Catchment - The Fly River and its tributaries head in the fold and thrust belt of New Guinea where nearby peaks reach up to 4000m. The river drains southward, descending through a narrow canyon thinly mantled with gravel, and emerges as a sand bedded river that crosses a low-relief but highly dissected plain (the Fly Platform) of uplifted Quaternary sediments to the Gulf of Papua. Only about 30% of the 75, 000 km2 basin is in the rapidly eroding uplands. The basin receives up to 10m/year rainfall in the uplands and still exceeds 2 m/year in the lowlands. The river has three primary branches, the Ok Tedi, the Fly and the Strickland. The Ok Tedi joins the Fly at about 18 m above mean sea level where the river is 846 km from the delta mouth. The Strickland (36,740 km2 ) joins the Fly (18,400 km2 ) at about 6 m above sea level, and below this junction the Fly gains an additional 19,860 km2 as it travels another 432 km to the delta mouth. Pickup (1984) and Dietrich et al . (1999) describe the overall geomorphology and sediment routing of the Ok Tedi- Fly System.
Before disturbance by mining, erosion rates were estimated to be about 2 to 4 mm/yr in the steepest part of the basin (Pickup et al., 1981; Pickup, 1984) and the combined sediment discharge of the Ok Tedi and the Fly where they join was about 10 million tonnes (840 t/km2-yr or about 0.4 mm/yr equivalent). The Strickland, which penetrates farther into the fold and thrust belt and drains twice the drainage area, discharges about 70 to 80 million tonnes/year (about 2000 tonnes/km2-yr or the equivalent of about 0.8 mm/yr) , hence the discharge to the delta was on average about 80 to 90 milllion tonnes/year, with perhaps only 30% of that load coarser than silt .
Natural sediment loading in the uplands area is dominated by landslides (Pickup et al., 1981; Pickup, 1984). Small landslides are common, and large ones can be spectacular. Blong (1991) reported a 7 km3 landslide that originated on a local limestone escarpment (up to 750 m high) called the Hindenburg wall about 8800 BP. Reworking and dispersal of landslide introduced sediment downstream is rapid. For example, in about a 6 month period, nearly 60% of a 70 million m3 landslide deposited in 1989 into a headwaters tributary of the Ok Tedi was flushed downstream (Parker and Dietrich,unpublished report).
In the Strickland tributary about 10 km2 has been glaciated (Loffler, 1972) and along its eastern border there is a volcano, Mt. Bosavi, that appears to have last erupted 30,000 to 50,000 years ago (reviewed in Dietrich, et al., 1999). Over the time period of the MARGINS study, then, landsliding will be the dominant sediment source in the mountainous headwaters.
The majority of the gravel entering the headwater streams breaks down (Parker, 1991; unpublished reports) to fine sediment which is then carried away in suspension. There is limited gravel floodplain areas once the rivers leave the headwater gorge (at about an elevation of 500 m). Net deposition of the remaining resistant gravel must occur, as none can cross the abrupt gravel- sand transition that happens at large breaks in slopes along Ok Tedi- Fly and Strickland systems.
Below the gravel-sand transition the floodplain (where active sedimentation is occuring) progressively widens downstream from about 4 to 18 km where the Fly and Strickland join. The floodplain system is composed of an inner scoll bar complex bordered by backswamp and blocked valley swamps. Meander loop cutoffs are common as are large blocked valley lakes (the biggest being Lake Murray). Blake and Ollier (1971) proposed that the sedimentation along the Fly River (Fly and Strickland) kept pace with Holocene sea level rise and that this sedimentation led to blocking of lowland tributaries and lake formation. Pickup (1984) estimated Holocene floodplain sedimentation rate along the Middle Fly to be about 1 mm/yr. Based on recent floodplain drilling (up to 15 m deep) and radiocarbon dating, J. Chappell, W. Dietrich and G. Day (unpublished data) agree with the Blake and Ollier inference and with the Pickup value as a rough average. Dietrich et al. (1999) conclude, however, that the proposed missing mass in the Harris et al. (1996) sediment budget of offshore sedimentation can not be found in the Fly floodplain deposits.
The Middle Fly floodplain shifts progressively from rainforest dominated to swamp grass (see also) covereged and correspondingly there is significant downstream in bed slope (from 6.6 to about 1 times 10-5) and downstream fining of the bed from about 0.3 to about 0.1 mm (Dietrich et al., 1999). Channel lateral migration rates are low throughout the system, averaging only 1 to 2 m/yr for the 250 m to 350 m wide channel and in the past 50 years it has been essentially zero in the swamp grass reach. The Strickland creates a backwater on the Fly and this may have contributed to swamp formation and reduction of channel migration rates (Dietrich et al., 1999). An important geomorphic feature of the Fly floodplain system are "tie channels" which connect the mainstem Fly and Strickland to off river water bodies and serve as significant avenues of water, sediment and biotic flux (Dietrich et al., 1997).
Beginning in 1985, rock waste and tailings were discharged to the Ok Tedi from a gold and copper mine (Ok Tedi Mining Limited). By 1992 the total input of sediment associated with the mine activity was 501 million tonnes. If the mine continues on current mine plan, the total sediment introduced during the life of the mine (scheduled to close in 2010) will be 1720 million tonnes. This is will be nearly 14 times the natural load on the Ok Tedi. In response to concerns about the environmental consequences of mine activity, a monitoring program was initiated and progressively expanded. Ok Tedi maintains precipitation and stream gauging sites as well as some of the sediment sampling sites. There are currently 9 flow and suspended sediment gauging stations along the Ok Tedi Fly system. At some of these stations, data go back to the early 1980s. Numerous studies have been conducted to document the change in the river system. Detailed topographic maps using aerial photograph have been generated in the Ok Tedi. High resolution aerial photographs have been flown in 1982 and 1992 of the Ok Tedi and Fly and repeat infrared surveys have been made in more recent years to document changes in vegetation. Considerable effort has been expanded using differential GPS to establish a high resolution network of benchmarks up the Fly and Ok Tedi. This surveying has provided a high quality river profile in very flat terrain. Numerous cross-sections have been established for repeat surveys. The Ok Tedi and Fly floodplain has been topographically mapped through a combination of aerial photography, laser profiling and ground surveying.
Geoff Day (Day and Dietrich, 1997) monitored over 70 water level recorders across the Middle Fly floodplain to document the controls on inundation frequency and resulting rates of floodplain sedimentation. For several years in the early 1990s annual sampling surveys were conducted to collect shallow core samples to document the spread of copper-rich, mine related sediment across the floodplain. These data indicate that floodplain deposition rates were about 1 to 2 mm/year in the forested reach but much less in the swamp grass reach. Local rates were locally much higher (Dietrich and Parker, 1994; Heffler et al., 1997). The low average rate in the swamp grass reach was apparently due to chronic local rain-induced flooding that reduces the incidence of sediment rich overbank floods (Dietrich and Parker, 1994; Dietrich et al., 1999; Day, in prep). These tracer studies indicate that roughly 3% of the sediment load entering the upper Middle Fly is deposited in the floodplain.
A sediment budget in 1992 of the Ok Tedi-Fly system indicated that about 66% of the gravel introduced from the mine broke down to finer sediment, 50% of the total mine-related sediment remained in the Ok Tedi, and another 27% was deposited in the channel and floodplain of the Fly. Hence only about 30% of the total mine-related load reached the junction with the Strickland. This load was still roughly 4 times the natural level on the Fly. At model calculation by Cui and Parker (unpublished report, 1999) suggests that by the time of mine closure about two-thirds of the mine-related sediment will be deposited along the Ok Tedi and upper Middle Fly.
Parker and colleagues (Parker, 1991, Parker et al. 1996, Cui and Parker, unpublished report) have developed a process-based model, based on their work on the Fly and Ok Tedi, for routing of sediment downstream through the gravel-bedded uplands and across the lowland plain. Separate models exist for 1) gravel routing, 2) sand and finer sediment routing along the channel and 3) fine sediment dispersal onto the adjacent floodplain. The models generally accurately predict the channel and floodplain sedimentation due to mining load upstream
A mine has also opened in the headwaters of the Strickland and is estimated to add about ½ to 1/3 the mine load discharged into the Ok Tedi. Given that the natural load on the Strickland is about 8 times greater than the Fly natural rate, this increase load is a much smaller proportion of the total load.
Geomorphic response to the increased load includes aggradation and accelerated bank erosion, increased incidence of overbank flows and deposition, and probably fining of the river bed. From the perspective of the MARGINS program, the large influx of sediment in the Fly offers a unique opportunity to examine a large sediment signal, with a distinctive chemical composition, that can be exploited to document and model sediment transfer processes through the entire system. The influx of sediment has not fundamentally changed the transport processes. It is, however, a strong, long-duration signal that differs from short term stochastic introduction associated with natural loading events. The extensive data already collected and the established precipitation and hydrologic program in the Fly greatly increases the opportunity and possibility of obtaining accurate sediment budgets through the system.
The Modern Fly River Marine Dispersal System - The "natural" estimated load entering the Fly delta is estimated to be 85 million metric tons per year; 90 % of this material is fine-grained (Ok Tedi Mining Ltd, 1988). Mass-wasting processes may cause significant fluctuations in this load. Harris et al. (1993) created a preliminary sediment budget over a 210Pb timescale (100-yr). Their results indicate that modern sediments are partitioned in the following fashion (assuming 1 m3 = 1 metric ton):
Delta Front: 24 ± 12 ´ 106 metric tons/yr
Prodelta: 22 ± 6 ´ 106 metric tons/yr
Distal Delta: 0.9 ± .2 ´ 106 metric tons/yr
This budget accounts for 47 ´ 106 metric tons/yr or 55% of the load. An additional 2% is estimated to be transported southward into the Torres Strait. The remainder is hypothesized to be advected northeastward along the Gulf of Papua (GOP) shelf.
The Fly River delta has the classic funnel-shaped geometry of a tide-dominated system, and has been used as the end-member example in a delta classification scheme created by Galloway (1975). Local tides range from about 3.5 m at the mouth to 5 m at the apex (Wolanski and Eagle, 1991). Strong tidal currents preclude fine-sediment deposition within the distributary channels. The sandy sediments observed are comprised of 90% quartz, 20% chlorite and some altered feldspar (Taylor, 1977). Fluid muds have been observed within the delta, but their significance in redistributing sediments is unknown (Wolanski and Eagle, 1991).
An idealized cross-section of the Fly delta created by Spencer (1978) is presented in Figure 12. Sub-tidal sandbanks and pre-Holocene clay deposits serve as "anchor points" for the many mangrove-stabilized islands within the delta. A small fraction of the fine-sediment entering the delta is thought to be trapped in the extensive mangrove forests (Harris et al., 1996; Wolanski et al, 1998). However, quantification of sediment accumulation within the delta is difficult due to ephemeral nature of the majority of the sediment deposited. Fly delta mangrove deposits are comprised of three facies: 1) a massive, bioturbated, clay-rich mud, 2) a thinly laminated sandy mud, and 3) a coarsely bedded sand (Walsh and Nittrouer, 1998b). Studies of nutrient and metal cycling within the delta indicate that the intertidal mudbanks act as a sink for these materials (Alongi, 1991; Alongi et al., 1991; Alongi et al., 1992).
The majority of fine-sediment is advected beyond the Fly mouth and accumulates offshore on top of Pleistocene floodplain deposits (Harris, 1994; Harris et al., 1996). The modern offshore facies transitions are similar to those observed by Kuehl et al (1986) in the Amazon system (Alongi and Robertson, 1995). Northward advection of Fly-derived sediment occurs due to the predominant southeasterly winds (the "trade winds") as well as the clockwise geostrophic gyre circulation present in the GOP (Wolanski et al., 1995). A detailed discussion of the offshore facies can be found in Alongi et al. (1992) and Harris et al. (1993). In brief, sediments become progressively finer in the offshore direction as wave and current velocities diminish. Simultaneously, benthic organism abundance increases, and the interbedded physical structure is replaced by homogenous, bioturbated muds. Nutrient and metal cycling offshore are described in Baker and Harris (1991), Alongi et al. (1993), Alongi and Robertson (1995), and Apte and Day (1998).
The Gulf of Papua - The "missing" 45% of Fly sediment is hypothesized to transport along the GOP shelf where it mixes with sediments supplied by the Kikori, Purari and other rivers. A fraction of the fugitive sediment is likely sequestered in the extensive mangrove forests that line the channelized GOP coastline; however, the majority is expected to accumulate on the shelf (Harris et al., 1993), possibly on clinoform deposits that extend along the 50-m isobath (Milliman et al., 1999). High sediment accumulation rates (> 1 cm/yr) are known to occur in this region (Walsh and Nittrouer, 1998a), but the source of these sediments is undetermined. Geochemical evidence also suggests that terrestrial sediment is escaping the shelf into Pandora Trough, but the magnitude of this loss is unknown (Bird et al.,1995; Brunskill et al., 1995). Several studies have investigated nutrient and metal cycling in the greater GOP including Alongi et al. (1992), Alongi (1994), Alongi et al. (1996) and Robertson et al. (1998). An interesting observation is that geochemical conditions suggest that deep, physical mixing occurs in the GOP, similar to that observed in the Amazon (Alongi et al., 1996). Further research in the GOP is needed to help constrain the Fly sediment budget and to better understand sediment and solute cycling throughout this region. Other attributes of the GOP that make it an excellent locality for MARGINS research include its long marine sedimentary of foreland-basin evolution (3000 m since the Jurassic, see Thom and Wright, 1983), as well as extensive coral reefs and Halimeda banks. To learn more about the Gulf of Papua, CLICK HERE.
Fly Selection criteria:
A. Natural Factors
1. Has strong well documented forcing
2. Has active sedimentation with opportunities to study interaction with many kinds of sediment storage elements.
3. Sediment is actively transferred among environments
4. System is closed to the sea. There is a question of whether in order to study off shore sedimentation it is necessary to monitor the other smaller basins entering the Gulf of Papua.
5.There is a high resolution stratigraphic record
6. Carbonate environments are well established offshore
7. If the alternate site is the Sepik, the Fly differs by having contributions from a mine sediment source, a shallow platform across which the floodplain extends and very different offshore environments. The Markham has less extensive floodplain system and different offshore environment.
B. Human considerations
1. There has been much published work on the tectonics of New Guinea. The many studies associated with the mine on the Ok Tedi are not yet published, but the data and the gauging station data are available. Recent research on floodplain sedimentation, channel and floodplain modeling and offshore sedimentation provide some framework to begin a project.
2. Logistics are mangeable on the Fly because of easy access via ship (up river), via plane and helicopter as well as small boat. Highlands are less accessible on foot, but remote sensing data are available. It is likely that the local mines will be supportive of this work.
3. The anthropogenic influence is well-defined and is locally not small. Significant sedimentation on the lower Strickland and lower Fly have not been observed. The large mine-related sediment loading offers a unique sediment signal to document sediment transfer processes.
4. Work in the Fly catchment will be relevant to society in at least two very different ways. One would be an enhanced ability to explain and predict the fate of mine related sediment through the environment. This is an issue found on many river systems. The second is an understanding of how a low gradient river responds to sea level rise.
5. Leverage opportunities exist with the TROPICS
program, with the PNG government and with the local mining companies.