Enginee ing and Technology Jou nal e-ISSN: 2456-3358
Volume 10 Issue 11 No embe 2025, Page No.- 7742-7751
DOI: 10.47191/e j/ 10i11.06, I.F. – 8.482
© 2025, ETJ
7742
ETJ Volume 10 Issue 11 No embe 2025, Thobani Maluleka
Assessing P e alen Open-Sou ce Insa Time Se ies Analysis Me hods o
G ound Subsidence Moni o ing In Mid aal, Sou h A ica
Thobani Maluleka1, Siphiwe Mphu hi2
1A chi ec u e Planning and Geoma ics, Uni e si y o Cape Town, Cape Town, Sou h A ica
2Geospa ial Resea ch and Inno a ions, Ea hSense Geospa ial Inc., Johannesbu g, Sou h A ica
ABSTRACT: G ound subsidence is a g owing geohaza d in he Mid aal egion o Sou h A ica, h ea ening in as uc u e,
economic ac i i ies, and communi y well-being. Cu en moni o ing echniques o en lack he necessa y spa ial and empo al
esolu ion, highligh ing an u gen need o mo e p ecise and e icien me hods. This s udy in es iga es he e ec i eness o p e alen
open-sou ce In e e ome ic Syn he ic Ape u e Rada (InSAR) ime se ies analysis me hods and ools o moni o ing g ound
subsidence in Mid aal. We used Sen inel-1 Single Look Complex image y om Janua y 2019 o Decembe 2021 and e alua ed
ou dis inc InSAR wo k lows: Pe sis en Sca e e (PS) InSAR using ISCE–S aMPS and SNAP–S aMPS, Small Baseline Subse
(SBAS) InSAR using ISCE–S aMPS and HyP3–Min Py. The AW3D, 30m esolu ion DEM was used o opog aphic phase
co ec ion, and con inuous Global Na iga ion Sa elli e Sys em (cGNSS) da a om Heidelbe g (HEID) and Ve eeniging (VERG)
s a ions alida ed InSAR-de i ed Line-o -Sigh (LOS) eloci ies using me ics like Roo Mean Squa e E o (RMSE), Mean
Absolu e E o (MAE), and Range E o . Ou esul s show clea spa ial pa e ns o g ound de o ma ion, wi h g ound subsidence
concen a ed in he no heas e n and sou heas e n egions and upli in he no hwes e n egion. ISCE-S aMPS PS-InSAR and
SBAS-InSAR demons a ed he highes p ecision wi h he lowes s anda d e o s, making hem sui able o de ec ing sub le
mo emen s. In con as , HyP3-Min Py SBAS-InSAR, while exhibi ing la ge s anda d e o s, p o ed aluable o cap u ing b oade ,
la ge-scale de o ma ions, including ex eme g ound subsidence a es o up o -233.9 mm/yea . Co ela ion analysis e ealed a s ong
posi i e co ela ion o 0.7 be ween ISCE-S aMPS and SNAP-S aMPS PS-InSAR, while o he me hod pai s showed weake
co ela ions, indica ing di e ences in he ype o sca e e s, dis inc s eng hs and limi a ions. Veloci y accu acy e alua ion agains
cGNSS da a showed ha ISCE-S aMPS SBAS-InSAR and SNAP-S aMPS PS-InSAR achie ed he lowes MAE and RMSE,
mee ing he NISAR mission’s alida ion c i e ion o secula g ound de o ma ion. This esea ch unde sco es he impo ance o
me hod selec ion based on speci ic s udy objec i es, whe he high p ecision o b oad spa ial co e age is p io i ized, and highligh s
he need o con inued e inemen o InSAR echniques o accu a e geohaza d moni o ing in dynamic en i onmen s.
KEYWORDS: G ound Subsidence, Insa Time Se ies Analysis, SBAS-Insa , PS-Insa
I. INTRODUCTION
G ound subsidence ep esen s a c i ical geohaza d ha a ec s
u ban and pe i-u ban a eas wo ldwide, wi h consequences
anging om in as uc u e damage o signi ican sa e y isks
o human se lemen s [1]. In Sou h A ica, he Mid aal
egion o Gau eng has expe ienced a g owing incidence o
g ound subsidence, posing a di ec h ea o i al
in as uc u e, economic ac i i ies, and communi y well-
being [2], [3], [4]. Despi e he se e i y o his issue, cu en
moni o ing and mi iga ion s a egies o en ely on
con en ional echniques, which a e equen ly cons ained by
limi ed spa ial and empo al esolu ion, as well as delayed
epo ing o g ound subsidence pa e ns [5]. These
sho comings highligh he u gen need o inno a i e, da a
d i en app oaches ha can enhance he p ecision and
e iciency o g ound subsidence moni o ing.
In e e ome ic Syn he ic Ape u e Rada (InSAR) ime se ies
analysis has eme ged as a ans o ma i e echnology o
moni o ing g ound de o ma ion [6], [7], [8]. I s capabili y o
de ec millime e su ace mo emen s ac oss ex ensi e a eas,
combined wi h high empo al and spa ial esolu ion, makes i
pa icula ly well-sui ed o add essing g ound subsidence
challenges [5], [9].
Ne e heless, a signi ican ba ie o i s widesp ead
applica ion lies in he a iabili y o ools and me hodologies
a ailable o InSAR da a p ocessing. Di e en so wa e
packages, algo i hmic amewo ks, and pa ame e
con igu a ions can yield di e gen esul s, complica ing
e o s o s anda dize and alida e indings. This a iabili y
necessi a es a sys ema ic e alua ion o InSAR ime se ies
echniques o de e mine hei eliabili y and applicabili y,
pa icula ly in egions like Mid aal whe e p ecise g ound
subsidence moni o ing is indispensable.
This s udy seeks o b idge hese gaps by in es iga ing he
e ec i eness o a ious open-sou ce InSAR ime se ies
analysis ools and me hods o moni o ing g ound subsidence
“Assessing P e alen Open-Sou ce Insa Time Se ies Analysis Me hods o G ound Subsidence Moni o ing In
Mid aal, Sou h A ica”
7743
ETJ Volume 10 Issue 11 No embe 2025, Thobani Maluleka
in Mid aal. By examining he pe o mance, accu acy, and
p ac icali y o hese ools, he esea ch aims o p o ide
insigh s in o hei sui abili y o add essing g ound
subsidence ela ed challenges in he Sou h A ican con ex .
The s udy no only con ibu es o ad ancing geospa ial
me hodologies bu also unde sco es he ole o mode n
echnologies in ackling geohaza ds. By ocusing on he
echnical and p ac ical aspec s o InSAR ime se ies analysis,
i aims o in o m u u e esea ch and ope a ional p ac ices,
pa ing he way o imp o ed moni o ing, mi iga ion, and
decision-making in egions suscep ible o g ound subsidence.
II. MATERIALS AND METHODS
A. S udy A ea
This s udy is conduc ed in Mid aal, loca ed in he
no heas e n egion o Sou h A ica, wi h coo dina es anging
om 26° 18’ 38’’ S o 26° 56’ 53’’ S in la i ude and 27° 50’
40’’ E o 28° 25’ 52’’ E in longi ude. Spanning app oxima ely
4,126.34 squa e kilome es, Mid aal is a highly popula ed
u al-u ban a ea, home o 112,254 esiden s as o 2022.
The egion’s geological landscape is no ably di e se, wi h
dolomi e o ma ions being pa icula ly ulne able o
pollu ion and g ound subsidence, which pose signi ican
challenges o sus ainabili y [2], [10]. Addi ionally, apid
u baniza ion, ag icul u al and indus ial expansion ha e
exe ed conside able s ain on Mid aal’s wa e esou ces,
leading o conce ns abou excessi e g oundwa e ex ac ion,
g ound subsidence, and he eme gence o sinkholes. Figu e 1.
p o ides a comp ehensi e map o he egion’s geog aphical
bounda ies, ele a ion and geological ea u es, illus a ing he
geological and opog aphic complexi ies o he a ea.
Figu e 1. Geog aphical loca ion and, (a) ele a ion and
(b) geological ea u es o Mid aal, Sou h A ica.
B. Da ase s
This s udy used a combina ion o sa elli e-based and g ound-
based da ase s o moni o g ound de o ma ion wi hin he
s udy a ea. The p ima y da ase was Sen inel-1 Single Look
Complex (SLC) image y acqui ed in descending
in e e ome ic wide (IW) mode, om he Alaska Sa elli e
Facili y (ASF) Dis ibu ed Ac i e A chi e Cen e (DAAC).
Sen inel-1’s C-band was selec ed due o i s high empo al
esolu ion (12-day e isi cycle pos -2017), wide swa h
co e age (250 km in IW mode), and p o en abili y o de ec
g ound de o ma ion e en in ege a ed o ag icul u al egions.
A o al o 84 descending o bi images spanning Janua y 2019
o Decembe 2021 we e used, cen ed consis en ly on ame
677 (pa h 50), ensu ing cohe ence and compa abili y ac oss
he ime se ies.
To co ec o opog aphic phase and suppo in e e ome ic
p ocessing, he “Ad anced Land Obse ing Sa elli e Wo ld
3D” 30 me e s esolu ion (AW3D30) Digi al Ele a ion
Model (DEM) was used. This 30 m esolu ion DEM,
e e enced o Wo ld Geode ic Sys em 1984 (WGS84) and
Ea h G a i a ional Model 1996 (EGM96), was selec ed
based on exis ing li e a u e [11] highligh ing i s demons a ed
supe io pe o mance in Sou h A ican condi ions, and
o e ing enhanced e ical accu acy o e o he publicly
a ailable DEM’s.
The alida ion o InSAR-de i ed Line-o -Sigh (LOS)
eloci ies was pe o med using con inuous Global
Na iga ion Sa elli e Sys em (cGNSS) s a ion da a om
Heidelbe g (HEID) and Ve eeniging (VERG). These
cGNSS/T igNe s a ions, ope a ed by he Chie Di ec o a e:
Na ional Geospa ial In o ma ion (CD: NGI) and p ocessed by
he Ne ada Geode ic Labo a o y (NGL), p o ided Eas -
No h-Up (ENU) displacemen ime se ies da a. Compa ison
wi h sa elli e InSAR de i ed eloci ies enabled independen
e i ica ion o g ound de o ma ion ends and de ec ion o
any disc epancies due o p ocessing e o s o localized
en i onmen al ac o s. I is no ed ha hese wo s a ions
ep esen he only g ound-based geode ic in as uc u e
a ailable o independen alida ion in his egion, hus some
limi a ion due o spa ial unde ep esen a ion. Despi e his
cons ain , he cGNSS da a s ill p o ides c ucial poin -based
alida ion o he Line-o -Sigh eloci ies, con ibu ing o he
o e all assessmen o me hod accu acy.
C. InSAR P ocessing Techniques
To assess he impac o a ious InSAR p e-p ocessing,
p ocessing, and ime se ies analysis echniques on he
de ec ion o g ound de o ma ion ac oss he s udy a ea, ou
dis inc wo k lows we e implemen ed. The i s wo k low
employed he Pe sis en Sca e e (PS) echnique using he
InSAR Scien i ic Compu ing En i onmen (ISCE) in
combina ion wi h he S an o d Me hod o Pe sis en
Sca e e s (S aMPS). The second used he Small Baseline
Subse (SBAS) app oach, also in eg a ing ISCE and S aMPS.
The hi d applied he PS me hod using he SeN inel
(a)
(b)
“Assessing P e alen Open-Sou ce Insa Time Se ies Analysis Me hods o G ound Subsidence Moni o ing In
Mid aal, Sou h A ica”
7744
ETJ Volume 10 Issue 11 No embe 2025, Thobani Maluleka
Applica ion Pla o m (SNAP) oge he wi h S aMPS. Las ly,
he ou h wo k low inco po a ed he ASF Hyb id Pluggable
P ocessing Pipeline (HyP3) wi h he Miami InSAR Time-
se ies so wa e in Py hon (Min Py) [12], [13], [14], [15], [16],
[17]. These me hods we e chosen o cap u e bo h localized
and egional pa e ns o g ound de o ma ion while enabling
c oss- alida ion o esul s h ough me hodological
edundancy.
1. PS-InSAR (ISCE–S aMPS)
This app oach in ol ed gene a ing w apped in e e og ams
using he InSAR Scien i ic Compu ing En i onmen , which
we e hen con e ed in o S aMPS-compa ible o ma . In
S aMPS, cohe en pe manen sca e e s we e iden i ied based
on ampli ude dispe sion and phase s abili y. Phase
unw apping was ca ied ou using SNAPHU, and spa ially
unco ela ed look angle e o s we e co ec ed. Addi ional
co ec ions add essed e o s a ising om DEM inaccu acies,
a mosphe ic e ec s, and sa elli e o bi misma ches. And he
a mosphe ic phase delays we e il e ed using he TRAIN
oolbox o imp o e he accu acy o he inal eloci y
es ima es by mi iga ing a mosphe ic phase delays.
2. SBAS-InSAR (ISCE–S aMPS)
This wo k low used ISCE o gene a e a ne wo k o small
baseline in e e og ams, which we e hen p ocessed in
S aMPS o de i e g ound displacemen o e ime. The
me hod op imized in e e og am selec ion o empo al and
spa ial cohe ence. Co ec ions we e applied o unw apping
e o s and a mosphe ic delays, and a e age eloci ies we e
es ima ed using leas -squa es in e sion o he ime-se ies
da a. This me hod p o ed e ec i e o mapping b oad-scale
g ound de o ma ion pa e ns.
3. PS-InSAR (SNAP–S aMPS)
In his wo k low, in e e og ams we e gene a ed using he
Sen inel Applica ion Pla o m, de eloped by Eu opean Space
Agency (ESA). The esul s we e hen o ma ed o S aMPS,
whe e PS-InSAR analysis was pe o med. The p ocessing
s eps we e simila o hose in he ISCE–S aMPS PS-InSAR
wo k low, including sca e e selec ion, phase unw apping,
and a mosphe ic co ec ion. This al e na i e p ocessing pa h
se ed o alida e he consis ency and eliabili y o PS-InSAR
esul s ob ained om di e en p e-p ocessing ools and
me hods.
4. SBAS-InSAR (HyP3–Min Py)
This me hod in ol ed he use o ASF HyP3 pla o m o
gene a e unw apped in e e og ams, which we e hen
analysed using Min Py. Key p ocessing s eps included
ne wo k e inemen , oposphe ic delay co ec ion, and
emo al o opog aphic and unw apping e o s. Re e ence
poin selec ion and op imiza ion o he in e e ome ic
ne wo k we e also pe o med p io o eloci y es ima ion.
The app oach was e icien o de ec ing g adual, spa ially
ex ensi e g ound de o ma ion signals.
Toge he , hese me hods enabled he gene a ion o InSAR
ime-se ies analysis g ound de o ma ion maps and a e age
eloci y ields. C oss compa ison be ween he ou pu s
p o ided insigh s in o he s eng hs and limi a ions o each
app oach, me hods and ein o ced he eliabili y o he
de ec ed g ound de o ma ion ends.
D. Co ela ion Analysis
To e alua e he le el o ag eemen be ween he di e en
InSAR ime se ies analysis p ocessing echniques, Pea son
co ela ion analysis was applied o he eloci y maps
gene a ed by ISCE-S aMPS PS-InSAR, ISCE-S aMPS
SBAS-InSAR, SNAP-S aMPS PS-InSAR, and HyP3-Min Py
SBAS-InSAR. All eloci y ou pu s we e con e ed in o 3 a c-
second as e g ids using he s amps_ o_ as e and
h5_ o_ as e modules wi hin he S oSAP p ocesso . S aMPS
CSV iles we e i s mapped o g id pixels based on
geog aphic coo dina es, while Min Py H5 ou pu s we e
aligned using a e e ence as e wi h EPSG:4326 p ojec ion.
Once co egis e ed, he as e s we e compa ed on a pixel-by-
pixel basis using only he common spa ial ex en ac oss all
me hods. Pea son co ela ion coe icien s we e calcula ed o
quan i y he ela ionships and used o isualize he
consis ency be ween da ase s.
E. F equency Dis ibu ion Analysis
A equency dis ibu ion analysis was pe o med using he
common g id pixels ac oss all ou InSAR ime se ies
analysis me hods o cha ac e ize he dis ibu ion o g ound
de o ma ion eloci y alues. The g idded da a we e classi ied
in o p ede ined eloci y in e als, and he numbe o
occu ences wi hin each class was allied. This p ocedu e was
used o obse e how each InSAR ime se ies analysis me hod
ep esen ed spa ial pa e ns in he g ound de o ma ion
eloci y ield and o assess a ia ion in he dis ibu ion o
measu ed g ound displacemen s.
F. Accu acy E alua ion
The accu acy o InSAR ime se ies analysis de i ed g ound
de o ma ion eloci ies was assessed by compa ing hem wi h
da a om wo con inuous GNSS s a ions, he Heidelbe g and
Ve eeniging s a ions. Fo S aMPS-based inhe en poin -
based ou pu s, cGNSS compa ison was conduc ed by
in e pola ing he a e age eloci y om he 25 nea es
PS/SBAS poin s loca ed wi hin a 2 km adius o each s a ion
using he s amps_nea by module wi hin he S oSAP
p ocesso . Fo Min Py-based g idded esul s, as e alues a
he s a ion coo dina es we e ex ac ed using he ‘Ex ac
Mul i Values o Poin s’ ool in A cPy a e con e ing he H5
iles in o as e o ma . While he eloci y
ex ac ion/in e pola ion me hods di e , hey a e chosen o be
mos sui able o he espec i e da a s uc u es p oduced by
each InSAR so wa e. Subsequen ly, accu acy was conduc ed
using me ics such as Roo Mean Squa e E o (RMSE),
Mean Absolu e E o (MAE), and Range E o . These me ics
“Assessing P e alen Open-Sou ce Insa Time Se ies Analysis Me hods o G ound Subsidence Moni o ing In
Mid aal, Sou h A ica”
7745
ETJ Volume 10 Issue 11 No embe 2025, Thobani Maluleka
we e compu ed o quan i y he di e ence be ween cGNSS
and InSAR ime se ies analysis es ima es.
G. Baseline E o Compa ison
The baseline eloci y e o e alua ion was unde aken o
de e mine he e ec i eness o each InSAR ime se ies
analysis me hod in cap u ing long- e m, s able g ound
de o ma ion. This assessmen in ol ed compa ing he linea
(secula ) eloci ies de i ed om InSAR wi h hose ob ained
om con inuous GNSS (cGNSS) s a ions, wi h he aim o
e alua ing consis ency wi hin a de ined accu acy h eshold.
The alida ion ollowed s anda ds simila o hose p esc ibed
by sa elli e missions such as ‘Na ional Ae onau ics and Space
Adminis a ion’ (NASA) – ‘Indian Space Resea ch
O ganiza ion’ (ISRO) Syn he ic Ape u e Rada (NISAR),
which equi e InSAR-de i ed measu emen s o achie e an
accu acy o 2 mm/yea o secula de o ma ion o
3 × (1 + L¹ᐟ²) mm/yea o ansien de o ma ion, whe e L is
he baseline leng h in kilome es wi hin he ange o 0.1 km
o 50 km. These h esholds p o ide a benchma k o
de e mining whe he he obse ed di e ences be ween
InSAR and cGNSS eloci y es ima es all wi hin accep able
limi s o mission-le el alida ion and long- e m geophysical
in e p e a ions.
III. RESULTS
A. G ound De o ma ion Pa e n
The eloci y maps de i ed om he di e en me hods and
so wa e showed clea spa ial pa e ns o g ound de o ma ion
ac oss he s udy a ea (Figu e 2.). G ound subsidence was
p edominan ly obse ed in he no heas e n and sou heas e n
egions, while upli was mo e localized in he no hwes e n
egion. Speci ically, he maximum g ound subsidence a e
obse ed in he ISCE-S aMPS PS-InSAR p ocessing was -8.7
mm/yea , while he maximum g ound upli a e eached 2.7
mm/yea . The s anda d e o s o hese measu emen s anged
om 0.1 mm/yea o 1.0 mm/yea , e lec ing he accu acy o
he PS-InSAR me hod. In con as , he ISCE-S aMPS SBAS-
InSAR me hod e ealed mo e ex ensi e a eas o g ound
subsidence, wi h a maximum a e o -25.8 mm/yea in he
no he n, sou hwes e n, and eas e n egions, and signi ican
g ound upli o up o 27.8 mm/yea in he sou heas e n a ea.
S anda d e o s o his me hod anged om 0.1 mm/yea o
3.5 mm/yea , showing a wide a iabili y in he esul s.
SNAP-S aMPS PS-InSAR also showed g ound subsidence in
he sou heas e n pa o he s udy a ea, wi h maximum g ound
subsidence o -6.4 mm/yea and g ound upli o 1.8 mm/yea .
The s anda d e o s anged om 0.1 mm/yea o 1.3 mm/yea ,
simila o ISCE-S aMPS PS-InSAR bu wi h sligh ly highe
e o s. Fo HyP3-Min Py SBAS-InSAR, g ound subsidence
was obse ed in he sou heas e n, no heas e n, and
sou hwes e n pa s o he s udy a ea, wi h he maximum
g ound subsidence a e eaching -233.9 mm/yea and
maximum g ound upli o 49.2 mm/yea . S anda d e o s o
his me hod anged om 0.0 mm/yea o 5.4 mm/yea ,
e lec ing bo h he la ge-scale g ound de o ma ion obse ed
and he highe a iabili y in measu emen s. The ex eme
g ound subsidence a es a e o en associa ed wi h highly
localized phenomena ha may include ac i e mining, in ense
g oundwa e ex ac ion, o localized geological ins abili ies,
which a e p e alen in pa s o he Mid aal egion.
(c)
(d)
( )
(e)
(a)
(b)
“Assessing P e alen Open-Sou ce Insa Time Se ies Analysis Me hods o G ound Subsidence Moni o ing In
Mid aal, Sou h A ica”
7746
ETJ Volume 10 Issue 11 No embe 2025, Thobani Maluleka
Figu e 2. G ound de o ma ion eloci y and hei
co esponding eloci y s anda d e o maps ob ained
om, (a, b) ISCE-S aMPS PS-InSAR, (c, d) ISCE-
S aMPS SBAS-InSAR, (e, ) SNAP-S aMPS PS-InSAR,
(g, h) HyP3-Min Py SBAS-InSAR.
B. Co ela ion Be ween Me hods
The Pea son co ela ion analysis be ween he di e en InSAR
ime se ies analysis me hods e ealed no able ends (Figu e
3. And Figu e 4.). ISCE-S aMPS and SNAP-S aMPS PS-
InSAR showed he s onges posi i e co ela ion o 0.70 wi h
a bes - i slope o 0.48, indica ing simila beha iou in g ound
de o ma ion measu emen and ype o sca e e s. By con as ,
he co ela ion be ween ISCE-S aMPS PS-InSAR and HyP3-
Min Py SBAS-InSAR was weak a 0.27 wi h a slope o 0.40,
highligh ing me hodological di e ences and a ia ions in
hei abili y o cap u e de o ma ion pa e ns. ISCE-S aMPS
PS-InSAR and ISCE SBAS-InSAR showed a weak nega i e
co ela ion (-0.13) wi h a nega i e slope (-0.15), unde sco ing
con as ing PS and DS poin cha ac e is ics. Simila ly,
SNAP-S aMPS PS-InSAR and HyP3-Min Py SBAS-InSAR
showed a weak posi i e co ela ion o 0.19 (slope 0.41), while
SNAP-S aMPS PS-InSAR and ISCE SBAS-InSAR had a
weak posi i e co ela ion o 0.23 (slope 0.38). The weakes
co ela ion was obse ed be ween HyP3-Min Py SBAS-
InSAR and ISCE SBAS-InSAR (0.11, slope 0.08), whe e
poin s we e symme ically dis ibu ed abou he bes - i line
bu o med an o al-shaped clus e , sugges ing s ong non-
linea pa e ns and ool- ela ed di e ences.
Figu e 3. Techniques and Tools Pea son co ela ion.
Figu e 4. Co ela ion sca e plo s be ween InSAR
echniques and ools, showing bes - i lines, clus e ing
beha iou , and di e ences in eloci y ange cap u e.
O e all, he sca e plo s (Figu e 4.) demons a e ha while
all echniques p oduce poin clus e s along bes - i lines, hei
clus e ing pa e ns a y. HyP3-Min Py SBAS-InSAR and
ISCE-S aMPS SBAS-InSAR a e pa icula ly concen a ed
wi hin -20 mm/y o 20 mm/y , which unde sco es hei
limi ed abili y o cap u e highe -magni ude eloci ies
compa ed o HyP3-Min Py SBAS-InSAR, which de ec s
de o ma ion beyond ±30 mm/y . These a ia ions align wi h
he eloci y maps and highligh how me hodological and ool-
based di e ences in luence he de ec ion and in e p e a ion o
g ound subsidence signals.
C. F equency Dis ibu ion o Veloci ies
Le e aging he common pixel poin s among all echniques,
we plo ed he equency dis ibu ion o he eloci ies o
unde s and hei class pa e ns (Figu e 5.). ISCE-S aMPS PS-
InSAR, HyP3-Min Py SBAS-InSAR, and ISCE-S aMPS
SBAS-InSAR showed highe eloci y equencies a 0
mm/yea , excep o SNAP-S aMPS PS-InSAR, which
displayed he highes equency a a ound -2 mm/yea . This
indica es ha eloci ies be ween ISCE-S aMPS PS-InSAR,
HyP3-Min Py SBAS-InSAR, and ISCE-S aMPS SBAS-
InSAR a e simila and di e en om SNAP-S aMPS PS-
InSAR, which may indica e e o s in he SNAP-S aMPS PS-
InSAR echnique. All ou echniques showed posi i ely
(g)
(h)
“Assessing P e alen Open-Sou ce Insa Time Se ies Analysis Me hods o G ound Subsidence Moni o ing In
Mid aal, Sou h A ica”
7747
ETJ Volume 10 Issue 11 No embe 2025, Thobani Maluleka
skewed eloci ies, hough ISCE-S aMPS and HyP3-Min Py
SBAS-InSAR showed much be e no mal dis ibu ion
pa e ns.
Figu e 5. Techniques and ools equency dis ibu ion.
D. Veloci y Accu acy E alua ion
Veloci y accu acy e alua ion was conduc ed by compa ing
he InSAR ime se ies analysis de i ed eloci ies wi h da a
om con inuous GNSS s a ions (HEID and VERG) (Figu e
6.). The esul s e ealed no able di e ences in pe o mance
ac oss he p ocessing me hods and so wa e used. SNAP-
S aMPS PS-InSAR exhibi ed he highes ela i e eloci y
e o s a bo h alida ion s a ions, 10.0 mm/yea a HEID and
8.0 mm/yea a VERG. In con as , ISCE-S aMPS SBAS-
InSAR demons a ed be e alignmen wi h cGNSS-de i ed
eloci ies, likely due o i s e ec i eness in mi iga ing phase
unw apping and a mosphe ic delay e o s, which commonly
a ec PS p ocessing in he e ogeneous o low-cohe ence
en i onmen s. O e all, HyP3-Min Py SBAS-InSAR showed
he bes pe o mance among he es ed me hods.
Figu e 6. Rela i e LOS Veloci y di e ences be ween
cGNSS and InSAR implied.
HyP3-Min Py SBAS-InSAR, while capable o cap u ing a
b oade ange o g ound de o ma ion eloci ies, including
ex eme displacemen s, showed he la ges s anda d e o
alues, pa icula ly in he wes e n pa o he s udy a ea,
whe e e o s eached up o ±5.4 mm/yea . A s a ion VERG,
HyP3-Min Py SBAS-InSAR epo ed a la ge absolu e
eloci y e o o −5.9 mm/yea , consis en wi h he s anda d
e o map and sugges ing po en ial limi a ions o p ecise
eloci y es ima ion a ha loca ion. ISCE-S aMPS PS-InSAR
also pe o med poo ly a VERG, u he indica ing ha si e-
speci ic ac o s such as a mosphe ic he e ogenei y o local
su ace cha ac e is ics may nega i ely impac Pe sis en
Sca e e analysis in ha a ea.
Based on he absolu e compa ison a e geo e e encing and
adjus ing he InSAR-de i ed eloci ies using s a ion HEID,
he pe o mance o he me hods was eassessed wi h
e e ence o s a ion VERG (Figu e 7.). Unde his amewo k,
ISCE-S aMPS SBAS-InSAR and SNAP-S aMPS PS-InSAR
exhibi ed he bes ag eemen wi h he cGNSS obse a ions,
ollowed by ISCE-S aMPS PS-InSAR, while HyP3-Min Py
SBAS-InSAR pe o med he weakes . The impo ance o
geo e e enced esul s lies in hei abili y o p o ide absolu e
eloci y es ima es ha a e consis en ac oss he s udy a ea,
he eby imp o ing he eliabili y o in e -s a ion compa isons
and enabling in eg a ion wi h o he geode ic da ase s. The
a ia ion in pe o mance ac oss me hods can also be
a ibu ed o he ype and dis ibu ion o sca e e s; Pe sis en
Sca e e echniques such as SNAP-S aMPS PS-InSAR,
gene ally ely on b igh , s able poin a ge s (e.g., buildings
o ock ou c ops), whe eas Small Baseline app oaches, like
ISCE-S aMPS and HyP3-Min Py SBAS-InSAR inco po a e
dis ibu ed sca e e s, which may be mo e sensi i e o
deco ela ion in he e ogeneous o ege a ed en i onmen s.
These di e ences in sca e e cha ac e is ics di ec ly
in luence e o p opaga ion and explain he con as ing
beha iou obse ed a s a ion VERG.
Figu e 7. Absolu e LOS Veloci y di e ences be ween
cGNSS and InSAR implied.
Addi ionally, s a is ical analysis o he absolu e eloci y
di e ences con i med hese ends (Table 1.). ISCE-S aMPS
SBAS-InSAR and SNAP-S aMPS PS-InSAR bo h yielded
he lowes mean absolu e e o (1.0 mm/yea ), na owes e o
ange (2.0 mm/yea ), and lowes RMSE (1.4 mm/yea ),
indica ing hei sui abili y o applica ions equi ing p ecise
de ec ion o sub le, long- e m g ound mo emen s, such as
ec onic pla e moni o ing. Con e sely, HyP3-Min Py SBAS-
InSAR eco ded he highes RMSE (4.2 mm/yea ), wides
e o ange (5.9 mm/yea ), and la ges mean absolu e e o
“Assessing P e alen Open-Sou ce Insa Time Se ies Analysis Me hods o G ound Subsidence Moni o ing In
Mid aal, Sou h A ica”
7748
ETJ Volume 10 Issue 11 No embe 2025, Thobani Maluleka
(3.0 mm/yea ), e lec ing lowe p ecision despi e i s b oade
de o ma ion de ec ion capabili y.
Table 1. Absolu e Veloci y di e ences s a is ics (uni s:
mm/yea ).
Me ic
Pe sis en Sca e
Small Baseline Subse
ISCES aMPS
SNAPS aMPS
ISCES aMPS
HyP3Min Py
MAE
1.5
1.0
1.0
3.0
Range
2.9
2.0
2.0
5.9
RMSE
2.1
1.4
1.4
4.2
E. Baseline E o Compa ison
The baseline e o compa isons be ween he InSAR ime
se ies analysis de i ed eloci ies and cGNSS da a e ealed
ha ISCE-S aMPS SBAS-InSAR and SNAP-S aMPS PS-
InSAR me he NISAR mission’s alida ion c i e ion o ±2
mm/yea o long- e m (secula ) g ound de o ma ion (Table
2.). In con as , ISCE-S aMPS PS-InSAR and HyP3-Min Py
SBAS-InSAR exceeded his h eshold, indica ing educed
accu acy in de ec ing s able mo ion. Despi e hese di e ences
in secula accu acy, all ou InSAR me hods complied wi h
he NISAR-de ined h eshold o ansien de o ma ion o
±24.16 mm/yea o e spa ial baselines om 0.1 km o 50 km.
This sugges s ha each me hod emains capable o
iden i ying sho - e m o highly localized g ound
displacemen e en s ac oss he s udy a ea.
Table 2. Baseline (HEID-VERG) LOS eloci y di e ences
be ween cGNSS and InSAR implied (uni s: mm/yea ).
Me hod ID
Baseline E o
Dis ance
(km)
NISAR E o
Th eshold
ISCEPS
2.9
49.73
24.16
ISCESBAS
2.0
49.73
24.16
SNAPPS
2.0
49.73
24.16
HyP3SBAS
5.9
49.73
24.16
Ca ea : The p o ided MAE, RMSE, and Range alues a e
poin es ima es de i ed om he limi ed cGNSS s a ions
a ailable o his la ge s udy a ea.
IV. DISCUSSION
A. Compa ison o G ound De o ma ion Pa e ns
The analysis o g ound de o ma ion pa e ns de i ed om he
di e en InSAR ime se ies analysis me hods indica e clea
spa ial a ia ions in he measu emen s. Each me hod cap u ed
g ound subsidence and g ound upli di e en ly ac oss he
s udy a ea. The ISCE-S aMPS PS-InSAR echnique, which
showed p ecise measu emen s wi h lowe s anda d e o s,
cap u ed g ound subsidence and g ound upli p edominan ly
in he no heas e n and sou heas e n egions. This sugges s
ha he echnique’s p ecision makes i sui able o moni o ing
a eas whe e g ound de o ma ion is sub le and equi es
accu a e de ec ion.
In con as , he ISCE-S aMPS SBAS-InSAR me hod
highligh ed b oade a eas o g ound subsidence, including
signi ican egions in he no he n, sou hwes e n, and eas e n
pa s o he s udy a ea, as well as localized g ound upli . The
highe s anda d e o s associa ed wi h his echnique sugges
ha while i is mo e obus o la ge-scale g ound subsidence
mapping, i may be less p ecise han ISCE-S aMPS PS-
InSAR, pa icula ly in a eas wi h low g ound de o ma ion
a es.
SNAP-S aMPS PS-InSAR also iden i ied g ound subsidence
in he sou heas e n pa o he s udy a ea, wi h sligh ly highe
e o s compa ed o ISCE-S aMPS PS-InSAR. This could be
a ibu ed o di e ences in phase unw apping algo i hms and
a mosphe ic co ec ion me hods, which migh explain he
sligh disc epancies obse ed in he eloci ies. Finally, HyP3-
Min Py SBAS-InSAR, wi h i s b oad spa ial co e age,
de ec ed ex eme alues o g ound de o ma ion, including a
maximum g ound subsidence a e o -233.9 mm/yea . While
his high a iabili y in de o ma ion eloci ies e lec s he
me hod’s sensi i i y o la ge-scale mo emen , i also sugges s
he me hod’s po en ial limi a ions in de ec ing smalle ,
localized de o ma ions wi h p ecision. Fu he mo e, localized
in es iga ions employing g ound- u h su eys o high-
esolu ion op ical image y a e equi ed o de ini i ely
con i m he speci ic causes o hese ex eme de o ma ion
a es.
B. E alua ion o Technique P ecision
The p ecision o each InSAR ime se ies analysis echnique
is c ucial o accu a e g ound de o ma ion analysis. The
esul s showed ha ISCE-S aMPS PS-InSAR and SBAS-
InSAR exhibi ed he lowes s anda d e o s, making hem he
mos p ecise me hods o measu ing g ound subsidence and
upli , cap u ing bo h pe sis en and dis ibu ed sca e e s.
This is consis en wi h p e ious s udies which ha e
highligh ed ISCE-S aMPS obus algo i hms o handling
a ious g ound de o ma ion mechanisms [5]. SNAP-S aMPS
PS-InSAR, while sligh ly less p ecise, s ill p o ided aluable
da a bu wi h highe s anda d e o s, po en ially due o less
sophis ica ed phase unw apping me hods compa ed o ISCE-
S aMPS PS-InSAR.
HyP3-Min Py SBAS-InSAR demons a ed la ge s anda d
e o s, ye i s abili y o cap u e a wide ange o de o ma ion
eloci ies makes i a powe ul ool o de ec ing mo e
signi ican changes in g ound displacemen s. Howe e , he
me hod’s lowe p ecision sugges s ha i may be be e sui ed
o iden i ying la ge g ound de o ma ion pa e ns a he han
sub le o localized g ound mo emen s. The unique s eng hs
o each echnique unde line he impo ance o selec ing he
app op ia e me hod based on he speci ic equi emen s o a
gi en s udy, whe he p ecision o he abili y o de ec la ge-
scale g ound de o ma ion is p io i ized.
C. F equency Dis ibu ion and S a is ical Analysis
The equency dis ibu ion o eloci ies ac oss all echniques
p o ided impo an insigh s in o he pe o mance o each
me hod. The obse a ion ha ISCE-S aMPS PS-InSAR,
HyP3-Min Py SBAS-InSAR, and ISCE-S aMPS SBAS-
InSAR displayed highe equency dis ibu ions a 0
“Assessing P e alen Open-Sou ce Insa Time Se ies Analysis Me hods o G ound Subsidence Moni o ing In
Mid aal, Sou h A ica”
7749
ETJ Volume 10 Issue 11 No embe 2025, Thobani Maluleka
mm/yea , while SNAP-S aMPS PS-InSAR peaked a ound -2
mm/yea , sugges ing ha SNAP-S aMPS PS-InSAR migh be
mo e p one o e o s, pa icula ly in cases o low g ound
de o ma ion. This -2 mm/yea peak could also ha e
in oduced bias in he accu acy assessmen , po en ially
making SNAP-S aMPS PS-InSAR appea mo e accu a e han
i ac ually is when compa ed o a limi ed se o alida ion
s a ions. Inco po a ing addi ional alida ion s a ions in u u e
analyses would help o e eal and mi iga e such biases mo e
e ec i ely, p o iding a mo e obus e alua ion o me hod
pe o mance. Addi ionally, he simila i y be ween ISCE-
S aMPS PS-InSAR, ISCE-S aMPS SBAS-InSAR, and HyP3-
Min Py SBAS-InSAR eloci ies indica e ha hese me hods
can cap u e simila pa e ns o g ound mo ion, al hough hei
p ecision di e s.
The posi i ely skewed dis ibu ion o eloci ies o all
me hods u he suppo s he no ion ha g ound subsidence is
mo e p e alen in he s udy a ea han g ound upli , wi h he
dis ibu ion o eloci ies ending o clus e owa ds lowe
g ound de- o ma ion a es. ISCE-S aMPS and HyP3-Min Py
SBAS-InSAR, which showed be e no mal dis ibu ion
pa e ns, sugges ha hese me hods could be mo e eliable
o analysing g ound de o ma ion a he egional le el, whe e
ends can be expec ed o ollow mo e p edic able pa e ns.
D. Accu acy and E o Assessmen
The accu acy e alua ion e ealed ha ISCE-S aMPS SBAS-
InSAR and SNAP-S aMPS PS-InSAR we e mo e accu a e
when compa ed o cGNSS da a, wi h e o s emaining wi hin
he expec ed ange o secula g ound de o ma ion. HyP3-
Min Py SBAS-InSAR, while showing he leas pe o mance
in ela i e accu acy, none heless main ained e o alues
wi hin i s epo ed s anda d e o anges. This indica es ha
HyP3-Min Py is a obus app oach as i s accu acy and
associa ed unce ain ies e lec ealis ic es ima es and e o s,
he eby p o iding a eliable ep esen a ion o g ound
de o ma ion magni udes. Mo eo e , despi e i s la ge
s anda d e o s, HyP3-Min Py SBAS-InSAR was able o
cap u e la ge-scale g ound de o ma ion, pa icula ly in
egions wi h signi ican g ound subsidence. This inding
unde sco es he impo ance o conduc ing accu acy
assessmen s o e alua e he eliabili y o InSAR-based
measu emen s, especially when moni o ing la ge o highly
dynamic egions subjec o ec onic o an h opogenic
in luences
The di e ences in accu acy be ween me hods u he
emphasize he need o a comp ehensi e alida ion s a egy
ha includes bo h emo e sensing da a and g ound-based
obse a ions, such as GNSS measu emen s o ully assess he
pe o mance o each echnique. Addi ionally, he abili y o
de ec small-scale g ound mo emen s in a eas wi h ela i ely
low g ound de o ma ion a es emains a challenge o many
InSAR echniques, as e idenced by he la ge e o ma gins
in me hods like HyP3-Min Py SBAS-InSAR. Fu he mo e,
he HyP3-Min Py's SBAS-InSAR b oade de ec ion
capabili y, while aluable o la ge-scale changes migh
inhe en ly be mo e suscep ible o phase unw apping
ambigui ies o egions o high empo al deco ela ion
compa ed o PS-InSAR me hods, especially in a eas o apid
and se e e de o ma ion.
V. CONCLUSION
This s udy has demons a ed he u ili y and challenges o
a ious InSAR p ocessing echniques in moni o ing g ound
subsidence and upli . By compa ing ISCE-S aMPS PS-
InSAR, SNAP-S aMPS PS-InSAR, ISCE-S aMPS SBAS-
InSAR, and HyP3-Min Py SBAS-InSAR, he esul s e eal
ha each me hod o e s unique ad an ages depending on he
scale and p ecision equi ed. ISCE-S aMPS PS-InSAR and
SBAS-InSAR eme ged as he mos p ecise, while HyP3-
Min Py SBAS-InSAR p o ed aluable in de ec ing la ge-
scale g ound de o ma ions. The di e ences in accu acy and
s anda d e o s ac oss hese me hods highligh he impo ance
o choosing he app op ia e echnique based on speci ic s udy
objec i es, whe he p ecision o b oade spa ial co e age is
p io i ized. The indings o his s udy unde sco e he need o
ca e ul selec ion o InSAR me hods and ools, as each me hod
p o ides dis inc insigh s in o g ound de o ma ion. The
abili y o moni o bo h la ge and sub le g ound de o ma ions
is c i ical o applica ions anging om u ban planning o
ea hquake moni o ing. Howe e , he limi a ions o hese
me hods, pa icula ly in handling small-scale g ound
de o ma ions wi h high accu acy, sugges ha u he
imp o emen s a e necessa y.
Fu u e esea ch should ocus on enhancing he accu acy and
eliabili y o echniques like HyP3-Min Py SBAS-InSAR.
While HyP3-Min Py demons a ed obus ness by main aining
accu acy wi hin i s epo ed s anda d e o anges, u he
e inemen is needed o educe ela i e pe o mance gaps
wi h S aMPS-based me hods. Add essing limi a ions such as
he po en ial bias in oduced by he -2 mm/yea peak in
SNAP-S aMPS PS-InSAR, he sensi i i y o PS app oaches
o si e-speci ic sca e e ypes, and he in luence o
a mosphe ic he e ogenei y on eloci y es ima es will be
essen ial. Equally impo an is he assessmen o DEM e ec s
as esidual opog aphic e o s can p opaga e in o
in e e ome ic phase signals, he eby in luencing eloci y
accu acy in bo h PS and SBAS p ocessing chains. In addi ion,
expanding he numbe and spa ial dis ibu ion o alida ion
s a ions will help o de ec and mi iga e localized biases mo e
e ec i ely, ensu ing s onge alida ion o InSAR-de i ed
eloci ies. By e ining hese echniques and inco po a ing
di e se da ase s, InSAR can become mo e eliable and
b oadly applicable in dynamic and densely popula ed egions.
Ul ima ely, a mo e comp ehensi e and accu a e app oach o
g ound de o ma ion moni o ing will p o ide be e suppo
o decision-making in a eas a ec ed by g ound subsidence
and upli .
“Assessing P e alen Open-Sou ce Insa Time Se ies Analysis Me hods o G ound Subsidence Moni o ing In
Mid aal, Sou h A ica”
7750
ETJ Volume 10 Issue 11 No embe 2025, Thobani Maluleka
ACKNOWLEDGMENT
This esea ch was suppo ed by he Resea ch De elopmen
G an and he Pos g adua e Funding O ice a he UCT. The
InSAR da a used in his s udy we e p o ided by ASF DAAC.
The GNSS da a was ob ained om he Ne ada Geode ic
Labo a o y, cou esy o CD:NGI, and he AW3D30 DEM
was acqui ed om OpenTopog aphy, cou esy o JAXA.
Compu a ions we e pe o med using acili ies p o ided by
he Uni e si y o Cape Town’s ICTS High Pe o mance
Compu ing eam: hpc.uc .ac.za. The au ho s hank he UCT
high Pe o mance, he anonymous e iewe s and he edi o
o hei cons uc i e commen s.
DISCLOSURE STATEMENT
The au ho (s) decla e no con lic o in e es .
FUNDING
This wo k is suppo ed by he Uni e si y o Cape Town
Resea ch De elopmen G an and Pos g adua e Funding
O ice. We a e g a e ul o hei suppo , and all iews
exp essed he e a e hose o he au ho (s) no he unding
bodies.
NOTES ON CONTRIBUTORS
Thobani Maluleka:
Thobani Maluleka is a PhD candida e a he Di ision o
Geoma ics, Uni e si y o Cape Town, Sou h A ica. His
esea ch in e es s a e geospa ial da a science and a i icial
in elligence, InSAR ime se ies analysis, ealiza ion o
geode ic e e ence ames, geoid modelling and g a i y ield
analysis, da a usion, GIScience, emo e sensing,
en i onmen al modelling, and capaci y building in geospa ial
science and su eying.
Siphiwe M. Mphu hi:
Siphiwe Mphu hi is a Lec u e a he Di ision o Geoma ics,
Uni e si y o Cape Town, Sou h A ica. His esea ch in e es s
a e physical geodesy, wi h a pa icula ocus on g a ime ic
geoid modelling and he es ablishmen o e ical da ums.
This encompasses he mode niza ion o bo h ho izon al and
e ical geode ic e e ence ames, as well as he me iculous
adjus men and analysis o g a i y da a.
ORCID
Thobani Maluleka – h ps://o cid.o g/0000-0002-5156-2367
AUTHOR CONTRIBUTIONS
Thobani Maluleka – concep ualiza ion, manusc ip w i ing
o iginal d a , e iew and edi ing, me hodology, da a
cu a ion, in es iga ion, isualiza ion, o mal analysis.
Siphiwe M. Mphu hi – concep ualiza ion, manusc ip e iew
and edi ing, esou ces, alida ion, supe ision.
DATA AVAILABILITY STATEMENT
The da a ha suppo he indings o his s udy a e a ailable
om he co esponding au ho upon easonable eques .
REFERENCES
1. M. Baghe i-Ga kosh e al., ‘Land subsidence: A
global challenge’, Science o The To al
En i onmen , ol. 778, p. 146193, Jul. 2021,
h ps://doi.o g/10.1016/j.sci o en .2021.146193.
2. A. The on, J. Engelb ech , J. Kemp, W. Kleynhans,
and T. Tu nbull, ‘De ec ion o Sinkhole P ecu so s
Th ough SAR In e e ome y: Rada and Geological
Conside a ions’, IEEE Geosci. Remo e Sensing
Le ., ol. 14, no. 6, pp. 871–875, Jun. 2017,
h ps://doi.o g/10.1109/LGRS.2017.2684905.
3. H. Ash a and F. Cawood, ‘Geospa ial subsidence
haza d modelling a S e k on ein Ca es’, SA J o
Geoma ics, ol. 4, no. 3, p. 273, Aug. 2015,
h ps://doi.o g/10.4314/sajg. 4i3.8.
4. T. Maluleka and S. M. Mphu hi, ‘Assessing g ound
de o ma ion moni o ing echniques in Mid aal,
Sou h A ica’, Jou nal o Applied Geodesy, Jun.
2025, h ps://doi.o g/10.1515/jag-2024-0031.
5. P. Ki ui, S. Oi o, H. Wai haka, P. Ode a, B. Riedel,
and M. Ge ke, ‘De ec ion, cha ac e iza ion, and
analysis o land subsidence in Nai obi using
InSAR’, Na Haza ds, ol. 113, no. 1, pp. 213–236,
Aug. 2022, h ps://doi.o g/10.1007/s11069-022-
05296-w.
6. L. Chai, X. Xie, C. Wang, G. Tang, and Z. Song,
‘G ound subsidence isk assessmen me hod using
PS-InSAR and Ligh GBM: a case s udy o Shanghai
me o ne wo k’, In e na ional Jou nal o Digi al
Ea h, ol. 17, no. 1, p. 2297842, Dec. 2024,
h ps://doi.o g/10.1080/17538947.2023.2297842.
7. X. Qiao e al., ‘In e p e a ion and A ibu ion o
Coas al Land Subsidence: An InSAR and Machine
Lea ning Pe spec i e’, IEEE J. Sel. Top. Appl. Ea h
Obse a ions Remo e Sensing, ol. 17, pp. 4768–
4783, 2024,
h ps://doi.o g/10.1109/JSTARS.2024.3361391.
8. H. Guo e al., ‘La ge-Scale Land Subsidence
Moni o ing and P edic ion Based on SBAS-InSAR
Technology wi h Time-Se ies Sen inel-1A Sa elli e
Da a’, Remo e Sensing, ol. 15, no. 11, p. 2843,
May 2023, h ps://doi.o g/10.3390/ s15112843.
9. W. L. Hakim, M. F. Fadhillah, S. Pa k, B. P adhan,
J.-S. Won, and C.-W. Lee, ‘InSAR ime-se ies
analysis and suscep ibili y mapping o land
subsidence in Sema ang, Indonesia using
con olu ional neu al ne wo k and suppo ec o
eg ession’, Remo e Sensing o En i onmen , ol.
287, p. 113453, Ma . 2023,
h ps://doi.o g/10.1016/j. se.2023.113453.
10. S. Richa dson, ‘Sinkhole and subsidence eco d in
he chuniespoo g oup dolomi e, Gau eng, Sou h
A ica, Mas e ’s hesis. P e o ia: Uni e si y o
P e o ia, Depa men o Geology’. 2013.
11. M. Malindi and P. Ode a, ‘A new digi al ele a ion
model o e Sou h A ica based on g ound and
