NUUANU LANDSLIDE

Recent studies have shown that the collapse of large volcanoes due to gravitational instability plays an important role in shaping volcanic environments. Detailed offshore bathymetric surveys of the Hawaiian Ridge (Moore et al., 1994), Reunion Island (Lenat et al., 1989), and the Canary Islands (Masson et al., 2002) reveal extremely large landslides. In the case of Hawaii, one of the Nuuanu Landslides caused by the collapse of Koolau Volcano on Oahu extends more than 200 km from the island. In the Canary Islands, the debris extends to 30 km. Landslides on both island chains might have generated huge tsunamis (Moore, 1964; Moore and Moore, 1988; Moore et al., 1989). Herrero-Bervera et al. (2002) estimated the age of Nuuanu Landslides at 2.1-1.8 Ma. However, the size, the age, and the number of Nuuanu Landslides are still in question. Site 1223 is ~300 km from Oahu and ~100 km to the northeast of the presently defined Nuuanu Wailau debris field (Figs. F1, F25).

Geological Setting

The Nuuanu Landslide, which broke away from the northeast flank of Koolau Volcano on the island of Oahu, is the largest Hawaiian landslide. It is a debris avalanche that contains enormous blocks such as the Tuscaloosa Seamount, which is ~30 km long, 17 km wide, and at least 2 km tall. The landslide is spread over a 23,000-km2 area (Normark et al., 1993; Naka et al., 2000), with distal portions extending up the Hawaiian Arch. To reach the upper portion of the arch, the target site for drilling, the landslide would have had to traverse the deep moat on the northeast side of Oahu and travel over 100 km uphill.

Reaching the landslide deposit by gravity or piston coring has proven difficult because the deposit is overlain by a carapace of younger debris such as turbidites and associated deposits. Thus, the thickness and depositional history of the landslide are poorly known. Prior to drilling, the thickness of the distal portion of the landslide was estimated to be from 1 to 100 m (Rees et al., 1993; Naka et al., 2000). Similarly, the age of the landslide is poorly constrained, although it apparently occurred near the end or after the formation of the Koolau Volcano, which has surface flows with ages 1.8-2.6 Ma based on K-Ar dating by Doell and Dalrymple (1973).

Scientific Objectives

The objectives of drilling at the Nuuanu Landslide site are

  1. To resolve whether the Nuuanu Landslide occurred as a single distinct event or as multiple collapses;
  2. To determine the age of the landslide(s);
  3. To determine the thickness of the landslide deposit at the distal site and to obtain ground truth for the available seismic data in order to estimate the volume of the slide;
  4. To study the deposition history of the landslide(s); and
  5. To gain insight into potential hazards related to giant landslides on the flanks of ocean island volcanoes.

Operations

The JOIDES Resolution departed from the Honolulu Harbor at 1404 hr on 20 December for the Nuuanu Landslide Site NU-1 (ODP Site 1223). The 170-nmi voyage to Site 1223 required 17.0 hr at an average speed of 10.0 kt.

Hole 1223A

Hole 1223A was spudded with the APC at 2030 hr on 21 December at a depth of 4235.1 m (4245.8 mbrf). We took two APC and four XCB cores. We cored 41 m and recovered 23.54 m of core (57.4% recovery), with 12.7 m cored and 10.87 m recovered (85.6% recovery) with the APC and 28.3 m cored and 12.67 m recovered (44.8% recovery) with the XCB (Table T5). After two APC cores, we switched to XCB coring. The use of the XCB system at these shallow depths and the long time needed for coring was unexpected, as was the presence of lithified volcanic rocks. Core 200-1223A-5X was advanced only 1.0 m when it was recovered because there were indications of jamming. Core 200-1223A-6X was advanced 8.0 m to a depth of 41.0 mbsf when it was recovered because the time on site had expired. The drill bit cleared the rig floor at 0130 hr on 23 December, and we departed for Site 1224 (H2O).

Principal Results

Major Discoveries

The core recovered from Hole 1223A (Fig. F26) answered several of the questions posed precruise, but also resulted in some unexpected discoveries. One of the objectives of coring at this site was to determine if the Nuuanu Landslide occurred as a single or as a multistage event as indicated by the number of turbidites recovered. Several unconsolidated volcaniclastic turbidites of varying thickness were recovered in the first two cores in Hole 1223A. At least seven of these were >10 cm thick at 0.86-1.01, 2.11-2.53, and 3.76-3.99 mbsf in lithologic Unit 1 and the four or more turbidites that comprise nearly all of lithologic Unit 2, spanning the interval from 5.11 to 7.32 mbsf. Other turbidites that were <1 cm thick were also recovered. Paleomagnetic data indicate that all but the uppermost turbidite have an age between 1.77 and 1.95 Ma. The top turbidite has an estimated age between 0.99 and 1.07 Ma.

A surprising discovery was the recovery of two crystal vitric tuff layers. Preliminary geochemical analyses indicate these tuffs are geochemically similar to Hawaiian tholeiitic basalts. Olivine in the vitric tuff is fresh. Kink banding and fibrous structures were observed in some olivine. The fibrous structure may have been caused by crystallization of hematite. The kink banding and fibrous olivine may have been derived from deformed dunite cumulates from deep in the lower crust or upper mantle. Paleomagnetic data indicate that the vitric tuffs are older than 1.95 Ma.

Another important result is the identification of the minerals anhydrite, paragonite, and wairakite in sediments just below the deeper vitric tuff. These are hydrothermal minerals and are stable at a temperature range from 150 to 350C. This suggests that considerable heat was involved when the crystal vitric tuff was deposited. Tentative interpretations for the origin of crystal vitric tuffs are given in "Lithology."

Lithology

We identified 14 distinct lithologic units (Fig. F26):

Unit 1 (0-5.11 mbsf) contains yellowish brown clay and volcaniclastic turbidites.
Unit 2 (5.11-7.32 mbsf) consists of volcaniclastic turbidites only.
Unit 3 (7.32-7.90 mbsf) is a thin layer of dark-brown clay.
Unit 4 (7.90-10.78 mbsf) is unconsolidated black sand.
Unit 5 (12.70-15.06 mbsf) is a crystal vitric tuff.
Unit 6 (15.06-15.29 mbsf) is bioturbated claystone.
Unit 7 (15.29-16.10 mbsf) is volcaniclastic sandy siltstone.
Unit 8 (22.30-22.80 mbsf) is volcaniclastic silty claystone with carbonate granules.
Unit 9 (22.80-22.91 mbsf) is volcaniclastic claystone.
Unit 10 (22.91-24.92 mbsf) is volcaniclastic silty claystone.
Subunit 11A (32.00-33.00 mbsf) is altered vitric tuff, highly disturbed by drilling.
Subunit 11B (33.00-36.99 mbsf) is palagonatized crystal vitric tuff.
Unit 12 (36.99-37.47) is volcaniclastic silty claystone with scapolite amygdules.
Unit 13 (37.47-38.31 mbsf) is volcaniclastic silty claystone.
Unit 14 (38.31-38.70 mbsf) is volcaniclastic clayey siltstone.

Coring gaps of several meters exist between some of the cores, so additional units may exist or those identified may be thicker by several meters.

Volcanic material is present throughout the stratigraphic column. Although we identified 14 distinct lithologic units, they may be grouped into three main types of lithologies:

  1. Unconsolidated clay and volcanic sediments,
  2. Weakly consolidated claystones and siltstones, and
  3. Crystal vitric tuffs.

The turbidites are concentrated in the upper 12.7 m of the core. They are dominated by a volcanic fraction that varies from 45% to 100%. The main constituents are, in order of increasing abundance, glassy shards, vitric fragments, olivine phenocrysts and clasts, plagioclase, palagonitized glass, lithic fragments, and clinopyroxene clasts. An MgO-rich olivine and Ca-rich plagioclase composition indicates equilibrium with mafic magmas. The mudstones and siltstones are found both in the middle and the bottom of the sections cored; they have high contents of clay, indicating detrital sources, and they have a low but variable volcanic fraction (~1%-25%). The silty claystone at the bottom of the deepest crystal vitric tuff (Subunit 11B) is characterized by the presence of relatively large (up to 3.5 mm) amygdules filled by anhydrite, a sulfate associated with hydrothermally altered basic rocks. The vitric tuffs were found just below the sand and at the top and in the middle of the mudstones and siltstones cored. The "nonvolcanic" fraction is made up of claystone clasts, micritic clasts, and, more rarely, radiolarians (generally <1%).

Whole-rock ICP-AES analyses were conducted on the two vitric tuffs, as well as several of the siltstones and claystones. They have high MgO concentrations (12.4-15.8 wt%), which are not surprising because of the high percentage of olivine present. The tuff's major elements are SiO2 (47.6-49.6 wt%), TiO2 (~2 wt%), Al2O3 (11.3-11.9 wt%), Fe2O3 (11.3-12.7 wt%), MgO (12.4-15.8 wt%), CaO (6.74-7.17 wt%), Na2O (2.19-3.07 wt%), K2O (0.47-0.83 wt%), and P2O3 (0.16-0.23 wt%). Their trace elements are Ba (50-70 ppm), Sr (22-330 ppm), Y (~20 ppm), Zr (~120 ppm), and Ni (~430-580 ppm). Siltstones and claystones show similar values to those of the tuffs. The geochemistries of whole-rock crystal vitric tuff, siltstone, and claystone were compared with the basalt glass geochemistry of MORB, Kilauea tholeiitic basalt, Haleakala alkali basalt, North Arch alkali basalt, and Koolau tholeiitic basalt. The crystal vitric tuff, siltstones, and claystones have the most in common, geochemically, with the Hawaiian tholeiitic lavas. There are some ambiguities in the measurements and samples, however, because the tuff contains clay minerals, which may affect the chemical compositions.

Several of the claystones and siltstones contain effervescing white amygdules. XRD analysis of the filling from interval 200-1223A-6X-4, 0-20 cm, gave a complex spectrum. The major components of the material are anhydrite and paragonite. Additional components include wairakite and possibly pumpellyite of varying compositions as a minor component. Grains from two intervals in the turbidite were analyzed by XRD (intervals 200-1223A-1H-5, 94-95 and 114-115 cm). The upper interval has brown grains with dominant XRD peaks at wavelengths consistent with phillipsite. In addition, some of the smaller peaks have spectra consistent with clay minerals, mainly smectite and illite and minor plagioclase. Therefore, the composition of the material analyzed is mainly phillipsite with clay minerals and minor plagioclase. The lower interval has white granules identified as calcite by XRD analysis.

Wairakite was originally found in hot springs in the geothermal fields of Wairakei in New Zealand and of Onikobe in Japan (Miyashiro, 1973). It has also been found in hydrothermal areas in the Mariana Trough (Natland and Hekinian, 1982). Wairakite is stable at a temperature range from 200 to 300C. This suggests that considerable heat was involved when the crystal vitric tuff was deposited sometime thereafter. Anhydrite and paragonite are found together in altered basalts in the stockwork of the Trans-Atlantic Geotraverse (TAG) hydrothermal mound on the Mid-Atlantic Ridge where they formed at temperatures from 210 to 390C (Honnorez et al., 1983).

Some of the olivine clasts in the tuffaceous layers show kink banding. Many mantle rocks such as lehrzolites and dunites show kink banding in olivine crystals. Also, tectonized peridotites often show such microstructures (e.g., Ishii et al., 1992). The kink bands are formed where shear is applied to the olivine (e.g., Kirby, 1983). Dislocations by shear stress in the olivine crystals generate the kink bands. Another interesting feature is masses of fibrous lines seen in some olivine crystals. Iron oxide exsolution may be the cause of this structure. The presence of kink bands and fibrous structures indicates that olivine crystals in the crystal vitric tuffs were subjected to tectonic deformation.

Other Observations

We measured bulk density using GRA, magnetic susceptibility, natural gamma ray (NGR), and compressional velocity (VP) on whole-core sections with the multisensor track (MST). Moisture and density and compressional wave velocities were also measured on selected individual samples. The yellowish brown clay in Core 200-1223A-1H has 83% porosity and a compressional velocity of 1.5 km/s. GRA density values from 0 to 10 mbsf gradually increase downhole from 1.2 to 2.2 g/cm3. The GRA densities agree with the bulk densities of the individual measurements, except for Core 200-1223A-3X, where GRA densities of the vitric tuff in Core 200-1223A-3X average 1.8 g/cm3, but bulk densities are slightly higher. Corresponding compressional velocity values for the vitric tuff in Core 200-1223A-3X average ~3.3 km/s. Bulk density, grain density, and compressional velocity for the siltstone in Core 200-1223A-4X are 1.6-2.0 g/cm3, 2.8 g/cm3, and 1.8-3.3 km/s, respectively. The vitric tuff in Core 200-1223A-6X has a 2.1 g/cm3 bulk density, a 2.6 g/cm3 grain density, and a 4.0 km/s compressional velocity. Grain densities of vitric tuff in Core 200-1223A-6X are lower than those in Core 200-1223A-3X. This is caused by the higher levels of alteration in Core 200-1223A-6X. The compressional wave velocities of the crystal vitric tuffs in lithologic Unit 5 (just below sand) and lithologic Unit 11 are ~3 and ~4 km/s, respectively. However, the bulk densities for both are similar, at ~2.2 g/cm3. The compressional wave velocity for the upper crystal vitric tuff deviates from the generally expected compressional velocity-density relationship (e.g., Johnston and Christensen, 1997), possibly because of the weak consolidation of lithologic Unit 5. Compressional velocity and density increase with depth within the turbidites of lithologic Units 1 and 2, and the gradients can be associated with the graded bedding in the turbidites. Using this velocity and density increase with depth, we can identify the presence of the turbidite layers.

The magnetostratigraphy for Hole 1223A appears to record all the major chrons and subchrons from Chron C1n (the Brunhes Chron; 0.0-0.780 Ma) through Chron 2r (1.95-2.581 Ma). The Brunhes normal polarity interval spans only the top 14 cm of Core 200-1223A-1H, which is thinner than expected by ~1 m based on prior piston coring in the vicinity. Thus, we may not have recovered the very upper meter or so of the sedimentary section, or sedimentation rates may vary locally. The top and base of the normal polarity interval interpreted as Subchron C1r.1n (Jaramillo Subchron; 0.99-1.07 Ma) are at 0.79 and 1.23 mbsf, respectively. The top and base of the normal polarity interval interpreted as Chron C2n (the Olduvai Chron; 1.77-1.95 Ma) are at 2.02 and ~7 mbsf, respectively. All recovered core below ~7 mbsf appears to be of reversed polarity, which is interpreted to be the upper part of Chron C2r, possibly with the entire interval lying within Subchron C2r.1r (1.95-2.14 Ma).

We did not have a paleontologist on board, but Bob Goll and John Firth from ODP-TAMU did postcruise examinations of sediment samples for radiolarians and calcareous nannofossils, respectively. No calcareous nannofossils were found. Of six samples that were examined for radiolarians, two could be used for dating and indicate an early Eocene age. This conflicts with the magnetostratigraphic dates. We consider that reworking is likely in these landslide deposits and that the early Eocene age is not representative of the depositional age of the units.

Microbiological analyses from Site 1223 were conducted on sediments and tuffs. Most probable number (MPN) series were prepared from all samples in order to determine the concentration of sulfate-reducing as well as fermentative bacteria. From Site 1223, 25 distinct microbial colonies could be isolated to pure cultures and could be further characterized as facultatively anaerobic organisms forming stable cell aggregates under appropriate conditions.

Interpretation

As noted above, three types of material—unconsolidated clay and volcanic sediments, weakly consolidated claystones and siltstones, and crystal vitric tuffs—were recovered in the cores from Hole 1223A. The first type consists of interbedded layers of pelagic clays and volcaniclastic sediments thought to be turbidites. These turbidites potentially originated from one of the Hawaiian Islands. The sand, which is present in the upper 15 m of the section, has a high vitric component suggesting a volcanic source. The claystones and siltstones, on the other hand, which are present in both the middle and the bottom of the section cored, have a lower percentage of coarse volcanic material.

We considered two hypotheses for the origin of these claystones and siltstones. The first describes them as sedimentary rocks derived from detrital material from the Hawaiian Islands and emplaced by turbidity currents. The second hypothesis describes them as volcanic tuffs derived from submarine pyroclastic flows either from the Hawaiian Islands or from a local vent. Unwelded submarine pyroclastics in general tend to have a massive to poorly bedded and poorly sorted lower unit and an upper thinly bedded unit (Fiske and Matsuda, 1964; Bond, 1973; Niem, 1977). Fisher and Schmincke (1984) state that it is very difficult to distinguish between the deposits of these two types of events.

The vitric tuffs that are present below the sands and at the top and in the middle of the claystones and siltstones are also problematic. Hypotheses for their origin must address the following questions:

  1. Why are they indurated at such a shallow depth?
  2. Why are they so glass rich?
  3. Why do they have so much fresh olivine?
  4. Why is their general character tholeiitic?
  5. Were they warm to fairly hot at, or shortly after, deposition?

Alteration and cementation in the lower tuff and the transformations in sediments at its lower contact probably occurred at a high temperature, perhaps 150-320C, consistent with zeolite metamorphic conditions and the occurrences of wairakite/analcime, paragonite, and anhydrite described above.

Formation of palagonite was not isochemical. In this case, it involved loss of both CaO and Sr and addition of K2O to the bulk compositions of the rocks (see "Geochemistry" in "Principal Results" in "Hawaii-2 Observatory"). These exchanges required substantial flow of fluids derived from seawater through the porous tuffs. The presence of anhydrite and paragonite, a sodic mica, suggests that some of the fluids were brines.

We consider two origins for the crystal vitric tuffs, a Hawaiian Island source and a local source. If the source of the tuffs was local, they were presumably produced by igneous activity, either intrusion or extrusion, that has not yet been documented for this part of the Hawaiian Arch. Neither lava fields nor fissures associated with such hypothetical volcanism are evident in seismic data of the smoothly sedimented crest of the arch near Site 1223, but no high-resolution bathymetry data have been taken of the surrounding area. The nearest known young volcanism on the arch occurred at the North Arch volcanic field, some 200 km to the northwest. There, lavas and tuffs are alkalic olivine basalts, basanites, and olivine nephelinites (Dixon et al., 1997), very unlike the tholeiitic precursors to the indurated tuffs and other sedimentary rocks of Hole 1223A. Similar volcanism near the drill site thus would only coincidentally have driven hydrothermal fluids through the sediments that we cored.

A possible scenario for a Hawaiian source is that a very large eruption of Hawaiian tholeiite occurred when a deep magma reservoir was breached by catastrophic failure of the flank of a volcano, similar to the 1980 eruption of Mount Saint Helens (Fig. F27) (Moore and Albee, 1981). This may have occurred on Oahu when the northeast flank of Koolau Volcano collapsed, producing the giant Nuuanu debris avalanche. Sudden decompression caused pressure release, vesiculation, and expansion of the magma. The magma erupted as a directed blast and passed over the collapsing blocks now strewn on the seafloor as a submarine pyroclastic debris flow that reached over the Hawaiian Arch. In this scenario, if the material reached the area of Site 1223 (300 km away) quickly enough in a bottom-hugging density flow, it may have retained enough heat to cause the alteration and induration of the two crystal vitric tuffs. Water surrounding a hot pyroclastic flow may become vaporized, creating a water vapor barrier around the flow which helps to insulate the flow and prevent mixing (Kato et al., 1971; Yamazaki et al., 1973).

Subaerial pyroclastic flows have been observed with velocities ranging from 14 km/hr (Tsuya, 1930) to 230 km/hr (Moore and Melson, 1969). They have traveled great distances (>100 km) and surmounted obstacles >600 m high (Fisher and Schmincke, 1984). Their mobility has been attributed to several factors, namely exsolution of gas from glassy particles, gas being released when particles are broken, and the heating of the medium causing thermal expansion (e.g., Sparks, 1979). The gas reduces the friction between particles, allowing the flow to travel faster. In addition to moving at tremendous speeds, pyroclastic flows are very good at retaining heat. Boyd (1961) calculated that cold air has a minimal effect on a hot pyroclastic flow. Therefore, a hot pyroclastic flow may remain at almost magmatic temperatures during transport and even after deposition.

The above applies to subaerial pyroclastic eruptions; subaqueous pyroclastic flows are less well understood. A massive, coarse-grained, subaqueous pyroclastic flow deposit over 4.5 m thick and extending up to 250 km from its source was recovered in the Grenada Basin, Lesser Antilles. Carey and Sigurdsson (1980) interpreted the deposit to be a debris flow that originated when a hot subaerially erupted pyroclastic flow entered the ocean. The debris flow incorporated pelagic sediment and seawater, decreasing internal friction, giving the flow great mobility and the ability to suspend large fragments. Thermal remanent magnetism of a similar deposit (Pliocene-Miocene in age) was interpreted by Kato et al. (1971) to have been deposited at temperatures of ~500C.

An alternative is that the tuffs were deposited containing some heat. Nevertheless, the amount of heat in a few meters of such materials is unlikely to have driven fluid flow for very long, at least if the deposit was small. On the other hand, a widespread blanket of hot material deposited suddenly might have acted as a compressive load on uncompacted surface sediments and, where sufficiently thick, as an impermeable barrier to fluids mobilized by sudden compaction. The fluids, thus forced to flow laterally, may have sustained high temperatures at the base of the tuff for some time and there produced the most concentrated effects of alteration and contact metamorphism. A somewhat similar effect was postulated for the pattern of fluid flow at the top of basaltic basement, beneath ~100 m of volcaniclastic turbidites in the eastern Mariana Trough at Deep Sea Drilling Project Site 456 (Natland and Hekinian, 1982). Greenschist facies hydrothermal conditions were reached, and both wairakite and cristobalite formed in the sediments at the contact.

This hypothesis provides a mechanism for directing fluid flow and thus concentrating the most pronounced alteration effects in the tuff beds themselves, rather than in adjacent sediments. This probably would not have been the case if hydrothermal flow was directed along vertical fissures associated with local igneous action. Admittedly, the geometry of fluid flow in variably permeable sediments is difficult to extrapolate over long distances from the vantage of a single hole.

At Site 1223, the two tuffs, although about equally indurated, experienced alteration that was different either in type or in degree. The upper tuff is not palagonitized, although it is cemented by clay minerals and zeolites, and almost all of its original glass is still fresh. Probably the essential difference was temperature—lower for the upper tuff—although the upper tuff may have experienced less fluid flow as well. Lower temperature and reduced fluid flow should mean the same thing, less heat was available to drive fluids, whether or not it was derived locally or from a more distant source.

Both the whole-rock ICP-AES analyses and the preliminary electron microprobe analyses of individual glass fragments within both crystal vitric tuffs (Unit 5 and Subunit 11B) support a Hawaiian Islands source. Unlike the whole-rock ICP-AES analyses, the effects of olivine accumulation and alteration do not complicate the glass geochemistry. Therefore, it is easier to compare the glass geochemistry from the tuffs with the geochemistry from other potential sources (i.e., Hawaiian Islands and North Arch volcanic field). The glass geochemistry suggests that the tuffs have tholeiitic basalt compositions and are similar to Hawaiian tholeiitic basalts.

Futhermore, the glasses in both tuffs have low S concentrations (majority <300 ppm) (S.B. Sherman, unpubl. data). Sulfur concentrations can be used to estimate depths of eruption (e.g., Moore and Fabbi, 1971; Dixon et al., 1991). Low S concentrations (<200 ppm) are indicative of significant degassing and predominantly result from subaerial eruptions. Deep submarine glasses with low S contents were recovered from Puna Ridge, which is a submarine extension of Kilauea Volcano. The low S abundances for these glasses were attributed to mixing between lavas that had degassed at or near the subaerial surface with the rift lavas that were erupted underwater (Clague et al., 1995). Therefore, the low S concentrations in the glasses in the tuffs suggest that they were erupted subaerially or at least mixed with subaerial lavas. A subaerial eruption supports a Hawaiian rather than a North Arch source for the tuffs. Not only is the North Arch volcanic field submarine, the majority of the glasses from the North Arch volcanic field have S contents >300 ppm (Dixon et al., 1997). It may be difficult to imagine the source for the tuffs from Site 1223 as one of the Hawaiian Islands, but the geochemical evidence indicates it is the most likely option.

Further discussion of the provenance of the Site 1223 materials is given in "Provenance and Petrogenesis".

Provenance and Petrogenesis

It is necessary to determine the provenance of the Site 1223 materials in order to confirm whether these materials did or did not originate in the Nuuanu Landslide event(s). An island provenance would be a prerequisite if this were indeed the case. On the other hand, a local source of heat seemed required to explain the induration of the two welded tuffs so near the seafloor. A local provenance for the tuffs, perhaps a nearby seamount, would be consistent with this hypothesis. Accordingly, we compared bulk compositions of samples from Hole 1223A with varieties of Hawaiian basaltic rocks, with lavas of the North Arch volcanic field, some 200 km to the northwest, and with both normal and enriched abyssal basalts (N-MORB and E-MORB). Our initial petrographic interpretation was that the olivine-bearing glass shards in the tuffs resembled types of Hawaiian picritic tholeiite, but the issue still remained whether tholeiite itself might have erupted recently along the Hawaiian Arch and provided a local source for the tuffs essentially identical to the islands.

Existence of a significant component of aluminous detrital clay, produced ultimately by subaerial erosion, confirmed by chemical analyses, sets most of these issues at rest. The source of the aluminous component in all of the sediments and tuffs was clearly the islands, making it extremely unlikely that a separate local source provided the volcanic glass shards and associated minerals and lithic fragments. But what can now be said in more detail about the composition of source materials? Tholeiitic basalt is, of course, the most voluminous of Hawaiian lava types (e.g., Macdonald and Katsura, 1964; Clague and Dalrymple, 1987). However, alkalic olivine basalts, basanites, and olivine nephelinites erupted during both the earliest and latest stages of Hawaiian volcanism, and it is possible that these at least contributed volcaniclastic materials to the sedimentary succession at Site 1223.

Major oxide discriminant diagrams (Fig. F28A, F28B) show the similarity of analyzed samples from Hole 1223A to Hawaiian tholeiite, represented by basalt glasses from Kilauea Volcano and its undersea extension, Puna Ridge (Clague et al., 1995). The diagrams also show strong differences between our samples (as well as Kilauea tholeiites) and Hawaiian alkalic olivine basalts, basanites, and olivine nephelinites from three localities—the North Arch volcanic field (Dixon et al., 1997), the Honolulu Volcanic Series of Oahu (Jackson and Wright, 1970; Clague and Frey, 1982), and the Hana Volcanic Series of Haleakala Volcano on the island of Maui (Chen et al., 1991). Data are also plotted for a representative suite of abyssal tholeiites from the East Pacific Rise (J. Natland, Y. Niu, and P. Castillo, unpubl. data). This suite includes a wide range of primitive and differentiated abyssal tholeiites (N-MORB). Samples from Site 1223 clearly differ from these as well.

Because we compare bulk sediment compositions with compositions of glasses from Kilauea and the East Pacific Rise, the effects of abundant olivine in Site 1223 samples need to be taken into account. Olivine forms 9-13 wt% of the mode of the vitric tuffs. If its composition is about Fo85, the effect of subtracting 13 wt% of olivine from the bulk composition of a tuff with 10.5 wt% Al2O3, 47.5 wt% SiO2, and 11 wt% iron as Fe2O3 is shown by the arrows in Figure F28A and F28B. This is about the composition of Group 1 tuff in Figure F29 with the least amount of detrital clay in its makeup. The tip of the arrows in Figure F28 therefore is approximately that of an aphyric basalt or basalt glass still very closely resembling the composition of Kilauea tholeiite. In amounts of these oxides, it is far from the compositions of Hawaiian alkalic basaltic lavas and from MORB.

In Figure F28C and F28D, compositions of samples from Hole 1223A are explicitly compared with E-MORB glasses, using a compilation drawn from the literature, and with eight analyses of magnesian tholeiites and tholeiitic picrites from Koolau Volcano (Frey et al., 1994). E-MORB resembles N-MORB, except in being slightly more aluminous. Most of the several Koolau lavas have slightly higher SiO2 contents and lower iron as Fe2O3 than Kilauea tholeiitic glasses. In these diagrams, the effects of the addition of detrital clays and of authigenesis make it difficult to describe samples from Hole 1223A as more like one of the Hawaiian volcanoes than the other.

Figure F30 provides additional hints about the particular Hawaiian provenance of samples from Hole 1223A. Again, data for MORB, E-MORB, Kilauea-Puna Ridge, and Koolau Volcano are plotted for comparison. Again, samples from Hole 1223A resemble Hawaiian tholeiites rather than MORB or E-MORB. Several of the tuff samples from Hole 1223A, including those falling in Group 1 in Figure F29A and F29B, have higher SiO2, lower Ba, and lower Zr than Kilauea glasses at given MgO content (Fig. F30). In all these respects, they more closely resemble Koolau. For SiO2 and Ba, these estimations may be complicated by the presence of some detrital clay, and in two samples, Ba clearly is too high. However, the comparison holds for those samples with the least amounts of clay or, that is, highest CaO contents. In addition, Zr is an element that is usually unaffected by alteration. It is also usually precisely and consistently measured from one laboratory to the next. The measurements should give a relatively solid estimation of its original concentration in volcanic glass and lithic fragments in the sediments and tuffs, diluted by up to 13% with olivine and only small amounts of clay in several of the tuffs. The effect of subtraction of olivine with ~45 wt% MgO and no Zr is given by the arrow in Figure F30C. Addition or subtraction of olivine cannot direct residual liquid compositions from Koolau into the field of Kilauea tholeiites. Dilution by clays in samples of Hole 1223A will draw compositions nearly toward the origin (no Zr and <1 wt% MgO), but in several samples, this effect should not be too important. The diagram suggests, then, a Koolau provenance for the vitric tuffs. The higher SiO2, lower Ba, and lower iron as Fe2O3 (Fig. F28C) of the same samples support this contention.

A great number of glass shards in the vitric tuffs have olivine phenocrysts that enclose small Cr spinel crystals. The bulk samples are fairly olivine rich, approaching the bulk compositions of Hawaiian tholeiitic picrites in many respects. The high Ni, Cr, and MgO contents of all the samples indicate the importance of olivine tholeiite, and perhaps even picrite, in their provenance. Olivine tholeiite is not atypical of Hawaiian volcanoes at the shield-building stage, but many Hawaiian tholeiites are far more differentiated than this and either have no olivine on the liquidus or very little olivine in their general petrography. Average Hawaiian tholeiite has ~7.5-8 wt% MgO (Engel and Engel, 1970).

Picritic Hawaiian tholeiite is thought to reside at deep levels in the conduit-reservoir systems of Kilauea and Mauna Loa Volcanoes, where it crystallized to produce abundant dunite, a common type of xenolith in late-stage Hawaiian basalts (Jackson, 1968). More differentiated lavas develop in the shallow reaches of rift zones, where they mix with more primitive lavas during eruptive cycles (e.g., Wright and Fiske, 1971). If the magmas that supplied most of the glass in the two vitric tuffs of Hole 1223A are truly primitive with, for example, high MgO in the glasses; if, in addition, the associated olivines are also forsteritic; and if, in particular, the tuffs each have a restricted range of primitive glass compositions; this evidence together would indicate that the sources were very large eruptions of tholeiitic magma, dwarfing the volume of any eruptions known from the islands themselves. Such a quantity of magma could only be derived from the deep, high-temperature reaches of Hawaiian magma reservoirs during the main shield-building stage. To reach into the main reservoir, below the shallow active rift system, would require one or more massive failures of the flank of the volcano of the type inferred for the Nuuanu Landslide.

Summary

One of the objectives of coring at this site was to try to determine if the Nuuanu Landslide occurred as a single or as a multistage event by inference from the number of turbidites recovered. Several unconsolidated volcaniclastic turbidites of varying thickness were recovered. Paleomagnetic data indicate that the uppermost turbidite has an estimated age between 0.99 and 1.07 Ma, that the other turbidites have an age between 1.77 and 1.95 Ma, and that other underlying units are older than 1.95 Ma. A surprising discovery was the recovery of the two crystal vitric tuff layers. Preliminary geochemical analyses indicate these tuffs are MgO-rich tholeiitic basalts and have geochemical similarities to Hawaiian tholeiites. However, there remain some other possibilities for the sources of crystal vitric tuffs, such as a part of the Hawaiian Arch or a nearby seamount. The genesis of crystal vitric tuffs can be complicated. Questions arise about why they are indurated so close to the seafloor, why they are so glassy, why they are so rich in fresh olivine, why they include kink banding and fibrous structures in the olivine crystals, and why they were warm or even hot when emplaced.

Preliminary results are summarized below:

  1. We recovered several lithologic units that were transported to the site most likely by a number of distinct landslide events. The origin of the deposits, as indicated by petrographic inspection and geochemistry, is the Hawaiian Islands. Furthermore, the age of the transported units is coeval with the age estimate for the Nuuanu Landslides. The turbidites associated with the landslides were also identified by physical properties changes.
  2. Two pyroclastic events similar to the 1980 Mount Saint Helens' eruption but an order of magnitude larger occurred on Koolau Volcano at ~2 Ma. These events may correlate with the collapse of the flank of the volcano and the formation of the Nuuanu debris field.
  3. The thickness of turbidites and pyroclastic material corresponding in age to the Nuuanu Landslide (1.8-2.4 Ma) is >38 m at Site 1223, over 300 km from Oahu. We did not core to the bottom of this sequence; thus, the related deposits may be thicker, and additional landslide events may have occurred.

NEXT