FIRE and MUD Contents

Preeruption and Posteruption Digital-Terrain Models of Mount Pinatubo

By John W. Jones1 and Christopher G. Newhall1

1U.S. Geological Survey.


Digital terrain-elevation data for Mount Pinatubo, gathered before and after its June 15, 1991, eruption, show broad features of both the preeruption geology and eruption-related changes. Two major conjugate faults, one trending northeast and the other southeast, seem to cut the ancestral and modern Pinatubo edifices and to bound the north edge of pyroclastic-flow deposits in the Sacobia River valley. The June 1991 eruption created a 2.5-kilometer-wide collapse caldera and filled valleys around Pinatubo with about 5.50.5 cubic kilometers of pyroclastic-flow deposits. Products of prehistoric eruptions that were larger than that of 1991 appear as rough-textured inliers (kipukas) surrounded by smooth-textured 1991 pyroclastic-flow deposits. The new summit elevation of Mount Pinatubo is approximately 1,485 meters above sea level, reduced from a preeruption elevation of 1,745 meters; the elevation of the caldera lake is between 820 and 840 meters above sea level, or about 650 meters below the highest point on the new caldera rim.

Attempts to quantify volume loss in the summit region and volume gain in the valleys by subtracting preeruption digital data from posteruption data were unsuccessful. Interferences may have included low resolution of the data (3-arc-second pixels), imprecision in registration of preeruption and posteruption data, loss of forest cover during the eruption, syneruption or posteruption erosion and (or) subsidence, and artifacts from cloud cover or splicing of data sets. Difficulties encountered here are a useful caution against use of similar methods without independent corroboration of volume changes.

Note to readers: Figures open in separate windows. To return to the text, close the figure's window or bring the text window to the front.


Before its June 15, 1991, eruption, Mount Pinatubo consisted of a rounded, steep-sided, domelike mass that rose about 700 m above a broad, gently sloping, deeply dissected apron of pyroclastic and epiclastic materials. Some relics of older volcanic edifices, including an ancestral Mount Pinatubo, lay south, east, and northeast of Mount Pinatubo (Delfin, 1983, 1984; Newhall and others, this volume). In comparison to well-known stratocones such as Mayon or Fuji, Mount Pinatubo was small and inconspicuous, but its extensive pyroclastic apron told of large prehistoric explosive eruptions.

Eruption of about 5 km3 of magma on June 15, 1991 (W.E. Scott and others, this volume) created a new, 2.5-km-diameter collapse caldera centered slightly northwest of the preeruption summit. The preeruption summit was included in the area of collapse, so the posteruption height of Mount Pinatubo was substantially reduced. Valleys that had existed in the pyroclastic apron were largely filled by eruptive products; valleys that had been carved into older volcanic terrain and partly filled by prehistoric eruptions of Mount Pinatubo were partly filled once again.

Preliminary posteruption digital topography was compared with preeruption topography to correlate preeruption and posteruption features and to estimate the volumes lost from the summit area and added to the surrounding pyroclastic apron. The resolution of our posteruption topography is relatively poor, so much of the discussion that follows is qualitative.



The Defense Mapping Agency provided us with tapes containing preeruption and posteruption digital-terrain elevation data (DTED), with 3-arc-second (90-m) lattice resolution. The data span 15 minutes of latitude and 30 minutes of longitude, centered on Mount Pinatubo. Posteruption data are spliced together from several data sets that were collected from November 1991 to November 1992. Data were read from tape into the Earth Resource Data Analysis System (ERDAS), which had an appropriate translator for DTED format files. In ERDAS, we used the nearest-neighbor resampling method (Lillesand and Kiefer, 1987) to rectify data from latitude/longitude referencing with the 3-arc-second lattice resolution to the UTM coordinate system with a 100-m resolution.

After some initial display, contouring, and examination of the posteruption DTED, some artifacts in the data became readily apparent. For example, phantom linear hills appeared where none actually exist. Therefore, the projected files were transferred to ARC/INFO geographic information system software and edited by use of grid-edit tools in ARC (Environmental Systems Research Institute, Inc.).

Various filtering and visual editing methods to mitigate the effects of the artifacts were tested and found to be inadequate. Finally, artifacts in the posteruption DTED were minimized by inserting terrain information from the preeruption DTED, where necessary, into the posteruption DTED. Because of this procedure, posteruption topography is locally incorrect, but contours of the map as a whole approximate the present terrain better than they would had the artifacts been allowed to remain. The largest patch of inserted, preeruption topography is northeast of the caldera.

Similar substitution of preeruption terrain information into the posteruption data file was done for 10-km-wide strips at the east and west edges of the 15' x 30' area. Thus, no change, real or otherwise, can be described in these margins, which are mostly areas of alluvial fans and likely posteruption sediment deposition.

An additional correction was required for a slight discrepancy in registration of the preeruption and posteruption DTED. We tried shifting the posteruption DTED zero, one, and two pixels north, south, east, west, and diagonally. A shift of posteruption data by one pixel (100 m) to the east minimized residual differences between the two data sets for points which we knew did not change in elevation.


Contour maps were produced for two areas: one, 6 km on a side, and the other, encompassing 15 minutes of latitude and about 30 minutes of longitude, both approximately centered on the caldera. Files of these areas were reformatted into a triangulated irregular network (TIN) data structure, in which triangles fitted to the input DTED lattice were subdivided three times in order to produce relatively smooth contour lines in the final maps. A contour interval of 20 m was selected to match the interval used on preeruption published topographic maps.

For both the preeruption and posteruption cases, maps were plotted at 1:250,000 and 1:100,000 scales for the area of 15 minutes x 30 minutes and at an approximate scale of 1:15,000 (actual scale 1:15,030) for the 6 km x 6 km area. The posteruption contour map still required some interactive editing in order to overcome the influence of artifacts in the digital elevation data. This editing was done by comparison of apparent posteruption topography with the broad features of topography known to us from fieldwork in the area.


We delineated watershed boundaries, by hand, on the contour map created from the preeruption digital terrain data. Similar preeruption watershed boundaries had been drawn on published preeruption topographic maps, and we checked our new boundaries against previously drawn boundaries for consistency. The fit was good. The hand-drawn boundaries (fig. 1) were, in turn, digitized as vectors, tagged, and gridded into raster file format using the ARC/INFO software. A resolution of 100 m was used in the gridding process to provide for overlay analysis of the terrain files with the watershed boundaries.

Figure 1. Pinatubo watersheds and other geographic features noted in text.


Apparent loss of volume from Mount Pinatubo and gains of volume on surrounding terrain were estimated by computer subtraction of the preeruption DTED from the posteruption DTED. Inputs for this procedure were the pixel-shifted, artifact-cleaned DTED. Elevation data were then transferred into PCI software (PCI Inc.), where they were differenced on a cell-by-cell basis. An elevation change was calculated for each 100 m x 100 m cell. Then, the volume of change for each cell was estimated by multiplying each pixel's elevation change by 104 m2, and these results were summed over each watershed area, including the caldera (table 1). Positive and negative changes are listed separately to show any anomalous data.

Table 1. Comparison of various estimates of volume loss (-) and volume gain (+) resulting from eruption of Mount Pinatubo on June 15, 1991.

[-, not estimated]


Estimates of Volume Change (this study)


Independent Estimates of Volume Change

volume change1

Negative volume change1

Tree and erosion corrections2 (km3)

Corrected volume change3


Punongbayan and others4 (km3)


Scott and others (this volume) (km3)

U.S. Army Corps of Engineers (1994)


Daag (1994)

0.76 -1.01 0.11+0.04 0.91   +0.65 +0.28 +0.245 -
0.36 -0.72 0.04 0.40   - +0.06 - -
0.70 -0.88 0.02 0.72   - - - -
0.78 -0.49 0.22+0.14 1.14   +0.9 +0.60 +0.606 +0.88
0.02 -0.05 0 0.02   - - - -
0.11 -0.10 0 0.11   +0.2 +0.10 - +0.01
1.02 -0.33 0.10+.02 1.14   +0.5 +0.30 +0.30 +0.40
1.35 -0.65 0.02 1.37   - - - -
1.81 -1.49 0.02 1.83   +.057 +0.037 - -
1.62 -1.53 0.30+0.21 2.13   +1.38 9 +1.268 9 +1.408 9 -
Santo Tomas (excl. Marella)
0.12 -0.06 0 0.12   - - - -
0.12 -0.18 0.03 0.15   - - - -
Balin Baquero
1.42 -1.27 0.63+0.60 2.65   - +1.40 - -
0.64 -0.72 0.34 0.98   +3.110 +0.88 +3.0010 -
0.06 -1.64 0.02 -1.62   - -2.5 - -
    Total volume change
10.77 -11.12 - +12.05   - +2.4 - -

1Major apparent volume losses and gains appeared in watersheds where we know that little actual topographic change occurred (Bangat, Marimla, Porac, Gumain). Most of the overestimates of volume gain, we think, resulted from slight misregistration of the preeruption and posteruption data, and are especially pronounced in rugged terrain where even slight misregistration can result in large error. Most underestimates of volume gain, we think, resulted from misregistration of preeruption and posteruption data, flattening of forest by the eruption, and posteruption erosion of pyroclastic deposits.

2 The first figure for each watershed, tree correction, is the estimated volume of the preeruption forest, as differenced from treetops to actual ground surface. For all watersheds except the caldera, this correction is taken to be 15 m (an average height of the forest canopy in valleys) times the area of each watershed that was buried by valley-filling pyroclastic-flow deposit. Areas of valley-filling pyroclastic-flow deposit are from W.E. Scott and others, this volume. For the caldera, this correction is taken to be 5 m (height of a scrub forest) times the entire area of that watershed. The second figure, where present, is a correction for the volume of 1991-92 erosion, as estimated by the U.S. Army Corps of Engineers (1994). The erosion correction for the Balin Baquero is for the Balin Baquero and Bucao Rivers combined.

3Corrected volume change for all areas except the caldera is calculated as column 1 plus column 3. Note that negative volume changes are suppressed except in the caldera. Volume change for the caldera is calculated as column 2 plus column 3. We do not believe either corrected or uncorrected estimates and show them only to illustrate that, when examined by watershed, some are obviously false.

4PHIVOLCS has reestimated volumes several times. This is the latest. The earliest estimates were by J. Daligdig and G. Besana in 1991; this revision is by N.M. Tungol and R.A. Arboleda.

5Includes Bangat watershed.

6Includes Abacan watershed.

7Includes Porac watershed.

8Includes upper Santo Tomas watershed.

9Includes Maloma watershed.

10Includes Balin Baquero watershed.


Independent estimates of volumes of pyroclastic-flow deposits (volume gain) were made by sketching 1991 valley fill from aerial oblique photographs and video onto published preeruption topographic maps, differencing preeruption and posteruption cross-valley profiles, and multiplying by the length of each applicable reach (Daligdig and others, 1991; W.E. Scott and others, 1991; this volume). Estimates by the U.S. Army Corps of Engineers (1994) were made by similar methods but from vertical aerial photographs with better photogrammetric control. Estimates by Daag (1994) for volume gain on the Sacobia-Abacan-Pasig pyroclastic fan used July 1991 aerial oblique stereo photographs, from which levels of fill were marked on preeruption topographic sheets, contoured, and digitized. The preeruption topography was then subtracted digitally from the posteruption topography.

An independent estimate of the volume of caldera collapse (volume lost from the summit area), by Scott and others (this volume) was made by drawing the new rim and floor of the caldera onto a topographic profile of the preeruption volcano, assuming 800 m as the elevation of the caldera floor, and summing the volumes of disk-shaped layers within this simple model.



Largely because of the coarse resolution of our digital topography (figs. 2A,B), and the opportunity to view this in shaded relief (figs. 3A,B), several large-scale features of interest can be seen. Many of these features were previously known but are particularly well displayed in the digital topography.

Figure 2. Pinatubo topography, 15 minute x 30 minute area. Contour interval 20 m. A, Preeruption. B, Posteruption, a composite of data gathered from November 1991 to November 1992. Sacobia and west-side pyroclastic fans are shown. Refer to figures 3A,B for identification of additional features.

Figure 3. Shaded relief views of Mount Pinatubo. A, Preeruption view showing preeruption summit. B, Posteruption view showing major fault lineaments (Iba-Maraunot and Sacobia), ancestral Pinatubo edifice, modern Pinatubo, and 1991 caldera.


The 1991 caldera is centered about 750 m northwest of the preeruption summit and about 350 m southeast of where a precursory lava dome began to form on June 7, 1991 (figs. 4, 5). The 1991 caldera is more or less concentric within the modern Pinatubo edifice that in turn lies concentrically within the prehistoric Tayawan caldera (fig. 3B) (Delfin, 1983, 1984; Newhall and others, this volume). Evidence for a prehistoric but relatively young caldera only slightly north of the 1991 caldera was discussed in Newhall and others (this volume). A fissure and line of phreatic explosion pits that formed on April 2, 1991 (Sabit, this volume; Wolfe and Hoblitt, this volume), extended northeastward from the northeast margins of both this young prehistoric and the 1991 calderas. A line of vigorously steaming vents that extended southwest of this fissure immediately following the phreatic explosions of April 2 trended nearly through the center of the 1991 caldera.

Figure 4. Pinatubo topography, 6 km x 6 km area centered on caldera. A, Preeruption view showing location of the June 1991 dome. B, Posteruption view showing location of the 1992 dome. Note that the 1991 caldera is offset north-northwest from the preeruption summit, and the 1992 dome is offset southeast from the 1991 dome.

Figure 5. Pinatubo topography, 6 km x 6 km area centered on the caldera, with preeruption 20-m elevation contours (blue) overprinted with posteruption 20-m contours (black). Major changes in the summit region are as seen in figure 4, but here are superimposed on each other for ease of checking change at specific points. Closely matching preeruption and posteruption topography northeast of the caldera is in an area that experienced little change during the eruption; the consistency is partly illusory, though, because a patch of preeruption topography was forced onto posteruption topography, as discussed in text, to avoid an even larger error. Several unpatched discrepancies are still present, including apparent postcaldera "hills" at center right and 1.5 km northwest of the caldera rim. Discrepancies are artifacts rather than real change; they can be ignored in map form but have not been successfully removed from estimates of volume change.

The location of the July-October 1992 dome can be seen in posteruption topography, centered within the 1991 caldera. Thus, judging from locations of the June 1991 and July-October 1992 domes, the active vent shifted about 350 m to the southeast between those episodes of dome growth, probably during the caldera-forming eruption of June 15, 1991.


In principle, differencing of preeruption and posteruption DTED should yield reliable volume changes. Qualitatively, the results are reasonable (fig. 6). Those valleys which we know to have received thick pyroclastic fill show significant volume gain, and the new caldera shows significant volume loss. However, quantitative estimates (fig. 5 and table 1) are inconsistent with volumes that others have previously estimated (Daag, 1994; Punongbayan and others, 1994a; U.S. Army Corps of Engineers, 1994; W.E. Scott and others, this volume).

Figure 6. Gray-scale map of volume gain and loss around Mount Pinatubo. An intermediate gray indicates no change, while black shows maximum loss and white shows maximum gain. As in figure 5, most of the change is real, but two important artifacts are easily seen. The first is the patch created by the authors northeast of the caldera; the second consists of 10-km-wide strips along the east and west edges of the figure. In both the patch and the strips, preeruption topography was ported directly to the posteruption data set, so no change appears, and those parts of the image are featureless gray. The strips, placed in the posteruption data prior to any processing by the authors, remove most areas of lahar deposition from calculations, even though much of that lahar deposit was derived from 1991 products whose volume we tried to estimate.

Major volume gains and losses even appeared in watersheds where little actual topographic change occurred (columns 1 and 2, table 1). Why is there such a poor result from our digital reestimation?

Random discrepancies (without tendency to inflate or deflate actual volume change) could have been caused by low resolution of the available DTED, imprecise registration of preeruption and posteruption DTED, or artifacts resulting from splicing of multiple data sets. Because severe errors are apparent in watersheds that have steep terrain and that received only minimal new deposit, we think that imprecise registration is a major problem. The problem remained even after we shifted the posteruption data 1 pixel to the east, as described earlier.

Systematic underestimates of valley filling could have been caused by

Systematic overestimates of valley filling (or an underestimate of volume loss, as in the caldera) could have been caused by

Serious, nonsystematic error from inaccurate registration of preeruption and posteruption data seems likely, especially in rugged terrain. In addition, because the apparent net volume change for the entire area (sum of columns 1 and 2) is much lower than we expect from field investigations, and, indeed, is negative, whereas the true change is probably positive, we also suspect some serious, systematic source of negative elevation change, certainly including posteruption erosion and possibly also including loss of forest. Preeruption topography may have been along the treetops; the eruption blew down trees, and the apparent ground surface dropped by the height of the trees. Further evidence for the influence of forest cover is found in the texture of preeruption and posteruption elevation data. Preeruption data lack the texture and erosional features visible in posteruption data (fig. 3A,B).

We cannot correct for imprecise registration of data. To correct for a possible forest artifact, we artificially reset all negative elevation changes except in the caldera region to zero before calculating volume changes. In addition, where pyroclastic-flow deposits filled valleys, our estimate of positive elevation change might be too low, by the height of the preeruption trees. Column 3 of table 1 includes a volume correction of 15 m (average height of trees) times the area of pyroclastic valley fill in each watershed, to be added to the previous estimate of volume gain. For the area of the former summit (now caldera), a 5-m average thickness of preeruption vegetation cover is multiplied by the entire area of that watershed. Column 3 also includes, for major drainages, a correction for the volume of 1991-92 erosion as estimated by the U.S. Army Corps of Engineers (1994). Column 4 is the sum of columns 1 and 3 except in the case of the caldera, for which column 2 and column 3 are summed.

Previously published, independent estimates are qualitatively consistent with what one can readily observe, and with each other, though estimates of Daligdig and others (1991) are generally higher than the other three estimates. Estimates of volume change from this paper, even after attempted corrections, are not consistent with what one can observe in the field. Therefore, we believe that our new estimates of volume change are incorrect and should not be used.

Ironically, we originally thought that it would be easier to quantify volume changes in the uplands, where changes in elevation were greatest, than it would be to quantify volume changes in the lowlands, where deposition from lahars increased elevations by only a few meters. The highlands turned out to be difficult; we do not have enough data for areas of lahar deposition to test whether volume change in those lowlands is, in fact, quantifiable from available DTED.

In short, our attempt at volume estimation from DTED was unsuccessful, and it raises some important cautions for similar applications where there is no independent information by which to corroborate DTED-based estimates.


Major features of preeruption topography show well in low-resolution (3-arc-second) digital data. Two major conjugate faults, one trending northeast and the other southeast, seem to bound and cut the ancestral Pinatubo edifice and to localize the present Sacobia pyroclastic fan. Products of prehistoric eruptions from the modern Pinatubo, larger than those of 1991, appear as rough-textured inliers (kipukas) on posteruption topography.

Massive topographic changes caused by the 1991 eruption are readily apparent in a comparison of preeruption and posteruption digital topography. The principal features are a new summit caldera and infilling of most valleys surrounding Mount Pinatubo by pyroclastic flows.

Our attempts to quantify volume changes by subtraction of 3-arc-second DTED were unsuccessful, especially in steep terrain. Higher resolution data, including posteruption data for the entire 15 minute by 30 minute area, would allow more precise registration of control points (points that were known to have not changed in elevation), more precise evaluation of the role of forest loss, an automatic correction for posteruption erosion, and thus more precise estimation of actual volume changes.


Bautista, B.C., Bautista, M.L.P., Stein, R.S., Barcelona, E.S., Punongbayan, R.S., Laguerta, E.P., Rasdas, A.R., Ambubuyog, G., and Amin, E.Q., this volume, Relation of regional and local structures to Mount Pinatubo.

Bruinsma, J.W., 1983, Results of potassium-argon age dating on twenty rock samples from the Pinatubo, southeast Tongonan, Bacon-Manito, Southern Negros, and Tongonan, Leyte areas, Philippines: Proprietary report by Robertson Research Ltd. to the Philippine National Oil Company-Energy Development Corporation, 24 p.

Daag, A.S., 1994, Geomorphic developments and erosion of the Mount Pinatubo 1991 pyroclastic flows in the Sacobia watershed, Philippines: A study using remote sensing and geographic information system (GIS): unpublished M.S. thesis, International Institute for Aerospace Survey and Earth Sciences, Enschede, The Netherlands, 106 p.

Daligdig, J.A., Besana, G.M., and Punongbayan, R.S., 1991, Overview and impacts of the 1991 Mt. Pinatubo eruption [abs.]: Abstracts, GEOCON `91, Geological Society of the Philippines, 4th Annual Geological Convention, p. 8.

de Boer, J., Odom, L.A., Ragland, R.C., Snider, F.G., and Tilford, N.R., 1980, The Bataan orogene: Eastward subduction, tectonic rotations, and volcanism in the western Pacific (Philippines): Tectonophysics, v. 67, p. 251-282.

Delfin, F.G., Jr., 1983, Geology of the Mt. Pinatubo geothermal project: Philippine National Oil Company, unpublished report, 35 p. plus figures.

------1984, Geology and geothermal potential of Mt. Pinatubo: Philippine National Oil Company, unpublished report, 36 p.

Lillesand, T.M., and Kiefer, R.W., 1987, Remote sensing and image interpretation: New York, John Wiley and Sons, 721 p.

Newhall, C.G., Daag, A.S., Delfin, F.G., Jr., Hoblitt, R.P., McGeehin, J., Pallister, J.S., Regalado, M.T.M., Rubin, M., Tamayo, R.A., Jr., Tubianosa, B., and Umbal, J.V., this volume, Eruptive history of Mount Pinatubo.

Punongbayan, R.S., Tuñgol, N.M., Arboleda, R.A., Delos Reyes, P.J., Isada, M., Martinez, M., Melosantos, M.L.P., Puertollano, J., Regalado, T.M., Solidum, R.U., Tubianosa, B.S., Umbal, J.V., Alonso, R.A., and Remotigue, C., 1994a, Impacts of the 1993 lahars, long-term lahars hazards and risks around Pinatubo Volcano, in Philippine Institute of Volcanology and Seismology, Lahar Studies: Quezon City, PHIVOLCS Press, p. 1-40.

Punongbayan, R.S., Arboleda, R.A., Tuñgol, N.M., and Solidum, R.U., 1994b, Secondary explosions and secondary pyroclastic flows as rapid mechanisms for stream piracy: PHIVOLCS, unpublished report, 19 p.

Sabit, J.P., Pigtain, R.C., and de la Cruz, E.G., this volume, The west-side story: Observations of the 1991 Mount Pinatubo eruptions from the west.

Scott, W.E., Hoblitt, R.P., Daligdig, J.A., Besana, G., and Tubianosa, B.S., 1991, 15 June 1991 pyroclastic deposits at Mount Pinatubo, Philippines [abs]: Eos, Transactions, American Geophysical Union, v. 72, p. 61-62.

Scott, W.E., Hoblitt, R.P., Torres, R.C., Self, S, Martinez, M.L., and Nillos, T., Jr., this volume, Pyroclastic flows of the June 15, 1991, climactic eruption of Mount Pinatubo.

U.S. Army Corps of Engineers, 1994, Mount Pinatubo recovery action plan: Long term report: Portland, Oreg., 162 p. + 5 appendices.

Wolfe, E.W. and Hoblitt, R.P., this volume, Overview of the eruptions.


FIRE and MUD Contents

PHIVOLCS | University of Washington Press | U.S.Geological Survey

This page is <>
Contact: Chris Newhall
Last updated 06.11.99