Comparison of l band and x band differential interferometric synthetic aperture radar for mine subsidence monitoring in central utah

Comparison of L band and X band differential interferometric synthetic aperture radar for mine subsidence monitoring in central Utah International Journal of Mining Science and Technology xxx (2016) x[.]

Comparison of L-band and X-band differential interferometric synthetic

aperture radar for mine subsidence monitoring in central Utah

Department of Mining Engineering, University of Utah, Salt Lake City 84112, USA

a r t i c l e i n f o

Article history:

Received 9 July 2016

Received in revised form 18 August 2016

Accepted 20 September 2016

Available online xxxx

Keywords:

Mine subsidence

DInSAR

TerraSAR-X

ALOS

Interferometry

a b s t r a c t Differential interferometric synthetic aperture radar (DInSAR), a satellite-based remote sensing tech-nique, has potential application for measuring mine subsidence on a regional scale with high spatial and temporal resolutions However, the characteristics of synthetic aperture radar (SAR) data and the effectiveness of DInSAR for subsidence monitoring depend on the radar band (wavelength) This study evaluates the effectiveness of DInSAR for monitoring subsidence due to longwall mining in central Utah using L-band (24 cm wavelength) SAR data from the advanced land observing satellite (ALOS) and X-band (3 cm wavelength) SAR data from the TerraSAR-X mission In the Wasatch Plateau region

of central Utah, which is characterized by steep terrain and variable ground cover conditions, areas affected by longwall mine subsidence are identifiable using both L-band and X-band DInSAR Generally, using L-band data, subsidence magnitudes are measurable Compared to X-band, L-band data are less affected by signal saturation due to large deformation gradients and by temporal decorrelation due to changes in the surface conditions over time The L-band data tend to be stable over relatively long periods (months) Short wavelength X-band data are strongly affected by signal saturation and temporal decorrelation, but regions of subsidence are typically identifiable over short periods (days) Additionally, though subsidence magnitudes are difficult to precisely measure in the central Utah region using X-band data, they can often be reasonably estimated

Ó 2016 Published by Elsevier B.V on behalf of China University of Mining & Technology

1 Introduction

Differential interferometric synthetic aperture radar (DInSAR) is

a satellite-based remote sensing technique that can be used to

measure surface displacement over large regions with high spatial

resolution Under good conditions, displacements can be measured

with centimeter to subcentimeter accuracy[1,2] DInSAR also has

high temporal resolution, and imaging periods typically range from

10 to 50 days[3,4] In the last two decades, the application of

DIn-SAR for mine subsidence monitoring has been demonstrated in

coal basins in Europe, Australia, China, and the United States

Over-all, these studies have demonstrated good data resolution, strong

relationships between mine development and subsidence, and

rea-sonable agreement between displacements measured by DInSAR

and displacements measured by conventional surveys[5–20]

In radar interferometry phase measurements from two nearly

coincident radar images are used to precisely measure relative

dis-tances[21] Surface deformation, topography, and changes in the

satellite position contribute most significantly to changes in the radar path length In general, changes in the path length due to changes in the satellite position and topography are known or can be estimated, and as a consequence, centimeter-level changes

in the path length due to surface deformation can be measured Because phase measurements are used to quantify distance, the wavelength characteristics of the radar are important Synthetic Aperture Radar (SAR) sensors most commonly use either L-band (24 cm wavelength), C-band (6 cm wavelength), or X-band (3 cm wavelength) radar, and the imaging characteristics of the radar bands are different Radar waves tend to interact strongly with structures similar in size to the radar wavelength, and as a result, surfaces appear rougher in images acquired using shorter wave-lengths Longer wavelengths tend to have some penetration of veg-etation, dry soils, and ice; phase measurements from longer wavelengths tend to be less sensitive to small changes in the sur-face conditions over time Additionally, the maximum deformation gradients measurable by DInSAR depend significantly on the radar band and on the ground resolution (pixel size) of the image[22] Longer wavelengths are less sensitive to deformation per pixel than shorter wavelengths, and larger deformation gradients are measurable in higher resolution data

http://dx.doi.org/10.1016/j.ijmst.2016.11.012

2095-2686/Ó 2016 Published by Elsevier B.V on behalf of China University of Mining & Technology.

⇑ Corresponding author Tel.: +1 801 5853029.

E-mail address: jwempen@gmail.com (J.M Wempen).

Contents lists available atScienceDirect

International Journal of Mining Science and Technology

j o u r n a l h o m e p a g e : w w w e l s e v i e r c o m / l o c a t e / i j m s t

Please cite this article in press as: Wempen JM, McCarter MK Comparison of L-band and X-band differential interferometric synthetic aperture radar for

Though DInSAR has significant potential as a method for

subsi-dence monitoring, SAR systems have variable characteristics Using

SAR data appropriate for the regional surface characteristics and

deformation rates is important for subsidence monitoring This

study evaluates the effectiveness of L-band and X-band DInSAR

for monitoring subsidence due to longwall mining in the Wasatch

Plateau region of central Utah L-band SAR data from the advanced

land observing satellite (ALOS) and X-band SAR data from the

TerraSAR-X mission are used

2 Location and data

The Wasatch Plateau is characterized by rugged topography

with flat topped mesas and steeply incised canyons It is

geologi-cally complex and exists in a region of transition from the Colorado

Plateau to the east and the Basin and Range province to the west

[23] Dominant structures include northward trending normal

faults and grabens, vertical joints, and vertical strike-slip faults

[24] The subalpine region of the Wasatch Plateau is heavily

vege-tated; grasses, forbs, and low shrubs are dominant Steep northern

exposures at higher elevations are heavily forested, and at lower

elevations, aspen, pine and tall shrubs are common [25] Fig 1

shows a TerraSAR-X image of a region of the Wasatch Plateau In

this image, topographic characteristics of the Wasatch Plateau

region are discernable

In the study area, coal has been mined from the upper and

lower Hiawatha seams These seams are present in the lower 75–

110 m of the Blackhawk Formation (Mesaverde Group), which

has a total thickness ranging from 190 to 245 m Prominent

near-seam geology includes the castlegate and the star point

sand-stones, both massive, medium- to course-grained sandstones The

castlegate sandstone overlays the Blackhawk Formation and has

a thickness ranging from 45 to 150 m The starpoint sandstone, lies

beneath the Blackhawk Formation and has a thickness ranging

from 25 to 300 m[26] The mining heights range from to 3.8 m

and the typical overburden thickness ranges from 305 to 550 m

Generally, the maximum vertical subsidence occurs near the center

of the longwall panels, with maximum magnitudes from 1.5 to

1.8 m The average reported angle of draw is 15°[27]

The L-band SAR data used in this study were imaged by the

ALOS satellite ALOS was operated by the Japanese Aerospace

Exploration Agency from 2006 to 2011, and acquired data globally

with a minimum recurrence cycle of 46 days The data used in this

study were imaged in fine-beam mode with both single and dual

polarization; only the co-polarized images were used in interfero-metric processing The images have swath widths of 70 km, azi-muth resolutions of 10 m, and ground range resolutions of 10 and 20 m for single and dual polarizations, respectively[3] Imag-ing dates and characteristics of the ALOS data are given inTable 1 The ALOS data were acquired from a repository of SAR data main-tained by the Alaska Satellite Facility

The X-band SAR data used in this study were imaged by the TerraSAR-X mission satellites, TSX-1 which launched in 2007 and TDX-1 which launched in 2010 These satellites are operated by the German Aerospace Center, and acquire data with a minimum recurrence cycle of 11 days The data used in this study were imaged

in stripmap mode with single horizontal polarization The images have 30 km swath widths and maximum ground resolutions of 3.3 m in azimuth and 1.7 m in range[28] Imaging dates and charac-teristics of the TerraSAR-X data are given inTable 2 The TerraSAR-X data were acquired from the German Aerospace Center

3 Processing

In this study, data processing was performed using SARscapeÒ and ENVIÒsoftware To generate a subsidence map using DInSAR, first the perpendicular and temporal baselines of paired SAR images are estimated Next, the paired images are co-registered and a differential interferogram is formed The interferogram is then filtered and the data coherence is estimated Next, the inter-ferogram is unwrapped and the absolute interferometric phases are determined Finally, vertical deformation is calculated from the absolute phases[29]

Before interferometric processing, SAR images are often multi-looked, or spatially averaged Significant amplitude and phase vari-ation of the radar signal from pixel to pixel, caused by varivari-ation in the surface characteristics, make the images appear speckled[30] Spatially averaging the data reduces speckle, but it also reduces the spatial resolution Short wavelength X-band data are very sensitive

to surface deformation, and deformation gradients on the order of 0.016 m/pixel are measurable However, in areas with large defor-mation gradients, phases tend to saturate In this study, the TerraSAR-X images were not multi-looked and were processed at full resolution to limit phase saturation Longer wavelength L-band data are less sensitive to deformation, and deformation gra-dients on the order of 0.118 m/pixel are measurable In this study, the ALOS images were multi-looked to produce images with 20 m

by 20 m pixels

4 Results For the central Utah study region, three interferograms were generated using L-band data in intervals over the period from June

16 to December 17, 2010 The interferometric data parameters are summarized inTable 3 Phase fringes due to subsidence are identi-fiable in all of the interferograms, but all the interferograms have topographic artifacts and the data quality is variable Coherence

is a statistics that quantifies the sameness of the radar signals in two paired images and it ranges from zero, which represents com-plete decorrelation, to one which represents perfect correlation In general, coherence reflects the quality of the phase measurements

[30] The average coherence of the ALOS data ranges from a low of 0.36 for the period from September 16 to December 17, 2010, to a high of 0.63 for the period from August 1 to September 16, 2010 Vertical displacement maps derived from the interferograms for each period are shown inFig 2,Fig 3shows accumulated displace-ment for the 184-day period from June 16 to December 17, 2010 In

Figs 2 and 3subsidence is contoured every 10 cm starting from

-10 cm of vertical displacement.Figs 4 and 5show the progression

Fig 1 TDX-1 intensity image of the Wasatch Plateau region (October 20, 2015).

of subsidence over time along sections AA0and BB0(Fig 3)

Maxi-mum measured subsidence during the June 16 to December 17,

2010 period is 1.5 m Gaps in the data in all of these figures are

due to pixels with low coherence

As noted inFig 2, subsidence is contoured every 10 cm, starting

from10 cm of displacement

As shown in Fig 3, subsidence is contoured every 10 cm,

starting from10 cm of displacement

Thirteen interferograms were generated using X-band SAR data

from TerraSAR-X in intervals over the period from June 10 to

November 22, 2015 The interferometric data parameters are

summarized in Table 4 The average coherence of the X-band

interferograms ranges from 0.54 to 0.66 Phase fringes due to

Table 1

ALOS SAR data characteristics.

Table 2

TerraSAR-X SAR data characteristics.

Table 3

ALOS interferometric data parameters.

Acquisition date Elapsed time

(d)

Baseline (m)

Average coherence

Fig 3 L-band cumulative vertical displacement map for the period from June 16 to December 17, 2010 (184 days).

Fig 2 L-band vertical displacement maps for the three periods.

Please cite this article in press as: Wempen JM, McCarter MK Comparison of L-band and X-band differential interferometric synthetic aperture radar for

surface displacement are identifiable in the majority of the

inter-ferograms, but in the areas with the largest magnitude subsidence,

the fringes are difficult to interpret Precise evaluation of the

max-imum subsidence magnitude is not attempted, but the subsidence

magnitudes can be reasonably estimated in many of the

interfero-grams.Fig 6shows an example of a filtered differential

interfero-gram for the period from June 10 to June 21, 2015 Phase fringe due

to subsidence are outlined in green There are at least seven fringes

indicating a maximum vertical displacement of more than 13 cm

and a maximum subsidence rate of more than 1 cm per day

5 Discussion

In the Wasatch Plateau, regions of subsidence can be identified

by both L-band and X-band DInSAR, but the effectiveness of

DInSAR for quantifying subsidence is dependent on the radar band

Generally, subsidence magnitudes are precisely measurable in the

L-band data The X-band data are more affected by signal

satura-tion and by temporal decorasatura-tion, and precisely measuring the

sub-sidence magnitudes using X-band data is more difficult Notably, though the data quality is variable, the L-band data and the band data have similar average coherence However, in the X-band data, coherence is spatially dependent: coherence is generally high over the valley floor and low in the vegetated subalpine region Variable surface conditions in the subalpine region con-tribute to both low coherence and significant phase noise in the X-band data, which make phases in the interferograms more diffi-cult to interpret

Spatial variation in the coherence of the L-band data is less apparent However, the L-band data are sensitive to significant changes in the surface conditions, and low coherence does affect the data quality Variable surface characteristics, including snow cover, likely caused low coherence in the L-band interferogram from September 16 to December 17, 2010 As a result of low coher-ence, the quality of the displacement map for this period ofFig 2c

is lower than the quality of the displacement maps for periods when the surface condition were more stable and the coherence was higher inFig 2a and b Additionally, in all of the L-band data, areas with very large deformation gradients are affected by low coherence due to phase saturation Although it is likely that the imaging period from June to December did not capture the full development of subsidence, phase saturation has the potential to cause subsidence to be underestimated by tens of centimeters in the L-band data The maximum cumulative subsidence reported for this area is on the order of 2 m[27]

In the X-band data, the interpretability of the phases is nega-tively affected by signal saturation as a result of large displacement rates and by temporal decorrelation, but the aerial extent of subsi-dence is clearly identifiable in most of the data Additionally, though the maximum subsidence magnitudes are difficult to pre-cisely measure, in most of the images the magnitudes of subsi-dence can be reasonably estimated Because the X-band imaging periods are shorter than L-band imaging periods, the X-band data provide a more timely report of the subsidence extent Conse-quently, short period X-band data has potential to accurately iden-tify periods when subsidence has ceased or is minimal

Acknowledgments

Funding for this research was provided by the National Institute for Occupational Health and Safety (NIOSH) The support of NIOSH

Fig 6 X-band filtered differential interferogram: June 10 to June 21, 2015 (11 days).

Fig 4 Time series subsidence profiles of section AA 0 from Fig 3

Fig 5 Time series subsidence profiles of section BB 0 from Fig 3

Table 4

TerraSAR-X interferometric data parameters.

Acquisition date Elapsed time

(d)

Baseline (m)

Average coherence

is thankfully acknowledged; however, the conclusions expressed in

this paper are those of the authors and do not represent the

opin-ions or policies of NIOSH Data for this research was provided by

the Alaska Satellite Facility, the Japan Aerospace Exploration

Agency and the German Space Agency The contributions of these

organizations are gratefully acknowledged

References

[1] Buckley SM Radar interferometry measurement of land subsidence Austin

(TX): The University of Texas at Austin; 2000

[2] Massonnet D, Feigl KL Radar interferometry and its application to changes in

the Earth’s surface Rev Geophys 1998;36(4):441–500

[3] Rosenqvist A, Shimada M, Watanabe M ALOS PALSAR: technical outline and

mission concept In: Proceedings of the 4th international symposium on

retrieval of bio- and geophysical parameters from SAR data for land

applications Innsbruck (Austria); 2004.

[4] Roth A TerraSAR-X science plan TX-PGS-PL-4001, vol 1; 2004 p 2–19.

[5] Chen BQ, Deng KZ, Fan HD, Hao M Large-scale deformation monitoring in

mining area by D-InSAR and 3D laser scanning technology integration Int J

Min Sci Technol 2013;23(4):555–61

[6] Carnec C, Delacourt C Three years of mining subsidence monitored by SAR

interferometry, near Gardanne, France J Appl Geophys 2000;43(1):43–54

[7] Yang CS, Zhang Q, Zhao CY, Ji LY, Zhu W Monitoring mine collapse by D-InSAR.

Min Sci Technol 2010;20:0696–700

[8] Ge L, Chang HC, Rizos C Mine subsidence monitoring using multi-source

satellite SAR images Photogram Eng Rem Sens 2007;73(3):259–66

[9] Ge L, Chang HC, Ng A, Rizos C Spaceborne radar interferometry for mine

subsidence monitoring in Australia In: Proceedings of first international future

mining conference and exhibition 2008 Carlton South (Victoria, Australia);

2008.

[10] Fan HD, Deng KZ, Ju CY, Zhu CG, Xue JQ Land subsidence monitoring by

DInSAR technique Min Sci Technol (China) 2011;21(6):869–72

[11] Ismaya F, Donovan J Applications of DInSAR for measuring mine-induced

subsidence and constraining ground deformation model In: Proceedings of

the GeoCongress 2012: state of the art practice in geotechnical engineering.

Reston (VA); 2012.

[12] Ng AHM, Ge L, Yan Y, Li X, Chang HC, Zhang K, et al Mapping accumulated

mine subsidence using small stack of SAR differential interferograms in the

Southern coalfield of New South Wales, Australia Eng Geol 2010;115(1–

2):1–15

[13] Perski Z The interpretation of ERS-1 and ERS-2 InSAR data for the mining

subsidence monitoring in Upper Silesian coal basin, Poland Int Arch

Photogram, Rem Sens Spat Inf Sci 2000;33:1137–41

[14] Perski Z, Jura D Identification and measurement of mining subsidence with SAR interferometry: potential and limitations In: Proceedings of 11th FIG symposium on deformation measurements Santorini (Greece); 2003 p 25–8 [15] Popiołek E, Krawczyk A Post mining deformation monitoring based on satellite radar interferometry (InSAR) In: Proceedings of the 3rd IAG/12th FIG symposium Baden (Austria); 2006 p 22–4.

[16] Wegmuller U, Spreckels V, Werner C, Strozzi T, Wiesmann A Monitoring of mining induced surface deformation using L-band SAR interferometry In: Proceedings of geoscience and remote sensing symposium, 2005, IGARSS 05 Seoul (Korea); 2005.

[17] Wright P, Stow R Detecting mining subsidence from space Int J Remote Sens 1999;20(6):1183–8

[18] Li XL, Liu DL, Song HJ, Chen RP, Li HY Key technology of D-InSAR at X-Band for monitoring land subsidence in mining area and its application J Electron 2014;31(5):441–52

[19] Zhao C, Lu Z, Zhang Q, Yang C, Zhu W Mining collapse monitoring with SAR imagery data: a case study of Datong mine, China J Appl Rem Sens 2014;8 (1):5946–57

[20] Liu ZG, Bian ZF, Lü FX, Dong BQ Monitoring on subsidence due to repeated excavation with DInSAR technology Int J Min Sci Technol 2013;23(2):173–8 [21] Rosen PA UNAVCO short course: principles and theory of radar interferometry In: Presented at InSAR: an introduction to processing and applications using ISCE and GIAnT Boulder (CO), August 4–6; 2014 [22] Baran I, Stewart M, Claessens S A new functional model for determining minimum and maximum detectable deformation gradient resolved by satellite radar interferometry IEEE Trans Geosci Remote Sens 2005;43(4):675–82 [23] Arabasz WJ, Burlacu R, Pankow KL An overview of historical and contemporary seismicity in central Utah In: Central Utah—diverse geology

of a dynamic landscape Salt Lake City (UT): Utah Geological Association; 2007.

p 236–53.

[24] Pankow KL, McCarter MK, Arabasz WJ, Burlacu RL Coal-mining-induced seismicity in Utah—improving spatial resolution using double-difference relocations In: Proceedings of 27st international conference on ground control in mining Morgantown (WV): West Virginia University; 2008 [25] Ellison L Subalpine vegetation of the Wasatch Plateau, Utah Ecol Monogr 1954;24(2):89–184

[26] Doelling HH Central Utah coal fields: Sevier-Sanpete, Wasatch Plateau, Book Cliffs and Emery Salt Lake City (UT): Utah Geological and Mineralogical Survey; 1972

[27] Monroe JK Subsidence report canyon fuel company, LLC SUFCO Salt Lake City (UT): Utah Division of Oil, Gas, and Mining, Department of Natural Resources;

2014 [28] DLR TerraSAR-X ground segment basic product specification document TX-GS-DE-3302 Wessling (Germany): DLR 2013.

[29] SARscape Ò User Guide Purasca (Switzerland): Sarmap; 2014.

[30] Richards JA Remote sensing with imaging radar New York: Springer; 2009

Please cite this article in press as: Wempen JM, McCarter MK Comparison of L-band and X-band differential interferometric synthetic aperture radar for