Spectral characteristics of asphalt road aging and ......Pavement Aging Asphalt pavements consist of rocky components and asphalt mix (or hot mix or bitumen). The mineral con-stituents

Spectral characteristics of asphalt road agingand deterioration: implications forremote-sensing applications

Martin Herold and Dar Roberts

We integrate ground spectrometry, imaging spectrometry, and in situ pavement condition surveys forassessment of asphalt road infrastructure. There is strong spectral evidence for asphalt aging anddeterioration. Several spectral measures derived from field and image spectra correlate well with pave-ment quality indicators (e.g., a pavement condition index). However, the potential for mapping is limitedby fine spatial resolution requirements (as fine as 0.5 m) and by the spectral confusion between pavementmaterial aging and asphalt mix erosion on the one hand and structural road damages (e.g., cracking) onthe other. © 2005 Optical Society of America

OCIS codes: 300.6340, 110.3080, 300.6190.

1. Introduction

The quality standards for transportation infrastruc-ture have evolved considerably over the past threedecades. The cost of frequent, comprehensive inspec-tion is high, and many jurisdictions limit their sur-veys to major roads, with minor roads surveyed in3-year cycles. For this purpose, a number of surveytechnologies have been applied to road-conditionmapping. The common practice today is extensivefield observations by experts who characterize apavement condition index (PCI) and a structure in-dex (SI) based on established physical parameterssuch as cracking, rutting, and raveling. Other tech-nologies are evolving; they include the application ofpavement management systems typically coupledwith Global Positioning System receivers and geo-graphic information system technology and withsemiautomated in situ pavement health surveys fa-cilitated by vans.1 This technology produces detailedand georeferenced road condition reports, with PCI

ratings for every �10 m of road. Nevertheless, suchsurveys remain expensive and challenging, althoughcost and safety considerations require that they bemade at regular intervals.

Previous studies of pavement condition mappingby remote sensing are rare. Early studies in the1970s dealt with the visual interpretation of large-scale aerial photographs to map physical surfacedistresses (e.g., cracks).2 Results showed that dis-tresses are distinguishable but only on detailedmap scales. Recent advances in imaging spectrom-etry provide the ability to derive physical and chem-ical properties of materials at a detailed level.3Consequently, one would raise the question: Can wemap road surface conditions with imaging spec-trometry? The utility of imaging spectrometry fortransportation studies was discussed previously,but a generic study of the spectral effects of roadsurface aging and deterioration is lacking.4,5 Thereis evidence that road properties such as aging andmaterial composition result in distinct spectralcharacteristics.6 However, it is unclear which spec-tral properties are associated with specific road sur-face characteristics and what remote-sensing dataconfigurations are needed to map such phenomena.

An experiment was conducted in the vicinity ofSanta Barbara and Goleta, California, to investigatethe effects of road conditions on pavement spectraand evaluate the utility of imaging spectrometry formapping road conditions. A comprehensive spectrallibrary of road surfaces and distress types was ac-quired with an Analytical Spectral Devices (ASD;

M. Herold ([email protected]) is with the European Space Agen-cy’s Global Observation for Forest and Land Cover DynamicsProject Office, Department of Geography, Friedrich-Schiller-Universität Jena, Loebedergraben 32, 07743 Jena, Germany. D.Roberts is with the Department of Geography, University of Cal-ifornia, Santa Barbara, Ellison Hall, Santa Barbara, California93106.

Received 14 December 2004; revised manuscript received 13March 2005; accepted 14 March 2005.

0003-6935/05/204327-08$15.00/0© 2005 Optical Society of America

10 July 2005 � Vol. 44, No. 20 � APPLIED OPTICS 4327

Raj Bridgelall

Highlight

Boulder, Colorado; full-range) hand-held spectrome-ter. The roads were imaged with a HyperSpectir(HST; Spectir, Inc., Goleta, California) submeter sen-sor. For field reference, roads were surveyed with anautomatic road analyzer (ARAN; Roadware Corpora-tion, Ontario, Canada) survey vehicle that acquires amultitude of in situ road-condition parameters.Based on this comprehensive database, we investi-gated the spectral properties of road surface dis-tresses and evaluated the potential of remote-sensingdata for discriminating them. The ultimate goal wasto explore relationships between remotely sensedparameters (i.e., spectral reflectance) and road-condition parameters such as the PCI. This relation-ship needs to be established if remote sensing is tosupport pavement health surveys.

2. Data and Methods

A. Study Area

The study focused on several urban roads in the Go-leta, California, area, located 170 km northwest ofLos Angeles in the foothills of the California CoastRange. Two roads with asphalt pavement, FairviewAvenue and Cathedral Oaks, were of main interest.Both are major urban roads with four lanes, two ineach traffic direction. Most of Cathedral Oaks is wellmaintained and reflects good to very good road con-ditions. Parts of this road were resurfaced just beforethe study. Only the western part of this road has fairto poor conditions. In contrast, pavement on FairviewAvenue is in particularly poor condition. Along thisroad, the central divider is the boundary between twotraffic management zones, in which rehabilitation ef-forts are funded and coordinated by two differentinstitutions, resulting in delays and failure of neces-sary maintenance. Continued deterioration of theroad surface was apparent during the study.

B. Road-Condition Data

Road-distress surveys provide information on thevarious distress types, their location, severity, andextent.1,7 The pavement evaluation is performed insitu, and the road quality information is aggregatedinto a PCI and a SI. The PCI is a single road perfor-mance indicator aggregating a variety of individualroad-condition characteristics with a scale usuallyfrom 0 and 100, with 100 being the best condition.8The SI works in a similar manner. The only differ-ence is that it includes only distresses that are re-lated to the structure, e.g., only alligator, block, andtransverse cracking.8

In this study, an in situ survey technique provideddetailed information about road distresses and theirspatial distributions. The observations were per-formed in December 2002 with the ARAN systemmounted on a specially modified vehicle that housesan extensive set of computers and sensors includinglasers, inertial measurement units, accelerometers,ultrasonic transducers, digital cameras, and otheradvanced technology subsystems. Global PositioningSystem receivers on the vehicle and at a base station

ensure accurate locational data. The survey providesgeocoded road-distress information on more than 30individual parameters aggregated for 50 m road seg-ments. The individual road-distress measures werecombined into a PCI and a SI available for each roadsegment. All road-condition information was inte-grated into a spatial database by use of the ARCGIS(ESRI, Redlands, California) geographic informationsystem.

C. Asphalt Road Spectral Library

In February 2004, a ground spectra acquisition cam-paign was conducted in the study area. Groundspectra were acquired with an ASD full-range spec-trometer. Full-range field spectrometer data arewidely used and considered to provide high-qualityspectral measurements. The full-range spectrometersamples the spectral range from 350 to 2400 nm. Theinstrument uses three detectors that span visible andnear-infrared (VNIR) spectra and shortwave infra-red (SWIR1 and SWIR2) spectra, with spectral sam-pling intervals of 1.4 nm for the VNIR detector and2.0 nm for the SWIR detectors. The measurementswere taken within 2 h of solar noon. Spectra of insitu materials were acquired from a height of 1 m byuse of a bare fiber optic, with a field of viewof 22° �0.147 m2 at a height of 1 m�. The road mate-rials were measured in sets of five spectra for eachfield target. Four to six sets of spectra were bracketedby measurements from a Spectralon (Labsphere,North Sutton, New Hampshire) 100% reflective stan-dard to convert the signals to reflectance values. Allspectra were inspected for quality, and suspect spec-tra were discarded. All targets were documented andintegrated into a spectral library. This library wasacquired as part of the National Center for RemoteSensing in Transportation at the University of Cali-fornia, Santa Barbara, and is available online forresearch purposes from http://www.ncgia.ucsb.edu/ncrst/research/pavementhealth/urban/road_spec.htm.Some spectra show sensor-specific features withsmall-scale variations. High-frequency noise from1100 to 1150 nm results from the transition betweenthe VNIR and SWIR-1 detectors. Major water-vaporabsorption bands (1340 to 1480 and 1770 to 1970 nm)were excluded from the analysis. Other sensor-induced spectral variations relate to the somewhatnoisy signal in the SWIR region beyond 2300 nm,particularly evident in low-reflectance targets. Themagnitude of the noise is less than 1% reflectance inany case and does not hinder the interpretations ofspectral characteristics from this spectral library.

D. Remote Sensing Data

Imaging spectrometry observations were providedfrom Spectir, Inc. The new HST sensor samples 227bands from 450 to 2450 nm. The main advantage ofthe HST is its fine spatial resolution. An onboardintegrated stabilization system enables a very lowflight altitude to produce a ground instantaneousfield of view of �0.5 m. Related investigations have

4328 APPLIED OPTICS � Vol. 44, No. 20 � 10 July 2005

shown that a spatial resolution of 4 m ground reso-lution was insufficient to produce good results in com-paring spectral data and pavement conditions.9Despite the spatial advantages of the HST data, thespectral calibration of the data was limited, andonly the bands from 450 to 900 nm covered by thefirst spectrometer of the HST could be used for theanalysis.

The analysis of the remote-sensing data considersonly the actual visible road surfaces. We excluded allother land cover types by using an accurate roadoutline that is available in digital format. Vegetationand shadows obscuring the road surface were ex-cluded from the HST data when they were clearlyvisible. The remote-sensing investigations applied aspecific reflectance difference that is discussed below.The local variance in the omnidirectional 3 � 3 pixelneighborhood of the band difference was also calcu-lated and is included in the analysis.

3. Spectral Analyses

A. Spectral Properties of Asphalt Road Surfaces andPavement Aging

Asphalt pavements consist of rocky components andasphalt mix (or hot mix or bitumen). The mineral con-stituents of the crushed stone rocky components canvary depending on the geological region, but the usualmajor components in the aggregate are dominated bySiO2, CaO, and MgO.10 The asphalt mix is a complexsubstance that can vary in composition depending onthe source of the crude oil and on the refining process.The chemical nature essentially is a mix of hydrocar-bons with 50 to 1000 carbon atoms plus enough hydro-gen, oxygen, sulfur, and nitrogen substituents to givesome of the molecules a polar character. More specifi-cally, the chemical components of asphalt mix arecarbon �80%�87%�, hydrogen �9%�11%�, oxygen�2%�8%�, nitrogen �0%�1%�, sulfur �0.5%�1%�, andsome trace metals.11,12

Diagram A in Figure 1 presents spectral samplesfrom the ground spectral library of pure road asphaltwith no obvious structural damages or cracks. Theage of the pavement, the PCI, and the SI are shown,with photographs of each surface. Spectrum A showsa recently paved road. The surface is completelysealed with asphalt mix. The spectral reflectance isgenerally low, and hydrocarbon constituents deter-mine the absorption conditions. The minimum reflec-tance is near 350 nm, with a linear rise toward longerwavelengths. Hydrocarbon compounds exhibit elec-tronic transitions that arise from excitation of bond-ing electrons in the UV and the visible, causing thisstrong absorption. The absorption is broad, and thereare no individual resolvable absorption bands in thisspectral region because of the complex nature of bi-tumen. A decrease in absorption strength towardlonger wavelengths results in a broad overall reflec-tance increase toward longer wavelengths, as is alsoseen in coals, oil shales, and chars.13

At longer wavelengths, spectrum A exhibits someobvious organic absorption features in the SWIR.

Fundamental absorption bands include an aromaticC—H stretch, symmetric and asymmetric stretchesand bends of CH3 and CH2 radicals, the carbonyl–carboxyl C—O stretch and aromatic carbon stretch,and numerous combinations and overtones.13 A lowoverall reflectance suppresses most of the distinctfeatures except the most prominent ones near1700 nm and from 2200 to 2500 nm. Various C—Hstretching overtones and combination bands domi-nate the feature in the 1700 nm region. When thisfeature is well developed it is asymmetric and reflectsa doublet, with the strongest absorption at 1720 nmand a second, less deep one at 1750 nm. The regionfrom 2200 to 2500 nm is affected by numerous over-lapping combination and overtone bands.13 Thiscauses the reflectance to decrease substantially be-yond 2200 nm, with the strongest absorptions locatedin the 2300 nm region, including a well-developeddoublet at 2310 and 2350 nm with the 2310 nm fea-ture usually deeper.

Spectrum C shows an old, deteriorated road sur-face. The photograph of the surface shows that theasphalt seal is widely eroded and the remaining as-phalt mix has undergone aging. The natural aging ofasphalt is caused by reaction with atmospheric oxy-gen, photochemical reactions with solar radiation,and the influence of heat, and it has three majorresults14: loss of oily components by volatility or ab-sorption, changes of composition by oxidation, andmolecular structuring that influences the viscosity ofthe asphalt mix (steric hardening). The loss of oilycomponents is relatively short term; the other two arelonger-term processes. With the erosion and aging ofthe asphalt mix the road surface is less viscous andmore prone to structural damages such as cracking.

The spectral effects represent a mixture of expo-sure of rocky components and asphalt aging. The lossof complex hydrocarbons causes a general increase inreflectance in all parts of the spectrum. This differ-ence is highest in the near IR and the SWIR, result-ing in an increase of more than 10% reflectance. Theelectronic absorption in the VNIR reflects the domi-nance of minerals and results in a concave shape withdistinct iron oxide absorption features that appear at520, 670, and 870 nm. The typical SWIR hydrocarbonabsorption properties near 1700 and 2300 nm vanishfor older road surfaces and are replaced by mineralabsorptions. For example, there is a significantchange in slope at the transition from hydrocarbon tomineral absorption. For older road surfaces the slopeincreases from 2120 to 2200 nm as the 2200 nm sil-icate absorption becomes more prominent. The slopeis higher for new pavement materials at 2250 to2300 nm, which correlates with the intensity of the2300 nm hydrocarbon feature.

Spectrum B represents road pavement of interme-diate age and condition. The surface exhibits both theasphalt mix and exposed minerals. The spectral char-acteristics reflect this intermediate stage by showingabsorption features from hydrocarbons and minerals.The intensity and characteristics of the features areless distinct than for pure very new and very old road


Fig. 1. Spectral effects of the asphalt surface characteristics of aging from the ASD ground spectral measurements (the major water-vaporabsorption bands are interpolated). A, asphalt aging and erosion; B, structural road damages (cracks); C, structural road damages(raveling); D, pavement maintenance.


surfaces. Aging and deterioration are gradual, andthere is strong spectral evidence that this transitionin surface material properties can be mapped by im-aging spectrometry. However, it should also be notedthat the spectral signal is more sensitive to the earlystages of aging and deterioration than it is for laterstages. For example, the spectral differences between1- and 3-year-old roads and a road that is 3 to 10 ormore years old were approximately equivalent, al-though the PCI and SIs showed significant changes;i.e., the PCI decreased from 100 to 86 between 1 and3 years and from 86 to 32 between 3 to 10 years ormore. The structure index decreased from a constantof 100 between 1 and 3 years to 100 to 63 over thesame age range.

B. Spectral Properties of Typical Asphalt RoadDistresses and Surface Maintenance

The most common road distress and indicator ofpavement quality is cracking. Diagram B in Fig. 1shows the spectral effects of structural damages orcracks with different degrees of severity. The mainspectral effect of cracking is a decrease in brightnessin all on parts of the spectrum. The increasing surfaceroughness causes shadows and reflectance decreasesof as much as 7%–8% in the near IR and the SWIRbetween the unshadowed pavement and high-severitycracks. The concave shape in the VNIR is more obvi-ous for brighter, noncracked road pavements. Thereis also a subtle indication that the cracked surfaceshave more-intense hydrocarbon absorption featuresin the 1700 and 2300 nm regions. The asphalt mixerosion and oxidations occur on the road surface.Cracking exposes deeper layers of the pavement withhigher concentrations of the original asphalt mix,which are then manifested as an increase in the ex-pression of hydrocarbon absorption features. Thisfact highlights the contrary spectral signal betweenroad deterioration of the pavement itself (diagram A)and the severity of structural damages (diagram B).Whereas an aging road surface becomes brighterwith decreasing hydrocarbon absorptions, structuraldistresses cause decreased reflectance but greater ex-pression of hydrocarbon absorptions. Although thedifferences in reflectance and intensity of the hydro-carbon absorptions are less for cracks than for newasphalt surfaces, this fact indicates some limitationsin spectrometry of road conditions.

A second common road distress is raveling. Rav-eling describes the progressive dislodgement ofpavement aggregate particles. Spectra of normalpavement are compared with a raveled road surfacein diagram C (Fig. 1). The spectrum with ravelingexhibits larger amounts of rocky components andraveling debris (gravel) on the surface. This results ina general increase in the brightness of the surfacecaused by increasing mineral reflectance and less-prominent hydrocarbon absorptions. The ravelingspectrum shows characteristics from both the normalpavement and Spectrum C. Spectrum C reflects agravel parking lot surface. Compared with those ofthe pavement, the gravel surface has a higher reflec-

tance in the visible and photographic near IR becauseof the missing hydrocarbon absorptions. The mineralcomposition is reflected in more-prominent featuresfrom iron oxide and other minerals, such as calcitefeature near 2320 nm.

Besides rehabilitation treatments there are severalmaintenance methods to improve and maintain thequality of road surfaces. Their spectral characteris-tics are compared to those of a common asphalt roadsurface (diagram D in Fig. 1). Spectrum I shows aslurry crack seal that helps to prevent water or othernoncompressible substances such as sand, dirt, rocks,and weeds from entering cracks. Slurry seal crackfillings are mixtures of emulsified asphalt or rubber-ized asphalt that are spread with a machine onto theasphalt surface. This treatment material has aconstant low reflectance of the order of 5% reflec-tance. Only minor hydrocarbon absorption features,similar to those found for parking lot surfaces, arerepresented.6

Patches are used to treat areas of localized roaddistress. The material is similar to fresh pavement,and Spectrum J has similarity to a newly paved road(Spectrum A). Chip seal treatments include sprayingan asphalt binder on the pavement and then imme-diately covering the binder with a single layer of uni-formly sized chips. The new surface treatment is thenrolled to seal the aggregate and broomed to removeany loose chips. Chip seal spectrum K has signifi-cantly higher reflectance than an untreated asphaltroad surface with more-prominent mineral absorp-tion features, similar to a raveled road surface(Spectrum G).

4. Remote-Sensing Data Analyses

The spectral interpretations of the road surfaces sug-gest several features that have utility for spectralidentification of road aging and deterioration. Thegoal of the remote-sensing image analysis was toevaluate the generic spectral understanding of as-phalt road conditions in a spatial mapping context.Given the fact that imaging spectrometers typicallyare coarser than spectral resolution, the image anal-ysis focused on a band difference in the visible region.The difference describes the spectral difference be-tween bands at 830 and 490 nm (VIS2 difference� �830 � �490). This measure was selected becausehigh-quality HST data were limited to a range of 450to 900 nm. The band at 830 nm reflects a spectralpeak between two iron absorption bands; 490 nm islocated in the middle of an iron absorption band. Thespectral difference between the two bands empha-sizes the increasing spectral contrast between roadsurfaces dominated by hydrocarbon absorptions (newroads) and mineral signals (older and deterioratedroads), which show increasing brightness and achange toward a more-concave spectral shape in theVNIR for older roads (Fig. 1). This band difference islow for new asphalt surfaces and increases with ageand level of deterioration, partially caused by in-creased iron oxide absorptions.

Figure 2 shows the VIS2 band difference calcu-


lated for the HST sensor, showing expected patternson surface roads. Roads with recently paved surfaces(upper left in Fig. 2; east Cathedral Oaks) show lowdifference values, of the order of 2%–4% reflectance(dark green and blue) between 830 and 490 nm. Thehighlighted part of west Cathedral Oaks presents thetransition between a newer split-seal refurbishedroad (green–yellow) and an older part with severalsealed transverse cracks (yellow–orange). The south-ern part of Fairview Avenue shows an area with se-vere alligator and block cracking. The crack patterns

are visible in the VIS2 difference and match expectedpatterns from field spectra. Such surface character-istics add spatial variance to the difference valuesand appear for roads with structural damage.

The HST VIS2 difference values show a significantcorrelation with the Roadware PCI and SI values(Fig. 3). Both regression relationships are statisti-cally significant at the 0.0001 probability level. Therelationship is quite distinct for roads in good condi-tion because the scattering of values is quite low. Thevariability increases for high difference–low PCI val-

Fig. 2. Spatial distribution of the VIS2 difference derived from the HST data.

Fig. 3. Comparison of the HST (HyperSpectir) VIS2 difference and the Roadware PCI and SI.


ues. The spectral signal of the roads in poorer condi-tion become more complex, and the relationship is notso obvious. This result agrees with observations fromthe field spectra. In situ material aging and asphalt-mix erosion cause an increase in the VIS2 difference.In contrast, structural road damages such as crackscan reverse this spectral gradient. The VIS2 differ-ence decreases if more numerous and more severecracks are present (Fig. 1) and make a deterioratedroad surface look more like a newly paved surface.

However, Fig. 2 also shows that structural roaddamage adds spatial heterogeneity to the road sur-face, given the 0.5 m spatial resolution of the HSTdata. A second remote-sensing measure, the spatialvariance of the VIS2 difference in the 3 � 3 pixelneighborhood, was investigated and compared to theRoadware SI (Diagram B, Fig. 3). This relationshipshows that structural road damage adds spatial com-plexity to the spectral signal. Diagram B highlights asignificant amount of variability for high SI values.This effect is caused by false positives, basically otherfeatures that add variability to the spectral signaland do not represent road damage such as shadowsfrom light poles, cars, and street paint. Despite thisvariability, the relationship between SI and the VIS2variance is significant and emphasizes that the boththe spectral signal and the spatial variance are im-portant indicators of road damage in hyperspectralremote-sensing data.

5. Conclusions

In this study we combined field spectrometry, in situroad surveys, and imaging spectrometry to gain aspectral understanding of asphalt road aging and de-terioration and to explore the possibility of mappingthese conditions by using image spectrometry. Forthe first time we were able to provide spectral evi-dence of the aging and degradation of in situ asphaltpavements. New asphalt pavements are dominatedby hydrocarbon absorptions. Pavement aging anderosion of the asphalt mix results in a gradual tran-sition from hydrocarbon to mineral absorption char-acteristics, with a general increase in brightness andchanges in distinct small-scale absorption features.Structural road damage (e.g., cracks) indicates asomewhat contrary spectral variation. Cracking de-creases the brightness and emphasizes hydrocarbonabsorption features.

The study used a band difference in the visibleregion and the local spatial variance to compare theremote-sensing signals to in situ pavement perfor-mance observations. The reflectance differences ofthe HST sensor data were compared to the PCI �R2

� 0.63� and showed that imaging spectrometry hasthe potential for representing road conditions. Thelocal variance correlated �R2 � 0.55� with a SI de-scribing structural road damage (e.g., cracks). Imag-ing spectrometry was successfully used to map roadsin good condition. Older roads in poorer condition aremore complex, and the map quality was lower. Thereare several reasons for this. The spectral effects of

asphalt aging are more sensitive to early stages ofasphalt deterioration. Furthermore, material agingand asphalt mix erosion on the one hand and struc-tural road damage on the other hand have oppositespectral effects. This results in less-clear spectral ev-idence for pavement quality. A measure of local vari-ance further represents the spatial cracking patternas an additional indicator of pavement condition.However, these measures are sometimes confoundedby false positives.

This study provided a first investigation of thistopic. It should be mentioned that the focus was on asmall study area and on asphalt road surfaces. Therewere some problems with the spectral calibration ofthe high-resolution HST data that limited the spec-tral remote-sensing analysis to a VNIR band differ-ence. With better-calibrated data it should bepossible to explore other techniques that include theSWIR, small absorption features (continuum re-moved algorithms), and characteristics of spectralshape (match filter analysis) that were identified inthe spectral library analysis, as well as improvedspatial pattern analysis techniques. It would be ofinterest to extend spectral investigations to concreteroad surfaces. Previous studies have shown that con-crete road aging, unlike asphalt aging,6 results indecreasing reflectance. Ultimately, the remote-sensing measurements should be compared to fieldexpert observations of road quality and managementsuggestions to make this technology valuable totransportation asset managers.

The ASD field spectrometer was kindly suppliedby the Jet Propulsion Laboratory. The authors ac-knowledge the support of the Research and Spe-cial Programs Administration, U.S. Departmentof Transportation, grant OTA #DTRS-00-T-0002(NCRST-Infrastructure). Essential for this studywere vehicle in situ inspection data and the high-resolution hyperspectral data provided by the com-panies Roadware (www.roadware.com) and SpecTir(www.spectir.com), respectively; their support isgreatly appreciated. The authors thank the Califor-nia Department of Transportation, Val Noronha, P.Dennison, M. Gardner, D. Prentiss, J. Schuhrke ofthe University of California, Santa Barbara, and O.Smadi and R. Souleyrette of the Center for Transpor-tation Research and Education, Iowa State Univer-sity, Ames.

References1. R. Haas, W. R. Hudson, and J. Zaniewski, Modern Pavement

Management (Krieger, 1994).2. E. G. Stoeckeler, “Use of aerial color photography for pavement

evaluation studies,” Highway Res. Record 319, 40–57 (1970).3. R. N. Clark, “Spectroscopy of rocks and minerals and principles

of spectroscopy,” in Manual of Remote Sensing, A. N. Rencz, ed.(Wiley, 1999), Chap. 1, pp. 3–58.

4. R. B. Gomez, “Hyperspectral imaging: a useful technology fortransportation analysis,” Opt. Eng. 41, 2137–2143 (2002).

5. J. Usher, and D. Truax, “Exploration of remote sensing appli-cability within transportation,” final projects rep. (Remote Sens-ing Technologies Center, Mississippi State University, 2001),


http://www.rstc.msstate.edu/publications/99-01/rstcofr01-005b.pdf (access: March 2004).

6. M. Herold, D. A. Roberts, M. Gardner, and P. Dennison, “Spec-trometry for urban area remote sensing-Development andanalysis of a spectral library from 350 to 2400 nm,” RemoteSens Environ. 91, 304–319 (2004).

7. J. S. Miller and W. Y. Bellinger, “Distress identification man-ual for the long-term pavement performance program,” 4threvised ed., Rep. FHWA-RD-03-031 (Federal Highway Admin-istration, 2003), http://www.tfhrc.gov/pavement/ltpp/reports/03031/03031.pdf (access: March 2004).

8. U.S. Army Corps of Engineers, “Technical manual on pave-ment maintenance and management,” document TM 5-623(1982), http://www.usace.army.mil/publications/armytm/ (ac-cess: March 2004).

9. M. Herold, D. A. Roberts, O. Smadi, and V. Noronha, “Roadcondition mapping using hyperspectral remote sensing,” pre-sented at the 2004 AVIRIS Workshop, Pasadena, Calif., 31March–2 April 2004.

10. T. L. Robl, D. Milburn, G. Thomas, K. O’Hara, and A. Haak,“The SHRP materials reference library aggregates: chemical,mineralogical, and sorption analyses,” Strategic Highway Re-search Program (SHRP) Publ. SHRP-A/UIR-91-509 (NationalResearch Council, Washington, D.C., 1991), http://gulliver.trb.org/publications/shrp/SHRP-91-509.pdf (access: September2002).

11. Asphalt Institute, The Asphalt Handbook, No. 4 of ManualSeries (Asphalt Institute, 1989).

12. Y. H. Huang, Pavement Analysis and Design (Prentice-Hall,1993).

13. E. Cloutis, “Spectral reflectance properties of hydrocar-bons: remote-sensing implications,” Science 4914, 165–168(1989).

14. A. Bell, “Summary report on the aging of asphalt-aggregatesystems,” Strategic Highway Research Program (SHRP) Publ.SHRP-A-305 (National Research Council, Washington,D.C., 1989), http://gulliver.trb.org/publications/shrp/SHRP-A-305.pdf (access: March 2004).


Spectral characteristics of asphalt road aging and ......Pavement Aging Asphalt pavements consist of rocky components and asphalt mix (or hot mix or bitumen). The mineral con-stituents

Documents