qualitative and quantitative photogrammetric techniques ...antonio zanutta, paolo baldi, gabriele...

1067

ANNALS OF GEOPHYSICS, VOL. 49, N. 4/5, August/October 2006

Key words photogrammetry – DTMs generation –archival photographs – surface representation – land-slide analysis

1. Introduction

The historical aerial photos are of fundamen-tal importance not only for qualitative analysis of

the territory (traditionally conducted by photoint-erpretation of stereoscopic models), but for aquantitative assessment as well. Appropriate com-parison of photogrammetric surveys of a zone re-alised in different years allows identification ofgeometric changes occurring during the time in-terval in question (differential photogrammetry,Anzidei et al., 2000; Baldi et al., 2002, 2005).

This technique has many fields of application:for example, the study of changes in urban areas,farming activities and evolution of coastlines(Dolan et al., 1980; Overton and Fisher, 1996),checking glaciers (Kääb and Funk, 1999) andmonitoring unstable slopes (Kääb, 2000).

Stereoscopic interpretation of two or moresets of aerial photographs taken over a sufficient-

Qualitative and quantitativephotogrammetric techniques

for multi-temporal landslide analysis

Antonio Zanutta (1), Paolo Baldi (2), Gabriele Bitelli (1), Mauro Cardinali (3) and Alberto Carrara (4)(1) Dipartimento di Ingegneria delle Strutture, dei Trasporti, delle Acque,

del Rilevamento, del Territorio (DISTART), Università degli Studi di Bologna, Italy(2) Dipartimento di Fisica, Università di Studi di Bologna, Italy

(3) Istituto di Elettronica e di Ingegneria dell’Informazione e delle Telecomunicazioni (IEIIT), CNR, Bologna, Italy

(4) Istituto di Ricerca per la Protezione Idrogeologica (IRPI), CNR, Perugia, Italy

AbstractThe results of two survey methods, geological photointerpretation and historical photogrammetry, are comparedin order to evaluate the temporal evolution of a unstable slope located in the Tuscan-Emilian Apennines (Italy).Historical aerial photos of the area, derived from photogrammetric surveys conducted in 1954 (scale 1:60000),in 1971 (scale 1:20000), and in 1976 (scale 1:17000) were available. A photogrammetric flight was further con-ducted in 2000, at a scale of 1:4400, with a traditional GPS ground survey support. First, the results of photo-graphic analysis with the photointerpretation method are presented: the landslides are described from a geolog-ical point of view, showing its temporal evolution. To quantitatively assess the landslide movements, Digital Ter-rain Models were generated by means of an analytical plotter and a digital photogrammetric workstation, withsemi-automatic and automatic procedures. To generate these products, it was necessary to solve problems relat-ed to a lack of data concerning the aerial cameras used for the historical flights (internal orientation) and the dif-ficulty identifying control points on the photos in order to define the external orientation. An unconventionalphotogrammetric methodology, based on identification of homologous points in zones considered outside thelandslide area, has been there developed and tested to insert the various surveys into a single reference system.

Mailing address: Dr. Antonio Zanutta, Dipartimento diIngegneria delle Strutture, dei Trasporti, delle Acque,del Rilevamento, del Territorio (DISTART), Università de-gli Studi di Bologna, Viale Risorgimento 2, 40136 Bologna,Italy; e-mail: [email protected].

1068

Antonio Zanutta, Paolo Baldi, Gabriele Bitelli, Mauro Cardinali and Alberto Carrara

ly long time span (20-30 years) is the most effec-tive and popular technique for the identificationand morphological characterisation of landslides(multi-temporal landslide mapping) (van Westenand Getahun, 2003). This procedure assumesparticular theoretical and applicative importancedue to the fact that most landslide motion tendsto take place within or very close to pre-existinglandslide deposits. Therefore, knowledge of thelocation, type and distribution of landslides oc-curring over time in a territory, is an essentialtool for forecasting future events.

It must nevertheless be pointed out that thereliability of stereoscopic interpretation strong-ly depends on the experience of the photo-inter-preter. Thus, when studying landslide phenom-ena, it becomes important to utilise more quan-titative techniques, such as analytic or digitalphotogrammetry, in association with photo-in-terpretation, which is intrinsically subjectiveand qualitative (Brunsden and Chandler, 1996;Lane et al., 1998).

Applicability of the two methods obviouslydepends on the availability of aerial photos ofthe territory, covering a sufficiently long timespan. In this regard, it should be rememberedthat the first Italian national scale photogram-metric coverage, with classic pseudo-verticalphotos (23×23 cm), was conducted in 1954-1955 by the Istituto Geografico Militare (IGM).These photos, even though at a scale (1:33000-1:60000), may be a fundamental source ofknowledge for the study of the temporal evolu-tion of slope instability (there are also other oldIGM surveys, at different scales, taken of indi-vidual portions of the territory).

Starting from 1970, in order to produce largescale maps (technical regional maps) and also inthe context of specific projects (for example,monitoring of river valleys, glacier faces, slopelandslides, etc.), many public administrationsmanaged the realization of local photogrammet-ric surveys.

To assess the potential offered by the integra-tion of qualitative and quantitative photogram-metric techniques in the study of landslides, weidentified as test area an unstable slope located inthe territory of the Municipality of Vergato, nearthe city of Bologna (Tuscan-Emilian Apennines,Italy) where in recent years a warning system

was implemented installing inclinometers andpiezometers, measuring pore water pressure andground displacements by means of wire exten-someters, GPS surveys, laser scanning and digi-tal photogrammetric techniques (Mora et al.,2003; Bitelli et al., 2004; Pesci et al., 2004; Si-moni et al., 2004; Dubbini and Zanutta, 2005).

This work presents the results of photo-in-terpretive analysis of four sets of aerial photos.Such analysis characterises the slope geo-mor-phologically, pointing out its temporal evolu-tion based on morphological changes identifiedin the available stereoscopic models.

A quantitative analysis of the landslidemovements was performed comparing multi-temporal DTMs produced with an analyticalplotter (DIGICART 40, Officine Galileo), and adigital photogrammetric workstation (HelavaSystem).

2. Photo interpretation investigations of the landslide phenomena

The territory of the Municipality of Vergatocontains numerous slopes affected by land-slides, many of which are currently active. Dueto its geological-geomorphologic structure andtypes of land use and settlement, this area maybe considered highly representative of themountainous territory of the Emilia-RomagnaRegion, for which recent studies claim that ofmore than 32000 landslides identified andmapped, approximately 26% are to be consid-ered «active» (Bertolini and Pellegrini, 2001).

Table I. Main characteristics of aerial photosutilised for production of multi-temporal map oflandslides in instudy area.

Agency Photo scale Type Year Month

GAI, IGMI 1:60000 B&W 1954 JuneEmilia-Romagna 1:20000 B&W 1971 -

RegionEmilia-Romagna 1:17000 Colour 1976 May

RegionAlifoto Srl 1:4400 B&W 2000 April

for ENEL Hydro

Qualitative and quantitative photogrammetric techniques for multi-temporal landslide analysis

A multi-temporal landslide map was pro-duced for this area by means of a detailed inter-pretive study of photos taken during four pho-togrammetric flights, from 1954 to 2000, withphotographic scale ranging from 1:60000 to ap-proximately 1:4400 (table I; fig. 1a-d).

The multi-temporal landslide map was pro-duced by means of careful photo-interpretiveexamination of all the available aerial photos,

1069

Fig. 1a-d. Multitemporal aerial photos details concerning landslide bodies (Municipality of Vergato, Bologna,Tuscan-Emilian Apennines, Italy; φ=44°17l25mN, ω=11°07l16mE): a) 1954; b) 1971; c) 1976; d) 2000.

This work examines the instability phenom-ena affecting the slope on the right bank of theReno River, below the Vergato-Piandisetta roadin the vicinity of Cà di Malta and Cà del Bosco(Bologna). This slope is composed of terrainspertaining to the Scaly Clay Complex: theirpoor mechanical properties make it one of themost unstable geological complexes in the en-tire Apennine territory.

a b

c d

1070


using a stereoscope with continuous 2-20Xzoom. Interpretation began with an analysis ofthe 1954 aerial photos, in which it was possibleto identify both large-scale landslides (mostlyold and relicts) and smaller landslides, general-ly more recent. The aerial photos from the oth-er flights were then analysed, separately and incombination with those from 1954, comparingsimilarities and differences in morphology ineach portion of the slope under study. The land-slides were classified on the basis of estimateddepth, type of movement (slip-flow and flow)and relative age (old or recent). This last param-eter was deduced from the degree of morpho-logical «freshness» that the instability present-ed in each set of photos.

The photo-interpreted landslides were firstredrawn on map at 1:4400 scale and then digi-tised and stored in a GIS for data display andanalysis.

Figure 2 shows the distribution of the land-slides identifiable on the 1954 photos. The old-est landslides may be classified as slip and slip-flow within which (as described below) numer-ous landslides identified in the flights conduct-ed from 1954 to 2000 have developed.

The main landslide is a deep slip, mostlyrelict, with a surface area of about 160000 m2

and average slope of 15°, extending from 320 ma.s.l, below the Vergato-Piandisetta road, untilthe Reno River to 185 m a.s.l. Over time, thislandslide underwent partial reactivations with

Fig. 2. Distribution of old or very old landslide bodies identified in 1954 aerial photos, classifiable as slip andslip-flow, with elements of activity.


movements of various magnitude that masked orcancelled some of the features of the main land-slide, in some cases making it difficult to identi-fy. The alimentation area, shown by a markedconcavity on the slope, is characterized by amain scarp whose arched shape is recognisableat various points, and continues for more than akilometre. Locally, the slope’s concavity is morecomplex, marked by later landslides that createdsudden changes in dip. The accumulation zoneshows general longitudinal and transverse con-vexity to the slope, which becomes more accen-tuated at the foot, where the Reno River wasslightly deviated toward the west for more than350 m. Two old slip-flows, with a characteristicnarrow, elongated accumulation shape whichthen fans out at the foot of the landslide, have de-veloped at the sides of this relict landslide.

The phenomenon that has developed on theright side of the main slip (hereafter called «Càdel Bosco») presents a complex and deep scarparea that reduces almost to a funnel around 225m of high. On the other hand, the deposit pres-ents a sharply lobed convex shape on whichseveral generations or pulsations of landslidesmay be noted, superimposed on the accumula-tion of the old main slip.

The landslide that developed on the left sideof the main slip (hereafter called «Cà di Mal-ta») presents less evident morphological char-acteristics. Its scarp area is not very accentuat-ed and is highlighted by a large concavity,while the deposit presents a marked lobe at thefoot of the landslide that further reduces theReno River channel by a few meters.

Figure 2 also shows distribution of the land-slides classified as slip and slip-flow, recentand/or active, traceable to around 1954. In thephotos, these landslides show clear evidence ofmorphological «freshness», demonstrating move-ments in evolution. These movements are of lowor medium magnitude, with extensions rangingbetween 1000 and 15000 m2. They are located al-most exclusively in the scarp areas of the Cà delBosco and Cà di Malta landslides, without everpushing down to the base of the slope.

The landslides which occurred during the pe-riod 1954-1971, are essentially two deep slip-flows extending from 40000 m2 to 70000 m2,classifiable as reactivations of the previous Cà del

Bosco and Cà di Malta landslides, within and atthe borders of which small landslides no largerthan 5000 m2 may be seen.

In the period 1954-1971, at least two gener-ations of movement may be seen for the Cà delBosco landslide. The first involved the upper-middle part of the previous instability, withoutreaching the base of the slope, whose foot ap-pears changed at several points, the left sidemore convex downhill and advanced by about20 m compared to its appearance in the 1954photo. The second was a reactivation of the pre-vious landslide. The scarp has receded by about100 m, reaching almost near the Vergato-Piandisetta road, while the deposit has reachedthe base of the slope, very probably blocking thevalley road. The activity of this movement isdemonstrated by an irregular surface withmounds, landslips troughs, and numerous ten-sion fractures open both longitudinally andtransversely. The medium-high section of theflow (from high 250 to 220 m) is winding, witha sharply depressed and concave transverse pro-file developing within the centre of the valley.As evidence of the rapid flow of material, alongthe sides of this section of the flow one mayclearly see levees that form crests a few metershigh, extending for over 200 m in length. Thefoot of the flow is clearly convex, with an irreg-ular surface lobed by pronounced bumps.

In the Cà di Malta landslide area, the insta-bilities occurring in the period 1954-1971 reac-tivated the entire old body with slip-flow move-ments, considerably reducing the scarp and ad-vancing its foot by about 20 m to within theReno River bed. Specifically, the border has re-ceded by over 50 m, touching the head of theCà del Bosco landslide. The main scarp hassteps of more than 10 m and extends continu-ously, mainly along the left side of the landslidefor about 200 m. The deposit has an irregularsurface, with a generally convex shape that hasbeen remodelled and regularised in the centresection as the result of drainage works withchannels and surface drains running rectilinearor transverse to the slope. As shown in the 1971photos, the foot of the landslide is limited byblocks aligned in the Reno River bed, seen forthe entire width of the landslide front at about20 m from the bank, which precisely in this sec-

1071

1072


tion is eroded and receded due to lateral under-mining. In addition, the road’s path has been al-tered along this section of the landslide, makingit straighter than it was in 1954.

The movements which occurred in the period1971-1976, identifiable in the colour aerial pho-tos taken in 1976, are mainly due to small flowsor slip-flows occurring in the centre channel ofthe Cà del Bosco flow or in the scarp area of theCà di Malta landslide. On the border of this lat-ter landslide, it is interesting to note the presenceof an anthropic intervention that caused about 20m of stripping and the creation of a rise of morethan 1000 m2 halfway up the hill. It is importantto point out that the most extensive reactivationbegan precisely at the site of such work, as seenon the 2000 photos.

Lastly, the landslides occurring between1976 and 2000 were essentially a total or partialreactivations of previous Cà del Bosco and Cà di

Malta landslides, with deep slip-flows measuringbetween 0.5 and 4.0 ha and small flows with sur-face area not exceeding 0.5 ha. The most evidentreactivations might be traceable to movementsoccurring in October 1996 following long, heavyprecipitation, or in November 1998 in the ab-sence of exceptional rains.

These photos show the Cà del Bosco land-slide reactivated at its top, with the formation ofa few slip-flows, the most evident of which isthe one formed from the right side of the scarpat 290 m high near a few farm annexes of Cà diSotto homes. The alimentation area of this land-slide is characterised by a sharp depression,bounded by a main scarp of about 20 m and bya series of secondary scarps and tiers. The de-posit is channelled in the medium-high sectionof the old 1971 flow, which is topographicallydepressed and concave, filling it with more than10 m of sediment (estimated thickness). The

Fig. 3. Three-dimensional perspective view of southern portion of slope under study, obtained by projecting the2000 digital orthophoto onto the corresponding 2×2 m DTM (heigh exaltation: 30%). The image represents avector/raster result which has metric characteristics, coming from photogrammetric restitution. The border of theCà di Malta landslide is indicated with a yellow line; red lines show local roads.


flow stops at 225 high, where it forms a topo-graphically pronounced and prominent lobe.

In the area of the Cà di Malta landslide (fig.3), the instabilities occurring in the period 1976-2000 reactivated most of the old 1971 body, withslip-flow and flow movements, considerablypulling back the scarp precisely at the site of theanthropic intervention observed during the 1976flight. At this point, the border has receded byabout 100 m, pushed back as far as the Vergato-Piandisetta road and creating, once again, a sharpdepression with a main scarp of over 20 m. Theother movements noted in the scarp area or at theborders of the Cà di Malta landslide also show adistribution of retrogressive activity, but withmore limited retrograding of about 20-30 m.These movements are clearly seen in the medi-um-high part of the slope, where they all meet ataround 200 m high. On the other hand, the medi-um-low sector of the slope has been heavily re-modelled by work done to reinforce the land-slide, with drainage and reprofiling of the slope.In these cases, man’s cancellation of shapesmakes it very difficult, if not impossible, to de-fine and map the real extent of the movement.Nevertheless, careful examination and compari-son of the last two aerial photo flights has pro-vided elements useful for defining and extendingthe landslide up to the Reno River. The hypothe-sis that the landslide may have reached the valleyis supported by bibliographic information (Mora et al., 2003) reporting partial obstruction of thewaterway following the reactivation taking placein October 1996.

3. Photogrammetric investigation of the landslide phenomena

In order to define the shape and position ofobjects or portions of land by means of pho-togrammetry, it is necessary to reconstruct thegeometric relationships on which the imageswere taken.

The cameras used in photogrammetry pro-duce photos, which, with sufficient approxima-tion, may be considered central perspectives ofthe photographed object, that is all the rays go-ing from the image to the object pass throughthe perspective centre.

The basic mathematical model used to con-vert from a two-dimensional system (photo) toa three-dimensional system (object) is ex-pressed by well-known colinearity equations(Kraus, 1997) expressing that the perspectivecentre, the object points, and the image pointsare lying on the same projective straight lines.

To transform the image coordinates in thecorresponding object coordinates, at least twophotos of the same object and the nine orienta-tion parameters internal (IOP), and external(EOP) are needed. As known, in traditional pho-togrammetry, IOPs are determined in laboratoryby means of camera calibration procedures;EOPs can be determined by spatial resection byusing known Ground Control Points (GCPs),visible on the photos, whose coordinates aremeasured in the image reference systems.

Orientation parameters defining the camerastation coordinates and the orientation anglesare often unavailable for historical aerial pho-tos. Therefore, to produce the 3D model of thearea, it is necessary to adopt an unconventionalapproach that gives results less precise than therigorous one. The aerial photos can be consid-ered taken by non metric cameras, and a self-calibration procedure can then be performed(self-calibrating bundle block adjustment),based on the availability of a sufficient numberof GCPs. Generally this method is not applica-ble because it is difficult to find enough GCPsmeasured at the same time of the flight.

Since large-scale maps of the zone are avail-able, one solution may be to detect the GCPsfrom them (Chandler et al., 1988a,b; Chandlerand Brunsden, 1995), even if a variety of prob-lems may be encountered: the maps may havebeen drawn from aerial photogrammetric sur-veys prior or subsequent to the one beinganalysed; it is difficult to assess the real error tobe assigned to the coordinates of the pointsmeasured on the map (for example, altitudes de-rived by interpolation from level curves); it isoften very difficult to identify equivalent points,i.e. the search for map GCPs on stereoscopicmodels.

The method adopted in this paper to solvethe orientation problem and to co-register themulti-temporal three-dimensional models, isbased on identification of a large number of

1073

1074


points, located outside the landslide bodies, andvisible on multi-temporal stereoscopic models.

As mentioned above, three sets of historicalaerial photos (from 1954, 1971 and 1976) plusthe large-scale photos obtained in 2000 with acalibrated aerial camera, are available for the areaunder study. Simultaneously with the 2000 flight,a large number of ground points (GCPs), welldistributed in the zone, were measured in theframe of a GPS campaign (Mora et al., 2003).

The procedure consisted of four steps:i) Orientation of the 2000 photos.ii) Identification of natural points on the

2000 stereoscopic model and editing of precisemonographs.

iii) Photogrammetric measurement of 3Dcoordinates of such points in the 2000 referencesystem and identification of the same points onthe old photos.

iv) Computation of EOPs of various histor-ical models using GCPs derived from the 2000survey.

The GCPs were selected outside the land-slide bodies, giving the preference to artificialelements (for example, buildings), homoge-neously distributed in the area.

The GCPs selection procedure was based onassessment of space resection program for thehistorical images, rejecting, with an iterativeapproach, points showing high external orienta-tion residuals respect to a theoretical valuecoming from the scale of the images. Obvious-ly, the ability to correctly identify points unus-able for support the orientation is greater if alarge number of points is examined.

At the end of the procedure, all of the «his-torical» stereoscopic models were oriented intoan absolute reference system derived from the2000 survey, making it possible to perform com-parisons between the DTMs, as later discussed.

The orientation and plotting steps for thestereoscopic models were first performed with aDIGICART 40 (Officine Galileo) analytical plot-ter. Because no calibration certificates wereavailable for the cameras used on the historicalflights (1954, 1971, 1976), the internal orienta-tion and image coordinates refinement procedurewas applied using the focal length pressed on thefilms, ignoring effects due to lens distortion,non-planarity of the film, and non alignment of

the principal point with its theoretical position.Correction was made for the earth curvature.

Relative orientation was performed using avery large number of points distributed uni-formly through the entire area of the modelscovering the landslide zone. Absolute orienta-tion was performed with the procedure de-scribed above. Once the orientation parameterswere known, it was possible to acquire three-di-mensional data for the slope under study.DTMs of the zone with grids of different size(2×2 m, 4×4 m, 8×8 m) were created from thefour models in order to reveal any mass move-ments and to simultaneously determine the bestcell size for describing morphological details.

In assessing the degree of uncertainty of theresults, it has to be considered that an uncon-ventional method was used to solve the orienta-tion problem, and that the available images arecharacterized by different scale, ranging from1:60000 to 1:4400.

This fact created many problems:1) Errors of a 1:60000 scale model are

greater than those of a 1:4400 scale model.2) It is difficult to identify homologous

points to use as reference (especially between1954 and 2000), due to the different resolutionof the photos and to the lack of useful elementsin the territory. The search procedure was con-ducted by dividing the common zone into sub-areas in which a large number of homologouspoints were then catalogued. Analysis of resid-uals deriving from the external orientation pro-cedure revealed, at any step, the points thatcould be considered useful for this purpose, be-cause assumed not subject to movements.

3) In the 1:4400 photos points cover the en-tire observed area, whereas in the lower-scaleimages they are concentrated in a small zone ofthe model.

In the absence of data from other sources,simple approaches such as computation of vari-ance propagation and empirical assessments ofthe quality of results were utilised to estimatethe precision of DTMs acquired in stereoscopyand therefore to evaluate the significance of themeasured deformations.

Assuming the so called Normal Case, in or-der to simplify the colinearity equation (thecamera axes are considered parallel to each oth-


er and perpendicular to the base line) and ap-plying the law of variance propagation, consid-ering both the error assigned to measurement ofhorizontal parallax (assuming σParallax=±10 µm)and, only for the pre-2000 surveys, the camerafocal length (assuming σCamera=±10 µm), wederived a theoretical mean square deviation val-ue expected for Z coordinate (elevation); if B isthe distance between the camera’s centres rela-tive to two stereo pairs, as known, for usualbase-ratios (B/Z>0.5) this is generally greaterthan the planimetric ones. Table II shows themain flight data and the value of theoretical σZ.

Table II shows a clear correspondence be-tween the mean standard deviations and thescale factors of the photos: the 1954 surveyshows theoretical errors of a higher magnitudethan those from 2000. Moreover 1971 presentstheoretical errors that are slightly lower thanthose from 1976 data; this is due to the base-ra-tios values. Obviously, the error related to un-certainty in collimation of natural points mustbe added to the σZ theoretical one.

A measurement repeatability test was per-formed: DTMs on sample area were generatedin different periods, by the same operator, withthe same analytical instrument, in semi-auto-matic mode. The comparison of such results

(see table III) shows good repeatability of themeasurements and confirms the close correla-tion between image scale factor and errors.

Considering the above observations as awhole and the values listed in table II, and in or-der to adopt a simple and homogeneous criteri-on, we have chosen 1 m as the significant rangeof movement in DTMs comparisons, from 1971up to 2000. The point differences lying withinsuch interval have therefore been considerednon-indicative of deformation due to landslidemovement. If the 1954 DTM is used for com-parisons, the significant range assumed is 2 m.

To guarantee the quality of data to be used inthis research, the above-described photogram-metric method was performed using both an an-alytical plotter and semi-automatic or manualvector procedures, with a significant amount ofwork of an expert operator, and the digital pho-togrammetry technique for the automatic gener-ation of terrain models and orthophotos. In par-ticular, in addition to the photogrammetricprocess performed by analytical stereoplotter,the 2000 survey images were used to generate aDTM with a high-level digital photogrammetricworkstation (Helava System), adopting semi-au-tomatic and automatic procedures.

The films were scanned with a RasterMas-ter (Wehrli & Associates Inc., NY, U.S.A.) pho-togrammetric scanner at a 1000 dpi resolution,

1075

Table II. Characteristics of available photogrammet-ric surveys and theoretical assessment of level error, as-suming σPξ= Standard Deviation (SD) in measurementof horizontal parallax =±10 µm and, for pre-2000 sur-veys, σc =SD of focal length of camera =±10 µm. Val-ues expressed in meters.

Year 1954 1971 1976 2000

Mean flight 9509.9 3376.8 2973.9 1011.3heigh (m)Nominal 0.154 0.152 0.153 0.152

focal lengthBase line 5729.2 1818.5 1137.7 365.6

Mean topographic 300.00 300.00 300.00 300.00heigh (m)

Mean 60000 20000 17500 4700scale factorMean B /Z 0.62 0.59 0.43 0.51

σZ ±1.45 ±0.58 ±0.72 ±0.18

Table III. Measurement repeatability test.

1971 1976 2000

No. points 214 214 850Standard deviation (m) 0.68 0.59 0.36

Table IV. Characteristics of images acquired withphotogrammetric scanner.

Year of the survey 2000Photo scale 1:4.400

Panchromatic film ColourImage resolution (µm) 24Ground resolution (cm) 12

File image dimension (Mb) 240

1076


Fig. 4. Comparison between the reference DTM (DIGICART 2×2 m) and DTM automatically generated withHelava system; colour classified differences are shown overlapped the shaded relief model.

Fig. 5. High differences obtained from 1976 and 1954 surveys. Differences (in meters) superimposed on 1976ortho-photo are indicated by different colours.


sufficient to guarantee a high level of detail(ground pixel size about 12 cm; table IV).

The DTM was firstly generated automatical-ly, with post-editing by the operator to correct er-rors deriving from the correlation procedure; ed-iting consisted mainly of manual insertion ofbreaklines into incorrect zones, with consequentlocal recalculation of the surface area.

The DTM automatically generated by digi-tal photogrammetry was then compared to theone generated by the analytical plotter; fig. 4shows the differences between the two models.It may be seen that for 96.5% of the points, thedifference between the two models is within aninterval of ±1 m, and that the largest differencesare localized mainly in zones with complexmorphology and shadowed areas.

In addition, together with DTM and classicvector products, a digital orthophoto has beenproduced, to give a photographic representa-tion, with metric characteristics, of the situationin a landslide area at a defined time. Moderndigital photogrammetry and GIS programs canalso generate highly effective three-dimension-al views that combine the expressive potentialsof ortho-photos and three-dimensional data ofDTM. Figure 3, generated with ErMapper soft-ware, refers to the 2000 orthophoto.

The comparison of multitemporal DTMs,performed without the use of interpolators, al-lowed us to quantify the geometric variations oc-curring as a result of evolution of the landslide.The two following figs. 5 and 6 graphically showthe results of comparisons of the 1954, 1976 and

1077

Fig. 6. High differences between the 8×8 m DTMs obtained from 2000 and 1976 surveys. Differences (in me-ters) superimposed on 2000 orthophoto are indicated by different colours.

1078


2000 surveys; the movement determined for in-dividual cells is highlighted in colour. For betterlegibility of the figures, the digital orthophotosgenerated for the 1976 flight and the 2000 flight,respectively, have been used.

Figure 5 shows the result of the comparisonbetween DTMs from 1954 and 1976, with 8×8 mgrid resolution. Figure 6 refers to differences rel-ative to the period 1976-2000, for which a his-togram of heigh differences is shown in fig. 7.

Other comparisons, for example those in-volving the 1971 model, are not discussed inthis paper, due to the lack of further informationwith respect to the presented data.

4. Landslide map and differential DTMcomparison

Comparative analysis of the multi temporalDTMs indicate that the landslide reactivated inthe time interval 1954-1971, while in the fol-lowing period (1971-1976) the slope was al-most inactive. The movements involved boththe Cà di Bosco and Cà di Malta landslide bod-ies (fig. 5). The alimentation area is character-ized by a main scarp with arched shape, locallycomplex, that is recognisable at various points.The movements observed in this area confirm adistribution of a moderate retrogressive activity

along the left side of the landslide area, as de-tected by means of the photo interpretation in-vestigations. The amount of material moved isless than assumed by qualitative inspection.

The time interval 1976-2000 was character-ized by a more important reactivation of thelandslide in 1996, after a long period of relativeinactivity (Mora et al., 2003). Subsequently, in1998, a small rotational slide involved the upperpart of the Cà di Malta main body. Part of theeffects of these movements, principally at thefoot of the landslide, were remodelled by hu-man activity in the frame of consolidation anddrainage works.

The results of natural and anthropic activi-ties are shown in fig. 6. The largest mass move-ments occurred at the scarp areas of the Cà diMalta and Cà del Bosco landslides, where nega-tive values ranging between 2 and 8 m are pres-ent in fairly limited areas. On the other hand, themost significant uplifts are also localized in veryrestricted areas, mainly at the foot of the land-slides, and they do not exceed 4.5 m. Over anarea of about 15×104 m2 the volumetric analysisgives a negative variation of 75×10 3 m3 and anaccumulation of about 25×103 m3.

Observing the statistic of the residual con-cerning areas assumed to be stable (differences<1.00 m) and assuming systematic errors due tomodel orientations of about 20 cm for the 1997-

Fig. 7. Statistic of differences between 2000 and 1976 DTMs.

1079


2000 data, the estimation of the volumes is giv-en with the precision of 20%.

The difference between the depletion andaccumulation volumes is principally due to theremoval of material along the river and abovethe road at the foot of the slope.

For more precise analysis of the relation-ships between the high residuals and the 1976and 2000 landslides map, fig. 6 shows (withcoloured rectangles and letters, points A to E)the areas in which the most significant surfacelowering or uplifts occurred. The significanttopographic changes may be seen at the slip-flow that develops along the left side of the Càdi Malta landslide.

Point A corresponds to an area of about 4000m2, with the largest negative variations, and co-incides with the landslide alimentation area,quantitatively defining the extent of the depres-sion and the distribution of the slopes of thescarp area. Degradation values increase rapidlytoward the centre of the area. This depressed areahas arched borders located precisely at the mainscarp and at the head of the landslide.

At point B, the scarp is clearly shown in anarea of the differentials map in which major lev-el degradations may be observed. The valuespresenting the most significant variations (be-tween 2 and 4 m) correspond to the alimentationarea of the 2000 slip-flow, which caused a sharpdepression bordered by a distinct main scarpand repeated secondary scarps and terracings.

Point D shows local uplifts. This zone coin-cides with the foot of the landslide, where theaccumulation is elongated and slightly lobed.

Other significant topographic changes maybe seen at points B and E, along the slip-flowthat develops starting from the right side of theCà del Bosco landslide.

Point E shows an area in which the differen-tial map presents a distinct uplift of about 4 m.This area corresponds to the point where thelandslide stopped, generating a lobe-shaped ac-cumulation at the foot.

Significant topographic changes may also beseen at point C where the slope was remodelledby man for reinforcement and drainage works.

Lastly in the area at the foot of the landslidethe topographic changes shown on the residualsmap are not traceable to landslides, but rather to

anthropic activity. In particular, the degradationsare due to construction of a short section of roadleading to the Reno River, while the uplifts arecaused by levelling of the slope by man to re-claim the foot of the Cà di Malta landslide.

5. Conclusions

Stereoscopic analysis of four sets of aerialphotos, taken over a time span of almost 50years, allowed reconstruction of the complexsequence of landslide movements occurring onthe slope under study. The 1954 photos, despitethe low-scale (1:60000), provided significantevidence of a slip-flow-type landslide body af-fecting the entire slope. This event may be con-sidered the primary cause of all of the subse-quent movements, which are therefore to be un-derstood as partial reactivations of such land-slide. Of these, the Cà di Bosco and Cà di Mal-ta landslides are the most extensive phenomena.

The event cannot be dated on the basis ofavailable data. Nevertheless, studies conductedin the Apennine area have demonstrated thatmost of the large landslides are hundreds or eventhousands of years old (Bertolini and Pellegrini,2001).

A quantitative analysis of the territory hasbeen obtained with an appropriate comparison ofphotogrammetric surveys of the zone realised indifferent years, allowing identification of geo-metric changes occurring during the time inter-val in question. The absence of fundamental in-formation on geometric parameters of the im-ages has been overcome by adopting an uncon-ventional method to co-register all the datasets inthe same reference frame, based on the detectionof homologous points in the multi temporalmodels. The areas and the amount of masses in-volved in the reactivations of the landslides havebeen estimated, giving a contribution to the mor-phological interpretation of the landslide defor-mation phenomena in the last years.

The availability of historical stereoscopic im-ages assumes particular importance in the studyof long time span phenomena, which producesurface deformations of the Earth. Furthermorethe possibility of adopting the digital photogram-metry approaches will yield faster and higher

1080


resolution results, at lower costs with respect tothe analytical analysis. The availability of high-resolution satellite images which may becomecompetitive with small-medium scale aerial pho-tos will lead to the low-cost creation of imagearchives with repetition intervals that depend onthe temporal resolution of the satellite.

Acknowledgements

The work was partially financed by thePRIN2003 Italian National Research Project«Tecnologie innovative per la previsione, ilcontrollo e la mitigazione dell’impatto delleemergenze ambientali» (Nat. Resp. Prof. SergioDequal). The authors wish to thank to Prof. Lu-ca Vittuari and Dr. Francesco Mancini for theirsupport in data processing.

REFERENCES

ANZIDEI, M., P. BALDI, F. CHIOCCI, M. MARSELLA, E. MAR-TORELLI and A. ZANUTTA (2000): Integrazione tra un rilie-vo aerofotogrammetrico e batimetria swath: il DTMMdel versante orientale e meridionale dell’Isola di Pal-marola (Lazio), Boll. Soc. Geol. Ital., 119, 767-775.

BALDI, P., G. BITELLI, A. CARRARA and A. ZANUTTA (2002):Detecting landslide long term movements by differen-tial photogrammetry, in Proceedings of the EuropeanGeophysical Society, XXVII General Assembly, 21-26April, Nice, France.

BALDI, P., M. FABRIS, A. MARSELLA and R. MONTICELLI

(2005): Monitoring the morphological evolution of theSciara del Fuoco during the 2002-2003 Strombolieruption using multi-temporal photogrammetry, ISPRSJ. Photogramm. Remote Sensing, 59 (4), 199-211.

BERTOLINI, G. and M. PELLEGRINI (2001): The landslides ofthe Emilia Apennines (Northern Italy) which resumedactivity in the 1994-99 period and required Civil Pro-tection interventions, Quad. Geol. Appl., 8 (1), 27-74.

BITELLI, G., M. DUBBINI and A. ZANUTTA (2004): Terrestri-al Laser Scanning and Digital Photogrammetry tech-niques to monitor landslide bodies, in Proceedings ofISPRS, 12-23 Luglio 2004, Istanbul, Turkey.

BRUNSDEN, D. and J.H. CHANDLER (1996): Development ofan episodic landform change model based upon theBlack Ven mudslide, 1946-1995, in Advances in Hills-lope Processes, edited by M.G. ANDERSON and S.M.BROOKS (J. Wiley, Chichester, U.K.), 869-896.

CHANDLER, J.H. and D. BRUNSDEN (1995): Steady state be-

haviour of the Black Ven mudslide: the application ofarchical analytical photogrammetry to studies of land-form change, Earth Surf. Processes Landforms, 20,255-275 .

CHANDLER, J.H. and M. COOPER (1988a): Monitoring thedevelopment of landslides using archival photographyand analytical photogrammetry, Land Miner. Surveying(Royal Institution of Chartered Surveyors), 6, 576-584.

CHANDLER, J.H. and M. COOPER (1988b): The extraction ofpositional data from historical photographs and theirapplication in geomorphology, Photogramm. Rec., 13(73), 69-78.

DOLAN, R., D. HAYDEN, D. MAY and S. MAY (1980): The re-liability of shoreline change measurements from aerialphotographs, Shore Beach, 48 (10), 22-29.

DUBBINI, M. and A. ZANUTTA (2005): Terrestrial laser scan-ning to landslide movements detection, in Proceeding toWorkshop Italy-Canada on «3D Digital Imaging andModeling, Applications of Heritage, Industry, Medicineand Land», 17-18 May 2005, Padova, Italy.

KÄÄB, A. (2000): Photogrammetry for early recognition ofhigh mountain hazards: new techniques and applica-tions, Phys. Chem. Earth., 25 (9), 765-770.

KÄÄB, A. and M. FUNK (1999): Modelling mass balance us-ing photogrammetric and geophysical data: a pilotstudy at Griesgletscher, Swiss Alps. Instruments andMethods, J. Glaciol., 45 (151), 575-583.

KRAUS, K. (1997): Photogrammetry, Advanced Methodsand Applications (Dümmler-Bonn), vol. 2, pp. 466.

LANE, S.N., K.S. RICHARDS and J.H. CHANDLER (1998):Landform Monitoring, Modelling and Analysis (JohnWiley & Sons), pp. 454.

MORA, P., P. BALDI, G. CASULA, M. FABRIS, M. GHIROTTI, E.MAZZINI and A. PESCI (2003): Global Positioning Systemsand digital photogrammetry for the monitoring of massmovements: application to the Cà di Malta landslide(Northern Apennines, Italy), Eng. Geol., 68, 103-121.

OVERTON, M.F. and J.S. FISHER (1996): Application of 3Dcomputer modeling using digital photogrammetry tomeasure shoreline change, in Proceedings of the 25thInternational Conference on Coastal Engineering, AS-CE, 3, 3750-3761.

PESCI, A., P. BALDI, A. BEDIN, G. CASULA, N. CENNI, M.FABRIS, F. LODDO, P. MORA and M. BACCHETTI (2004):Digital elevation models for landslide evolution moni-toring: application on two areas located in the RenoRiver Valley (Italy), Ann. Geophys., 47 (4), 1339-1353.

SIMONI, A., M. BERTI, M. GENERALI, C. ELMI and M. GHIOTTI

(2004): Preliminary result from pore pressure monitoringon an unstable clay slope, Eng. Geol., 73, 117-128.

VAN WESTEN, C.J. and F.L. GETAHUN (2003): Analyzing theevolution of the Tessina landslide using aerial photo-graphs and digital elevation models, Geomorphology,54, 77-89.

(received January 22, 2006;accepted July 17, 2006)

qualitative and quantitative photogrammetric techniques ...antonio zanutta, paolo baldi, gabriele...

Documents