Three-Dimensionallntegration of RemotelySensed Geological Data: A Methodologyfor PetroleumExploration* Abstract scribed through a seriesof non-quantitative,idealized, so The standard approach to integration of satelliteimagery block diagrams.But this approachdoes not get the most in- and sub-surfacegeological data has been the comparison of formation out of the data, and may result in unrealistic mod- a map-view image interprctation with a selection of sub-sur- eling of geologicalfeatures such as fault geometriesand lithologic boundaries. face cross-sections.The relationship between surface and subsurfacegeologr can be better understoodthrough quanti- This study teststechniques to integratea ro digital ter- tative thrce-dimensional (sn) computer modeling. This study rain model with sn subsurfaceinterpretations in order to teststechniques to integrate a so digital terrain model with construct quantitative models of the geology.In particular, it aims to demonstratemethods that will be beneficial in inter- 3D sub-surface interpretations.Data types integrated,from a portion of the Paradox Basin, southeast Utah, include Land- pretation of imagery data in frontier exploration areaswhere sat TMimagery, digital elevation data (ntu), and sub-surface few sub-surface data are available. The key objectives are gravity and magnetic data. Combined modeling of basement o To gain a betterunderstanding of the relationshipbetween and topographicfeatures suggeststhe baditional lineament subsurfacegeology and surfacegeology as rnterpretedftom analysis approach to structuruI interpretation is over-sim- satelliteimagery; and plistic. Integration of otu and image data displayed in sn . To testif, and by how much,the three-dimensionalmethods facilitatethe understandingof sub- proved more effective litholog discrimination than a developedin this study for surfacegeology. map-view approach. Automated strike and dip interpretation algorithms require DEMdata at resolutionsbetter than 70 m This study is multi-disciplinary in the sensethat it incorpo- by 90 m. The methodologr testedis beneficial to interpreta- rates concepts and methods from the fields of remote sens- tion of imagery data in ftontier exploration areas. ing, geographic information systems, subsurface geology, digital tenain modeling, computer mapping, and three-di- lntroduction mensionalmodeling to establisha methodologyto maximize Integration of remote sensing and digital terrain modeling the interpretation of the geological information. has been demonstrated to produce significant gains, and it is a well-establishedtechnique in areassuch'as environmental StudyArea impact assessment(Baker, 1991J, landslide hazard assess- The Blanding sub-basinof the ParadoxBasin in southeast ment (Huangand Chen, 1991),and surfacemineral explora- Utah was selectedto test the methodologyas it is a proven tion (Braux et o1.,1990), Software has been developedthat hydrocarbon province with plgntiful subsurface data and combines image data with digital elevation models (onv) and good surfaceexposures. The setting and a summary of the re- producesthree-dimensional surface displays with perspec- gional structural featuresare shown in Figure 1. Structural tive viewing (e.9.,Duguay et o/., 1gB9;McGuffie ef o/., 1989; relief on Mesozoicreference horizons is approximately6,000 Morris, 1991).In addition, the value of integratingremote feet from the baseof the Blanding Basin to the crest of the sensingwith geographicinformation systems(cts) has been Monument Upwarp. demonstratedand softwarepackages are available(e.g., Bult- The Paradox Basin was formed during Pennsylvanian man and Getting,1991; Runsheng et al., tgst). time as a transtensionalbasin, with the main deformationoc- However,in the field of trying to deducesub-surface ge- curring along northwest-to southeast-trendingfault zones, ology from satelliteimagery, three-dimensional (so) model- seenin surfaceimagery as prominent lineamentssuch as the ing techniques,beyond the display of a single surfacein Nequoia-AbajoLineament (Knepper, 1982; Stevensonand perspective,have not been fully employedin an integrated Baars,1986). This pattern is complicatedby a perpendicular fashion. The standardapproach to combining sub-surface series of regional lineament zones trending northeast to data with imageinterpretation has been the comparisonof a southwest.Within the study areaare three major Iineaments map-view (zn) imageinterpretation with a selectionof sub- (Figure1). The Four CornersLineament, trending NW-SE, surfacecross-sections. The resulting synthesiswould be de- definesthe southern limits of the ParadoxBasin, and is a strike-slip fault with right-lateral displacement. The NE-SW- "Presented at the Ninth Thematic Conferenceon Geologic Remote trending Coconino Lineament is interpreted by Davis (1978) Sensing,Pasadena, California, 8-11 February1993. as a major partitioning element in the basement extending beyond the Paradox Basin. The Monument Upwarp and Comb Ridge Monocline are features along the Coconino Li- Photogrammetric Engineering & Remote Sensing, Vol. 59, No. B, August 1993,pp. 1251-1256. MichelleI,..JVI'K*":; 0099-1 1 1 2/9 3/5 I 08-12 5 1 $ 03.00/0 Department of Geology and Petroleum Geology, @1993American Societyfor Photogrammetry University of Aberdeen,Meston Building, Kings College, and RemoteSensing AberdeenABg ztIE, Scotland,United Kingdom. PE&R5 t2sl neament. The House Creek fault zone consists of several NE- SW-trending en echelon faults that suggestleft-lateral move- ment (Stevensonand Baars,1986), which is compatiblewith these being antithetic shear faults to the NW-SE-trending fault zones. A deep-seatedfault underlying the Comb Ridge monocline lies en echelon to the House Creek fault zone (Stevensonand Baars,1986). During Laramidetime (Late Cretaceousto Early Eocene),the Colorado Plateau was sub- jected to compressional forces, and large-scaleinversion structures, such as the broad upwarps and monoclines, were developed(Blakey and Gubitosa,1984; Peterson,1986; Blakey,19BB). \ .c ;;;( {. \ The areaof interest is between 110oto 109" longitude ''.\i \\>f\l\\ =*h;-\' and 37oto 38" latitude North, in southeastUtah. Pennsylvan- \ =\B\ / €<--- ian and Permian age rocks are exposed on the Monument Upwarp fFigures 2 and 3). Late Permian and early Jurassic iir.Kt*: rocks are exposed in Comb Wash. The Jurassic-ageWingate, Kayenta, and Navajo formations are exposed.on Comb Ridge. , N.s Morico \. ( A major surface topographic feature, on the easternboundary -R^./ r\i'"{iare''-^v --,/ of Comb Ridge, is the Abajo Mountains which are formed by -, 1 i Oligocenelaccolithic intrusions. Eastof Comb Ridge are /,." lt-\ \/ younger lrI01020304050', Jurassic and Early Cretaceousunits exposed in river / DEFIANCE , valleys and on the mesa cliffs. Major drainages within the study area include the San |uan River that tracks across the southern portion from eastto west. As the San |uan crosses Figure1. Locationmap showingregional structural features Comb Ridge and the Monument Upwarp, there is a classic and studyarea (afterStevenson and Baars,1986). goose-neckincised meander-looppattern associatedwith an- tecedence.Major north-south drainagesinclude Montezuma Creek, Recapture Creek, Cottonwood Wash, and Comb Wash (Figure 3). South of the San Juan River, surface exposuresof using two-dimensional GISoverlay techniques and three-di- JurassicMorrison Formation to Navajo Sandstoneare masked mensional surface comparison methods to identify the rela- by Recentaeolian deposits. tionship between surface topography and sub-surface basement structures. The co-registeredimage and nslvl data were used to test the usefulness of automatic strike and dip, Methods using software developed by researchersat the University College, University of London. Finally, the RAINDRoPsoft- DataSet ware developed by Morris (1991) was run on the Dnv data to A seven-bandLandsat Thematic Mapper (rv) image in digi- identify major drainages. tal format was used in this study for lithologic discrimina- To test the proposed methodology, three commercially tion. Two images were tested: a three-band decorrelated available software packagesare linked: Image ProcessingSys- principal components(rC) image (bands7,4,5,7 with PC1,2,3 tem from Erdas,CPSa mapping packagefrom Radian Inc., and displayedin Red, Green,Blue (RGB),respectively), and a the Stratigraphic Geocellular Modelling (sctvt)Package from stretchedfalse-color composite (bands 1,4,5 displayedin Stratamodel Inc. The Erdas System used in this studv is a RGB,respectively). The stretched false-color composite image personal-computer version mounted on a Compaq 386 under was the most useful in regional lithological interpretations. Dos. The cPs3 and SGMsoftware run on a SiliCon-Graphics The noise introduced in the PCimage by the equal weighting IRISUnix workstation. Network software was used to translate of decreasingPCs was accentuatedwhen the imagewas re- files between DoS and Unix formats. sampled to larger pixels, negating any spectral enhancement that this method produced at full resolution. Digital elevation data for this study were obtained from Results the USGSl-degree digital data set (1" DEMdata are used to map at a scaleof 1:250,000).These data have a pixel resolu- StructuralInterpretation tion of 70 bv 90 metres with a vertical accuracv of 30 m and The methods that are traditionally used for lineament analy- a horizontai accuracyof t30 m. The gravity and magnetic sis in remote sensingare summarizedby Drury {1987).The interpretationsare from Caseand