The 1980 Eruptions of Mt. St. Helens. Lipman, P. and Mollineaux, D., editors. U. S. Geological Survey Professional Paper 1250, 1981. p. 815-820.

THE 1980 ACTIVITY—A CASE STUDY IN FORECASTING VOLCANIC ERUPTIONS

By ROBERT W. DECKER

ABSTRACT

Forecasting volcanic eruptions on a probabilistic basis appears encouraging. In general, the forecasts of the 1980 eruptions of Mount St. Helens were accurate, saving thousands of lives. However, in more specific terms of forecasting the timing, nature, and scale of eruptions, much needs to be learned. Eight significant eruptions of Mount St. Helens were preceded by multiple geological, geochemical, and geophysical phenomena that occurred from months to hours before the eruptions. Eight false alarms also occurred, but in almost all of these only one potential precursory phenomenon was observed.

INTRODUCTION

Forecasting the time, location, nature, and scale of volcanic eruptions is one of the most urgent aspects of applied volcanology. It is not a precise science, but it can be done on a probabilistic basis. Although only a few of the world's active volcanoes are being care fully studied, eruption forecasting attempts have been encouraging (Decker, 1978). These studies on active volcanoes involve two distinct but complementary approaches: (1) Documentation of the past eruptive activity of specific volcanoes by historic records, geologic mapping, stratigraphic studies, and radio metric dating techniques. This has been the rationale of the volcanic-hazard studies by Crandell and Mullineaux (1975). (2) Monitoring the changes in geophysical and geochemical data from volcanoes before, during, and after their eruptions. This is the major approach of the USGS Hawaiian Volcano Observatory to eruption forecasting (Tilling, 1977). The combination of these two approaches, historical record and data monitoring, is a logical basis for anticipating the future activity of specific volcanoes.

Some physical processes, such as celestial mechanics, are so regular that their histories can be ex trapolated with near certainty into the future. Predictions of the onset times, durations, and locations of solar eclipses can thus be done on a deterministic basis. Other physical phenomena, such as volcanic eruptions, are much more variable in nature. Extrapolating their histories and current processes into the future can only be done on a probabilistic basis. In the science of probabilistic forecasting, there are no sure predictions, only statistical gains over predic tions based only on historical averages. Pasteur said it well, "Chance favors the prepared mind."

Experience and insight are important in the interpretations of highly variable phenomena such as human nature, and they should not be discounted in the interpretation of highly complex and variable physical phenomena. However, what is needed in the science of probabilistic forecasting are objective criteria for interpreting the historical and monitoring data, and more basic research into the processes of volcanism so that the precursor events of eruptions can be interpreted in terms of their physical and chemical meanings. This should lead to better predictions of the nature, time, location, and magnitude of eruptive events.

____

816

The 1980 eruptions of Mount St. Helens make available a case study showing where eruption forecasting stands today and the directions it may take in the near future. In general, the forecasting of the Mount St. Helens activity was accurate and may have saved thousands of lives. In terms of the exact timing, nature, and scale of the huge May 18 eruption and avalanche, it was not accurate enough. Experience was painfully gained at the cost of 62 lives.

ACKNOWLEDGMENTS

This summary report could not have been written without the data and discussions generously provided by many persons who have studied the recent activity of Mount St. Helens. Much of the data is published in separate chapters in this volume. More specific credit goes to the following persons who, unless otherwise noted, are members of the USGS: Elliot Endo, Stephen Malone (University of Washington), and Craig Weaver, who provided seismic data; John Dvorak, Peter Lipman, James Moore, Arnold Okamura, and Donald Swanson, who provided deformation data; Jules Friedman and Hugh Kieffer, who provided ther mal data; Thomas Casadevall, David Harris, William Rose (Michigan Technological University), and Richard Stoiber (Dartmouth College), who provided volcanic gas-emission data; Robert Christiansen, Willie Kinoshita, and Donald Peterson, who provided eruption data; and Barry Voight (Pennsylvania State University), who provided data on the potential avalanche and eruption hazards. The research on Mount St. Helens has been a joint effort of many scientists and institutions working under difficult conditions. Their cooperation and generous exchange of data are gratefully acknowledged.

PAST ERUPTIVE ACTIVITY OF MOUNT ST. HELENS

Historic records indicate that eruptions occurred during 1831 to 1857, but aside from identifying Mount St. Helens as an active volcano, the records are too few to have any statistical meaning. Geologic studies and dating (Mullineaux and Crandell, this volume) have identified several eruptive periods hav ing diverse deposits in the last 4,500 years and have established some important patterns. Dormant intervals are of two lengths, 100-300 yr and 600-700 yr; the last two intervals prior to 1800 were the short variety. The eruptive products also show that pyroclastic explosions were common, and that prehistoric eruptions of Mount St. Helens affected large areas. This type of analysis led to Crandell and Mullineaux's forecast (1978, p. C1-C2) that Mount St. Helens was especially dangerous and likely to erupt within the next hundred years. Their words were prophetic:

Mount St. Helens has been more active and more explosive during the last 4,500 years than any other volcano in the conterminous United States. In the future, Mount St. Helens probably will erupt violently and intermittently * * * and will affect human life and health, property, agriculture, and general economic welfare over a broad area. The volcano's behavior pattern suggests that the current quiet interval will not last as long as a thousand years; instead, an eruption is more likely to occur within the next hundred years, and perhaps even before the end of this century.

The hazards report by Crandell and Mullineaux (1978) was accurate in predicting the areas affected by the May 18 eruption with regard to pyroclastic flows, air-fall ash, mudflows, and floods. The areas affected by the avalanche and directed blast are beyond the limits of any recognized prehistoric eruption of the same kind at the volcano and, therefore, were un precedented in scale in the 4,500 yr history upon which the hazards report was based.

Careful geologic mapping during several field seasons in the Mount St. Helens area gave Crandell and Mullineaux an additional benefit in dealing with hazard warning problems. They had become well ac quainted with the operations of the U.S. Forest Serv ice, lumber industry, and utilities in the area, and this knowledge led to much better credibility for their warnings during the phase of small eruptions prior to the May 18 catastrophe.

MONITORING THE ACTIVITY AT MOUNT ST. HELENS AND OTHER VOLCANOES

Geophysical and geochemical techniques are cur rently being used to monitor some active volcanoes in various parts of the world (UNESCO, 1971). However, most of the world's active volcanoes are not monitored. Current techniques include monitoring the following phenomena: seismicity, ground-surface

______

817

deformation, temperature, electric and magnetic fields, gas emissions, and the chemistry and mineral ogy of erupted lavas or pyroclastic materials. All of these techniques have been used during the recent activity at Mount St. Helens. Some techniques have worked better than others at forecasting, but all have yielded important data from the standpoint of obtaining a basic understanding of explosive volcanism.

PRECURSORY ACTIVITY AND ERUPTION POSSIBILITIES CONSIDERED

The first small phreatic eruption at Mount St. Helens occurred on March 27 at 1230 PST (Pacific Standard time). It was followed by hundreds of small steam and ash eruptions during March 27 to April 21 and May 7 to 14. The onset of these eruptions was preceded by 7 days of intense local seismic activity, including a period of exponential increase in seismic energy release on March 25. This seismic activity clearly signaled the high probability of the first eruption. The May 18 eruption occurred at 0832 PDT with out any distinct short-term warnings. However, the longer term precursory events were numerous and dramatic. The seismic swarm had continued with high and nearly constant energy release for 60 days. Short bursts of harmonic tremor began on March 31 and continued intermittently through April 5. Tremor recurred on April 12 and May 8.

Major visible deformation was first seen on March 27 and was monitored directly after April 23. The very large rates of deforrnation—1.5-2.5 m/day in an area of 1.5x2.0 km on the high north flank of Mount St. Helens—were of major concern. The close connection in time and space between the earthquake foci and the bulging area led most of the scientists studying the volcano to conclude that a shallow intru sion was taking place beneath an area just north of the summit. The possibility of a major avalanche and a lateral explosion were considered. Reports on the lateral explosions of Lassen Peak (Day and Allen, 1925), Bandai-san (Kuno, 1962), Bezymianny volcano (Gorshkov, 1959), Shiveluch volcano (Gorshkov and Dubik, 1970), and Mount Lamington (Taylor, 1958) were being reread by Mount St. Helens workers in late April and early May.

POSSIBLE ANALOGOUS ERUPTIONS

At Lassen Peak, Calif., after nearly a year of small steam explosions, a major eruption in 1915 formed an ash cloud that rose to an altitude of about 10 km, and an avalanche of hot debris destroyed an area 2x6 km on the east flank (Day and Allen, 1925). Bandai volcano in Japan erupted violently in 1888; the eruption was preceded only by a few felt earthquakes. An apparent hydrothermal explosion destroyed the summit, and formed a large horseshoe-shaped crater and a massive debris avalanche that covered 70 km2 and killed more than 460 people (Kuno, 1962). The gigan tic explosion of Bezymianny, Kamchatka, in 1956 occurred after 5 mo of smaller eruptions. It destroyed the top of the volcano, formed an ash cloud 45 km high, and devastated 500 square km in a huge lateral blast (Gorshkov, 1959). A similar but smaller lateral explo sion at Shiveluch, Kamchatka, in 1964 destroyed the volcano's summit and devastated 98 square km. No minor eruptions preceded the major blast, but recorded earthquakes had increased during the 10 mo prior to the explosion (Gorshkov and Dubik, 1970). Mount Lamington in Papua, New Guinea, erupted suddenly and violently in 1951. After only 6 days of felt earth quakes, landslides, and minor ash eruptions, a major pyroclastic flow devastated 176 km2 surrounding the volcano and killed 3,000 people (Taylor, 1958).

The possibility that a lava dome might extrude at Mount St. Helens without much accompanying damage, as happened at Syowa-Sinzan in Japan in 1944-45 (Kuno, 1962), was also considered, as well as the possibility that the apparent intrusion at Mount St. Helens might cease before it caused any major avalanche or eruption. One of the most perceptive possibilities suggested between March 27 and May 18 was by Barry Voight of Pennsylvania State Univer sity. He compared Mount St. Helens to the areas of large historic avalanches, at Gros Ventre, Wyo., in 1925, and Madison Canyon, Mont., in 1959, and decided that a potential failure of the north flank of Mount St. Helens could generate an avalanche of 1-3 km3. Voight (written commun., May 1, 1980) noted that:

A catastrophic event of the kind observed at Bandai-san—in which an explosively motivated fragmental flow devastated an area of more than 70 square km must be regarded as a legitimate possibility, particularly in view of the enhanced hydraulic pressure conditions implied by frequent summit steam explosions and the relatively high level of released seismic energy.

________

818

On the more prosaic side, a bulging slope associated with rock creep (increased tilt) may lead to an increase of rockfall hazards from exposed rock areas such as Goat Rocks, increased snow avalanche hazard, and increased risk of glacier falls.

In general, for most natural catastrophes, the larger the event, the more rarely it occurs. Therefore, most of the scientists studying Mount St. Helens in early May considered that a significant avalanche and eruption were distinctly possible, but that extremely large events were not very probable. Even so, the possibility of danger was clearly evident from prehistoric eruptions of Mount St. Helens, historic eruptions at other volcanoes, and the ongoing high rate of seismicity and deformation. These factors were considered by authorities responsible for land management around the volcano in their decision to continue restricting access to the volcano, even in the face of mounting public pressure to open the Spirit Lake recreation area.

PRECURSORY ACTIVITY AFTER MAY 18

Smaller but significant eruptions after May 18 were also preceded by geological, geochemical, and geo physical signals. The May 25 eruption at 0228 PDT was preceded by 7 days of harmonic tremor and by smaller eruptions as much as 10 hr in advance. Harmonic tremor for 9 hr and a smaller eruption 2 hr in advance also preceded the June 12 eruption at 2110 PDT. No tremor preceded the July 22 eruption, but the eruption was marked by increasing local seismicity for 8 hr before, as well as by several centimeters of outward deformation of the crater rampart during July 18-22. A pattern of decreasing CO2 emission and CO^2/SO^2 ratios for 9 days was also observed prior to July 22. Harmonic tremor for 4l/2 hr preceded the August 7 eruption at 1624 PDT. The CO^2 emission and CO^2/SO^2 ratios also showed a decrease for 5 days prior to August 7.

The October 16 eruptions, which began at 2158 PDT, were preceded by more than 30 hr of increased local seismicity, 12 days of small northward displace ments of the crater rampart, and a decrease of CO^2 emission and CO^2/SO^2 ratios between October 9 and 15.

On December 13-14, a small sector of the lava dome was apparently blown out. Because of bad weather, this eruption was not witnessed. Local seismicity began to increase on December 25 and reached a maximum by December 26-27, and the

     Table 111.—Geological, geochemical, and geophysical changes that occurred before significant eruptions of Mount St. Helens
             [hr, hours; leaders (---) phenomenon did not occur or was not detected]
______________________________________________________________________________
		Onset of changes before eruptions Eruptions
 					   		    			      Thermal    Gas       Small
		      SeIsmicity   Tremor   Deformation   anomaly emission  eruptions
_____________________________________________________________________________		  

Mar. 27,    7 days    	   ---		---			  ---		---        ---
  1230 PST. 			
May 18,     60 days    49 days (1) 53 days        53 days   ---     53 days (1)  
  0832 PDT. 	
May 25,		---     	 7 days		---	-		  ---		---	 	10 hr	
  0228 PDT. 
June 12,	---     	 9 hr		---		      ---		---	 	2hr
  2110 PDT.
July 22,    8 hr      	 ---       	4 days		  ---       9 days  ---
  1714 PDT.			
Aug 7,      ---          41/2 hr	---			  ---       5 days	---
  1624 PDT.
Oct. 16,    30 hr  		 ---		12 days       ---       7 days   6 days
  2158 PDT.
Dec. 27-28  2 days     	 ---		14 days		   ---      ---      14 days
____________________________________________________________________________
(1) Intermittent .
___

819

crater rampart moved tens of centimeters northward. On December 28, renewed growth of the lava dome was observed. This growth continued through January 2-4, 1981, doubling the volume of the dome. The summary of eruptions in table 111 shows that more than one precursory phenomenon was evident before the beginning of each eruption. Most of the precursors were clearly recognized and were reported before the eruptions. The pattern of tremor before the May 25 eruption and the pattern of gas emissions before the July 22 eruption were recognized as ap parent precursors after the eruption had occurred. Even though there was no short-term warning of the May 18 eruption, five different, long-term phe nomena indicated the possibility of a future eruption. Precursory events not followed by eruptions (false alarms) did occur, and these probably are inevitable in any probabilistic forecasting system. Table 112 lists eight occurrences of precursory-type events that were not followed by significant eruptions. Only before August 15 was there more than one potential precursory event.

Although there is ample evidence that several phenomena, which can be observed and measured, give warnings of impending eruptions for hours to months in advance, no single pattern has clearly repeated itself before all or even most of the eruptions. Basic scientific research, as well as human experience and judgment, are still needed in large measure.

THE FUTURE

The major eruptions of Mount St. Helens in 1980 not only indicate that science and technology can par tially anticipate and ameliorate volcanic hazards, but also give some indication of priorities for future in vestigations. These are as follows: (1) carry out geologic mapping, stratigraphic studies, and radio metric dating to establish the geologic history of each potentially active volcanic center or area; (2) install limited networks of seismometers to monitor these volcanic centers or areas; (3) make baseline geophysical and geochemical measurements on these volcanic centers or areas for comparison with future changes; (4) install additional seismometers and other continuous monitoring devices at volcanic centers or areas when increased seismicity or other changes suggest that increasing activity is occurring.

Because of the generally long repose intervals between eruptive periods of explosive volcanoes, several volcanic centers or areas should be studied and monitored from one observatory. Most impor tant of all, the geologic, geophysical, and gochemical studies on several volcanoes should be done by a group of scientists at one headquarters. Volcanology is such a broad field of research that many of its problems can only be solved by scientists from different disciplines—geology, chemistry, physics, and mathematics—working closely together.

Table 112.—Precursory phenomena that were not followed by significant eruptions at Mount St. Helens
         [Leaders (---), phenomenon did not occur or was not detected]
_____________________________________________________
				Type and operation
Date						  Thermal	    Gas	   		Small
	Seismicity  Tremor  Deformation anomoly	emission	eruptions

June 3      ---		Several	---	   ---		---		    ---		
	                 hours 
June 26    ---		--do---	---	   ---		---		    ---		
July 28     ---		---		---	   ---		---	        20 min.
Aug. 15   ---		---		---	   ---		2 days    	Several
                  								         minutes -
Aug 25    ---		---		---	   ---		4 days		---	
Sept 24   ---		---		---	   ---		---			4 hr	
Oct. 1	  ---		---		---	   ---		---		   Duration
                      									unknown
Nov18-19  ---		---		---	   ---		---		   Do
___

820

REFERENCES CITED

Return to The Electronic Volcano