia o Uni e si ies and Resea ch o he Gene ali a de Ca alunya).
FUNDING STATEMENT
P ojec PDC2022-133222-I00 and PID2022-141429OB-I00 unded by
MCIN/AEI/10.13039/501100011033/FEDER, UE.
DATA AVAILABILITY STATEMENT
Da a gene a ed o analyzed du ing his s udy a e a ailable om he co esponding au ho upon
easonable eques .
6
REFERENCES
Ad ian, R.J. (1991) Pa icle-imaging echniques o expe imen al luid mechanics. Annual Re iew o
Fluid Mechanics, 23(1):261–304. doi:10.1146/annu e . l.23.010191.001401.
GiD (2020) The pe sonal p e- and pos -p ocesso . Ba celona (Spain): CIMNE.
www.gidsimula ion.com.
Ha ley, R., and Zisse man, A. (2003). Mul iple View Geome y in Compu e Vision. Camb idge
Uni e si y P ess.
Leu, D.J. (1977). Visible and nea - in a ed e lec ance o beach sands: a s udy on he spec al
e lec ance/g ain size ela ionship. Remo e Sensing o En i onmen , 6(3):169–182.
Pa e a, F., Pinyol, N.M., and Alonso, E.E. (2021). Massi e, con inuous, and non-in asi e su ace
measu emen o deg ee o sa u a ion by sho wa e in a ed images. Canadian Geo echnical
Jou nal, 58(6):749-762. doi:10.1139/cgj-2019-0051.
Pinyol, N.M., and Al a ado, M. (2017). No el analysis o la ge s ains based on pa icle image
elocime y. Canadian Geo echnical Jou nal, 54(7):933–944. doi: 10.1139/cgj-2016-0327.
Sadeghi, M., Jones, S. B., and Philpo , W. D. (2015). A linea physically-based model o emo e
sensing o soil mois u e using sho wa e in a ed bands. Remo e Sensing o En i onmen ,
164:66-76. doi: 10.1016/j. se.2015.04.007.
Schwa z, R.C., Casano a, J.J., Pelle ie , M.G., E e , S.R., and Baumha d , R.L. (2013). Soil
pe mi i i y esponse o bulk elec ical conduc i i y o selec ed soil wa e senso s. Vadose
Zone Jou nal, 12(2): 1-13. doi: 10.2136/ zj2012.0133.
Take, W.A., Bol on, M.D., Wong, P.C.P., and Yeung, F.J. (2004). E alua ion o landslide igge ing
mechanisms in model ill slopes. Landslides,1(3):173–184. doi:10.1007/s10346-004-0025-1.
Wang, Y., Hu, Y., and Hossain, M.S. (2020). Soil low mechanisms o ull- low pene ome e s in
laye ed clays h ough pa icle image elocime y analysis in cen i uge es . Canadian
Geo echnical Jou nal, 57(11):1719–1732. doi:10.1139/cgj-2018-0094.
FIGURE CAPTIONS
Figu e 1. Schema ic ep esen a ion o he S measu emen o soils du ing mo ion, using
PIV-NP disc e iza ion (one NP pe elemen ).
Figu e 2. SWIR-S calib a ion cu e o used ma e ial (85% na u al sand + 15% dyed sand).
7
Figu e 3. Visual ames o he small-scale dam model a di e en imes, , in seconds
including ini ial and de o med geome y senso s’ loca ion, wa e le el (Hw), and c acks and
de eloped ailu e mechanism.
Figu e 4. PIV-NP-S plo s o S , accumula ed displacemen , de ia o ic s ains and olume ic
s ains: (a) a igge ing, =723s; (b) a ailu e, =724s; (c) a pos - ailu e, =773s; and (d) a
he end o he es , =810s. The de ia o ic s ains a e de ined as �2 3
⁄𝑒𝑒𝑖𝑖𝑖𝑖𝑒𝑒𝑖𝑖𝑖𝑖, whe e 𝑒𝑒𝑖𝑖𝑖𝑖 is
he de ia o ic pa o he s ain enso .
Figu e 5. PIV-NP-S based measu emen s o S dis ibu ion and e ical p o iles compa ed
wi h punc ual senso -based measu emen s (a) Hw/Hdam = 0.20, = 240s; (b) Hw/Hdam = 0.40,
= 435s; and (c) Hw/Hdam = 0.60, = 645s. (Hw/Hdam is he a io o he wa e heigh o he
heigh o he dam).
Figu e 6. Valida ion o he SWIR image-based measu emen o S agains SEN0193
capaci i e soil mois u e senso s (S1, S2, and S3) and EC-5 capaci i e soil mois u e senso s
(E1, E2, and E3).
