Located adjacent to the NE Pacific convergent boundary, Cascadia Basin has a global impact well beyond its small geographic size. Composed of young oceanic crust formed at the Juan de Fuca Ridge, igneous rocks underlying the basin are partially insulated from cooling of their initial heat of formation by a thick layer of pelagic and turbidite sediments derived from the adjacent North American margin. The igneous seafloor is eventually consumed at the Cascadia subduction zone, where interactions between the approaching oceanic crust and the North American continental margin are partially controlled by the thermal environment. Within Cascadia Basin, basement topographic relief varies dramatically, and sediments have a wide range of thickness and physical properties. This variation produces regional differences in heat flow and basement temperatures for seafloor even of similar age. Previous studies proposed a north-south thermal gradient within Cascadia Basin, with high geothermal flux and crustal temperatures measured in the heavily sedimented northern portion near Vancouver Island and lower than average heat flux and basement temperatures predicted for the central and southern portions of the basin. If confirmed, this prediction has implications for processes associated with the Cascadia subduction zone, including the location of the “locked zone” of the megathrust fault. Although existing archival geophysical data in the central and southern basin are sparse, nonuniformly distributed, and derived from a wide range of historical sources, a substantial N-S geothermal gradient appears to be confirmed by our present compilation of combined water column and heat flow measurements.
 Cascadia Basin has been recognized as an area of unusually high geothermal heat flow for over 40 years [Korgen et al., 1971; Davis and Lister, 1977]. Located adjacent to the North American margin, rapid sedimentation resulting from the combination of high biological productivity induced by coastal upwelling and turbidites from multiple sea level excursions covered and insulated the young oceanic crust formed at the nearby Juan de Fuca plate [Davis et al., 1992, 1999]. Isolation of crustal hydrothermal circulation cells from the overlying seawater reservoir soon after formation results in higher geothermal heat flow for the northern portion of the basin compared to global averages of the same age. Within the overlying water column, the NE Pacific has also been recognized as the distal end of the global thermohaline circulation of seawater, and the high geothermal heat flow within Cascadia Basin warms and elevates centuries-old Pacific bottom water to midwater depths [Talley and Joyce, 1992; Hautala et al., 2009]. The resulting buoyancy-driven circulation contributes substantially to a midwater plume of dissolved silica that can be traced from the North American continental margin across the entire North Pacific [Talley and Joyce, 1992; Johnson et al., 2006].
 The northern third of Cascadia Basin west of Vancouver Island has been well studied by numerous heat flow surveys and ODP Leg 168 and IODP Leg 301 drill holes [Davis et al., 1997; A. T. Fisher et al., 2005; Hutnak et al., 2006, and references therein]. In contrast, the central and southern portions of the basin located offshore Washington and northern Oregon and north of the Blanco fracture zone (Figure 1) have only had a small number of ad hoc heat flow measurements. This lack of any systematic survey of heat flow in the central and southern sectors severely limits conclusions regarding an N-S gradient in basement temperatures within the basin. However, the importance of the thermal conditions of the incoming oceanic crust to the subduction process in an area of potential megathrust earthquakes with substantial societal impact on the North American margin [Hyndman and Wang, 1993, 1995; Satake et al., 1996; M. A. Fisher et al., 2005; Wells et al., 1998; Goldfinger et al., 2003, 2012] justifies compilation of all available data from the full range of both conventional and unconventional sources. For this study, we compiled geophysical and physical oceanographic data from a wide variety of archival sources. We combined these data with a small number of new heat flow measurements using a modified multicorer, in order to make an estimate of the first-order spatial variation in geothermal flux within the basin.
2. Local Geology
 Cascadia Basin is a small and relatively well-studied geological feature in the NE Pacific bounded by the Juan de Fuca Ridge, the Blanco fracture zone and the North American margin. Located off the coasts of Washington, northern Oregon and British Columbia, the young igneous basement of the Juan de Fuca plate is primarily covered with interlayered pelagic and turbidite sediments from the adjacent North American continent [Goldfinger et al., 2012]. These turbidites form large, thick fans of terrestrial sediments within the basin, where only the tallest basement outcrops rise above the seafloor and remain exposed (Figure 1) [see also Underwood et al., 2005]. In the central and eastern portions of the plate, thick turbidites are overlain by a young layer of pelagic sedimentation produced by the relatively high biological productivity associated with the adjacent North American continental margin [Hedges et al., 1999]. The Juan de Fuca plate is a true enclosed basin, bounded by the Juan de Fuca Ridge on the west, the high-standing north wall of the Blanco fracture zone in the south and the North American margin and Cascadia subduction zone (CSZ) on the east. Limited gateways for the exchange of bottom water with the open NE Pacific occur only through narrow topographic gaps in the western and southern barriers [Talley and Joyce, 1992; Hautala et al., 2009]. Although Cascadia Basin consists of only 0.05% of the area of the global seafloor, the seafloor heating from the Juan de Fuca plate appears to provide 2% to 3% of the total energy required to close global thermohaline circulation [Hautala et al., 2005, 2009]. Finally, the eastern edge of the Juan de Fuca plate, along with the smaller Gorda plate in the south, provides the oceanic crust that is being subducted beneath the adjacent North American plate, producing the CSZ. This active subduction zone has been a major tectonic hazard in the recent geological past, with Mw9+ earthquakes that produce tsunamis on a scale that impacts the entire Pacific [Nelson et al., 2006; Goldfinger et al., 2012; Atwater and Hemphill-Haley, 1997; Satake et al., 1996].
 Basement underlying Cascadia Basin sediments ranges in age from zero at the Juan de Fuca spreading center to 8.5 Ma at the western edge of the deformation front of the subduction zone, adjacent to the North American margin (Figure 1). Passage of the Juan de Fuca spreading center over a region of mantle with anomalous chemical composition has produced abundant seamounts, which are clearly visible even in low-resolution bathymetry data and are particularly dense on the western flank of the Juan de Fuca Ridge [Davis and Karsten, 1986]. Within the basin, the seafloor slopes southward with a gradient of 1:1000, producing a maximum water depth of approximately 2900 meters in the SE corner [Underwood et al., 2005]. The mix of turbidites and pelagic sediments fill the basin to a 2 km thickness near the edge of the accreted margin in the northeast [Davis and Hyndman, 1989; Westbrook et al., 1994]. Sediment thickness in the center of the basin ranges from 200 meters to 600 meters [Moran and Lister, 1987; Flueh et al., 1998; Underwood et al., 2005; Nedimović et al., 2009], and the pelagic sedimentation rate in midbasin at the latitude of 47°N is approximately 3 cm/yr, increasing to 15 cm/yr near the continental margin at the eastern edge [Hedges et al., 1999].
 On the western side of the basin, the sediments thin and on-lap exposed igneous basement near the active Juan de Fuca spreading center at crustal ages of less than 1 Myr. In the SW corner of the basin near the Blanco fracture zone, the eastern Juan de Fuca Ridge flank is distant from turbidite sources from the continent, and has thinner sediment cover with more exposed basement relief derived from a complex tectonic history [Embley and Wilson, 1992]. When considered at a nominal distance of 50 km west of the CSZ deformation front, crustal age increases from 6 Ma in the north and 8 Ma in the south of the basin, but there is a systematic decrease in sediment thickness with decreasing latitude. Specifically, on this easternmost edge of the basin, sediment cover thickness is 2500 meters at ODP Site 888 near 48°N in the north [Westbrook et al., 1994], 1864 meters thick at 46°N latitude [Moran and Lister, 1987], and 900 meters thick at 44° 45′N (DSDP Site 174) [Kulm et al., 1973].
3. Water Column Results
 A compilation of CTD (conductivity-temperature-depth) data, subject to stringent quality control, was used to quantitatively estimate bottom water circulation within the basin, and identify the large-scale pattern of geothermal input into the water column [Hautala et al., 2005, 2009]. CTD casts in the water column overlying the basin are both more numerous and more uniformly distributed than traditional heat flow stations, allowing an estimate of the geographical distribution of geothermal heat content and bottom heat flow over the entire Juan de Fuca plate (Figure 1).
 Heating near the seafloor produces a characteristic near-bottom deviation toward higher temperature from the linear background potential temperature-salinity relationship. At any location, the difference between the observed potential temperature and a background profile (from Blanco Saddle, a key passage hosting bottom water inflow to the basin) gives a measure of the integrated effects of bottom heating along a pathway defined by the bottom water currents. After vertically integrating to obtain the total geothermal heat content, the net rate of heating may be determined from a heat budget, if advection and diffusion terms in the water column are known. In practice, water column velocity, diffusivity terms and bottom heating are simultaneously estimated on a 0.5° × 0.5° grid over the entire basin by least squares solution to a large set of linear equations that conserve geothermal heat content, potential vorticity and volume flux [Hautala et al., 2009]. Figure 2 shows relevant spatial transects through the bottom geothermal heating field determined by this “inverse model.” Note that prior to analysis, water column fields are statistically smoothed over a ∼200 km lateral scale in order to eliminate spurious effects of small-scale time-dependent variability on the estimation of the steady state large-scale circulation. Thus the spatial pattern of the bottom heat flux arising from the circulation inverse model also represents an average over similar spatial scales.
 While the bottom water circulation, and thus the advective pathway, is strongly constrained by potential vorticity conservation, the strength of the bottom heat flux relative to diffusive terms is influenced by a scaling factor between water column heat content and source flux that is related to background vertical temperature and salinity gradients [see, e.g., Lavelle et al., 1998]. Hautala et al.  focuses on the physical oceanography of the bottom water circulation and assumed a value of 1.0 for this factor. Regardless of the value of this scaling factor, the average geothermal heat flux is strongly constrained by the magnitude of the vertical diffusivity, and a low value was selected in the 2009 study to agree with general background values from direct measurements. Since the focus here is on the bottom heat flux itself, Figure 2 incudes the sensitivity experiments conducted in Hautala et al.  as well as results from a range of scaling factors from 1.0 to 2.0, thought to span a reasonable range of background temperature and salinity gradients.
 As seen in Figure 2, northern areas (latitude >47°N) are predicted by the inverse model to have a significantly higher seafloor heat flux than the southern portion (<46°N). Negative values in the south are likely related to an aliasing of cold water renewal across the Blanco fracture zone through unresolved small passage flows, particularly via the Cascadia Seachannel, and may bias the average slightly low. However, the increase in seafloor heat input with latitude that is demanded by the lower water column energy budget is readily apparent in the inverse model results, whether or not the estimated fields are sampled near the base of the continental slope (Figure 2a) or along a midbasin line parallel to the Juan de Fuca Ridge (Figure 2b).
4. Sediment Geothermal Heat Flow
 In order to further test the hypothesis that northern Cascadia Basin has higher geothermal heat flux than the southern portion using direct sediment heat flow measurements, we compiled archive heat flow data distributed over the middle and southern portions of the Juan de Fuca plate. Geothermal heat flow in much of the northern third of Cascadia Basin has been previously well studied and extensively reported [Davis et al., 1990, 1992, 1999; Hyndman and Wang, 1993, 1995; Thomson et al., 1995]. In the northern area, traditional thermal probe measurements are relatively densely spaced, although there is a data gap between crustal ages of 4 Myr and 6 Myr old crust, just west of the deformation zone at 6 Ma [Davis et al., 1990, 1999]. Comparison of Cascadia Basin data with global compilations indicates that the heavily sedimented area on the eastern flank of the northern portion of the Juan de Fuca Ridge younger than 2.5 Myr has a higher average geothermal flux than for global, non–Cascadia Basin crust of the same age [Johnson and Pruis, 2003; Hasterok et al., 2011]. For seafloor older than 2.5 Myr in the heavily sedimented northern corridor, where data are available, the heat flux also appears significantly elevated, i.e., 200 mW/m2 at 3 Ma and 120 mW/m2 at 6 Ma [Davis et al., 1990, 1999] above the equivalent global averages of 128 to 105 mW/m2 for world average crust that is 3 to 6 Ma in age, respectively [Stein, 2003].
5. Heat Flow Data Sources
 In contrast to the northern area, measurements of heat flow in the central and southern portions of Cascadia Basin are sparse, widely spaced, have been obtained with a variety of instruments and field programs, were analyzed by different methods, and in some cases, are incompletely described in the archives. In addition to compiling archival data for the middle and southern sections of the basin, we made additional heat flow measurements over the central and southern portions of the basin using thermistors mounted on a sediment multicorer during a 2006 cruise as part of a program to determine the source of the mid-Pacific silica plume [Talley and Joyce, 1992; Johnson et al., 2006; Esther et al., 2010]. These new stations are distributed over the basin, but also include a detailed heat flow and water column survey of an exposed basement outcrop, informally named “Nubbin Seamount.” This outcrop is located on 7 Ma crust in the SE corner of Cascadia Basin, and the heat flow values are included here to demonstrate clearly that crustal hydrothermal systems are still exchanging heat with seawater even in older Cascadia crust.
 In spite of the proximity of Cascadia Basin to the west coast of North America, the entire Juan de Fuca plate has been incompletely surveyed by modern swath bathymetry, and the best plate-wide map remains the hand-contoured integrated profiles of the NOS Map Series [National Ocean Survey, 1974]. The limited modern swath bathymetry surveys in the area have been largely focused on specific geological targets. The axis of the Juan de Fuca Ridge that forms the western basin boundary has been surveyed by NOAA with swath bathymetry, along with the Blanco fracture zone at the southern boundary [Embley and Wilson, 1992]. A compilation of magnetic anomaly data of the Juan de Fuca and Gorda Plates has been interpreted in Wilson , and we used those anomaly identifications for our crustal age assignments. More recently, 2-D multichannel seismic profiles over the Juan de Fuca Ridge extended out onto the westernmost flank of the basin, although this coverage does not extend to the central and eastern portions of the basin [Carbotte et al., 2008; Nedimović et al., 2009].
 Archival heat flow data for this compilation are generally from the Abbott compilation (available from GeoMapApp, http://www.geomapapp.org/), specifically those measurements obtained during cruises TT31, TT40, AII112, and W836 from sites on the Juan de Fuca plate south of latitude 47° 30′N, plus published heat flow data from Moran and Lister  and Shi et al. . Our geographic boundaries excluded heat flow data from sites located (a) landward of the deformation front on the eastern margin of the basin, (b) west of the Juan de Fuca Ridge, (c) south of the Blanco fracture zone, and (d) from the well-studied and previously published sites [e.g., Davis et al., 1992, 1997, 1999] in the northern third of the basin (Figure 1). Uncorrected heat flow measurements, without further modification for recent sedimentation, sediment compaction, or upward migration of pore fluids, represent the actual upward heat flux transferred from the seafloor to the bottom layers of the water column, and these uncorrected values are appropriate for comparing to those inferred from the bottom water heat budget [Hautala et al., 2009]. The only additional filter applied to the compiled heat flow data was to eliminate those sites located within 10 km of an exposed basement outcrop [Thomson et al., 1995], where these circumstances could be identified from existing bathymetry maps. The single exception to this restriction is the new heat flow data taken in proximity to Nubbin Seamount, which were included to demonstrate that crustal fluid emissions are occurring from even 8 Myr old exposed basement outcrops in the SE portion of the basin.
6. Central Basin Heat Flow Corridor
 Compiled heat flow data within Cascadia Basin were grouped into the Central and Southern corridors arbitrarily divided by the latitude 45.5°N, based on a combination of the geographic distribution of heat flow stations, on differences in sediment thickness adjacent to the margin and on the abundance of basement outcrop exposures between the middle and southern sections of the plate (Figure 1). Figure 3 shows station locations for the area of Cascadia Basin between latitudes 45.5°N and 47.5°N that includes archive data, stations from Moran and Lister  and Shi et al. , and two of our new multicore measurements from 2006. Figure 4 shows heat flow as a function of crustal age for this latitude corridor, where age is determined from the location within the magnetic anomaly map of Wilson . Stations connected with straight lines are from Moran and Lister , where the authors calculated site averages using multiple pogo reentries, with between 4 and 8 penetrations at each site, and acquired high-quality seismic reflection data along the same profile. Our 2006 multicore deployments and data from the GeoMapApp archive are included. The Shi et al.  stations are the cluster of data located in the upper right hand corner of the station map (Figure 3), and are also shown at 8 Ma in the corresponding age plot (Figure 4).
 The smooth solid curve in Figure 4 is the theoretical value for a crustal plate cooling by conduction only; correction for sedimentation to this no-advection trend would reduce the predicted values by less than 15% for sediment thickness in mid–Cascadia Basin [Davis et al., 1999], and this correction is shown as a thin dashed curve in the plot. However, the data set shown in Figure 4 suggests that the heat flow values versus age do not rise to the conduction-only value by the crustal age of 6 Ma even when reduced 15% by sedimentation correction (170 mW/m2 at 6 Ma). The measured heat flow values decrease again with age in crust older than 6 Myr, and depart further from the conduction-only curve as the deformation front of the margin is approached. This is consistent with the original conclusion of Moran and Lister  that their heat flow profile did not reach the predicted “conductive-only” values at any crustal age along their profile. In crust younger than 6 Myr, heat flow values are also much lower than conductive-only predictions, with the exception of what appears to be—interpreted from the high, scattered values—a zone of active hydrothermal flow in approximately 2 Myr old crust.
 Although the bathymetry data along most of the middle corridor of the basin are only of low resolution, it appears that there are very few exposed basement outcrops penetrating the seafloor on crust older than 2 Myr. This lack of basement exposures at the present time is similar to the relatively complete sediment cover of basement in the northern section of the basin [Davis et al., 1992, 1999]. The apparent absence of large exposed seamounts or basement outcrops in the middle corridor of Cascadia Basin is supported by the similar lack of associated gravity anomalies in the GeoMapApp compilation that would indicate the presence of large undetected basement exposures and by systematic single channel seismic data [National Ocean Survey, 1974] in the same area, although the low resolution of these older methods does not preclude the possibility of undetected small exposed basement outcrops in this area. We believe that the lower-than-conductive heat flow values observed in the central corridor represent an environment where basement is initially cooled for millions of years by the vigorous exchange of crustal hydrothermal fluid and seawater through exposed outcrops, which is followed by reheating and increased heat flow values after subsequent coverage of these outcrops by sediments in the Late Pleistocene. This thermal rebound model was originally described by Hutnak and Fisher  and provides a plausible explanation for heat flow values that are lower than conductive in seafloor that is presently completely sediment covered, but where the sedimentation is only very recent.
7. Southern Basin Corridor
 The diagram in Figure 5 shows the location of the heat flow stations on GeoMapApp multiresolution bathymetry for the southern corridor of Cascadia Basin (latitudes 43.5°N to 45.5°N). Because of the 020°N orientation of the Juan de Fuca Ridge axis and the roughly N-S orientation of the boundary of North America, the southwestern section of Cascadia Basin is more distant from the continental sources of turbidites that cover the northern and middle portions of the basin [Underwood et al., 2005]. The additional distance of the SW portion of the basin from terrestrial sediment sources results in seafloor with reduced sediment cover and more abundant basement outcrops than the northern section of the basin (Figure 5). Further, many of the basement outcrops in the southern corridor are not simple volcanic cones, but have a complex morphology and may consist of “high inside corners” associated with rift propagation [Severinghaus and Macdonald, 1988]. The southern latitudinal corridor of Cascadia Basin has been an area of considerable tectonic complexity, including the presence of three wide crustal shear zones associated with southward propagating rift traces [Wilson, 1988], episodic extension of the Blanco fracture zone [Embley and Wilson, 1992], and the presence of recent growth faults on the ridge flanks [Nedimović et al., 2009].
 Heat flow measurements along the southern corridor consist primarily of the compiled archive data plus 10 new sites from our 2006 cruise. This profile includes 6 sites on a linear profile within 2 kilometers of the Nubbin basement exposures and 4 sites that are >10 km distant from any outcrop. Similar to the central corridor, heat flow data in the southern corridor lie well below both the reference “conductive-only” curve and even below a conservative 15% reduction of that predicted curve due to recent sedimentation (Figure 6). The obvious anomalies in this southern corridor data set are the 6 heat flow measurements from the Nubbin outcrop profile in heavily sedimented 7 Myr old crust (Figure 6). With the large amount of unsedimented exposed basement rock in the southern part of the basin available as an interface for fluid exchange between the porous upper crustal reservoir and seawater (Figure 5), these low heat flow values are not surprising. Large areas of the 24 Ma seafloor off the Costa Rica margin show much lower than predicted heat flow values, and are interpreted to have been cooled by extensive fluid exchange through exposed basement outcrops that penetrate the overlying sediment cover [Fisher et al., 2003; Hutnak et al., 2007; Wheat and Fisher, 2008]. If the reduced heat flow values observed all along the southern third of Cascadia are the result of large-scale fluid exchange between the crustal reservoir and seawater through the overlying insulating sediment cover via permeable igneous basement outcrops, then even the oldest exposed volcanic rock exposures within the corridor should be sites of active hydrothermal fluid venting [Villinger et al., 2002; Harris et al., 2004; Wheat and Fisher, 2008]. The high heat flow values observed near 7 Myr old Nubbin Seamount supports this interpretation (Figure 6).
8. Results and Discussion
 If assumed to be an accurate representation of thermal flux, the new data compilation reported here agrees with the conclusion drawn from the bottom water heat budget derived from CTD casts. The southern corridor of the basin has average geothermal heat flow values less than the seafloor of similar crustal age in the northern section of the basin. However, it suggests that within Cascadia Basin, heat flow is not strongly dependent solely on crustal age and may not depend heavily on the completeness of sediment cover—at the present time. This nonintuitive result is particularly evident for the extensive areas of the central portion of Cascadia Basin. Here, the eastern Juan de Fuca Ridge flank between 45.5°N and 47.5°N latitudes appear to be as sparsely populated with exposed basement outcrops as the northern section of the basin, but average geothermal heat flow values for this central portion still remain substantially lower than the conduction-only model. Normally, this heat flow deficit would imply continuing fluid exchange between the crustal reservoir and seawater. The middle Cascadia Basin section in the 2 to 5 Myr age range has sediment cover that is hundreds of meter thick and has heat flow values that are as low as those for the southern portion of the plate of the same age interval, where basement exposures are abundant. The history of massive and repeated turbidite fill within Cascadia Basin in the late Pleistocene [Underwood et al., 2005] suggests that the process of thermal rebound previously described by Hutnak and Fisher  may be responsible for the lower-than-conductive values observed in the central corridor. In the Hutnak and Fisher  model, the central corridor crust would have been initially well ventilated and the underlying igneous crust and mantle deeply cooled for millions of years by exposure of basement to seawater prior to the late Pleistocene, similar to the present southern corridor. The igneous outcrop ventilators in the central corridor would then have been quickly buried by the Late Pleistocene sediments and are no longer visible bathymetrically. The central corridor received massive turbidite flows from the adjacent Cordilleran ice sheet at the end of the Pleistocene, and has been exposed to reasonably heavy sedimentation that has continued to the present time [Hedges et al., 1999]. As Hutnak and Fisher  point out, rapid and abrupt sediment cover of previously exposed basement, followed by continued relatively heavy sedimentation are conditions under which the thermal rebound process would have the most impact on reducing heat flow below the conductive-only values. This would reduce underlying plate temperatures by up to 25% below theoretical values, in addition to the 15% reduction in observed heat flow from more recent sedimentation.
 In the southern third of Cascadia Basin, extensive basement exposures still penetrate the sediment cover up to crustal ages of 7 Ma, and the discharge of warm hydrothermal fluid from the crustal reservoir continues to occur from the oldest and easternmost basement exposure located near the subduction zone; an environment that precludes thermal rebound. In the southern composite profile of heat flow measurements (Figure 6), heat flow values stay well below the reference conduction-only curve for crustal age and only rise to the value predicted (reduced by 15% correction for sedimentation) within 30 km of the western edge of the CSZ deformation front, where the sediment thickness exceeds 1000 meters and basement exposures appear to be absent.
 A simple comparison of heat flow versus age profiles of the northern section of the Basin [Davis et al., 1999] with the similar compiled age profile within just the southern section (Figure 6) would be in agreement with the traditional paradigm—heat flow values rise quickly to conduction-only values when sediment cover of basement is complete. This model has been successful in explaining regions of extremely low heat flow in the much older crust with abundant basement outcrops in the TicoFlux region on the Costa Rica margin [Fisher et al., 2003; Hutnak et al., 2008]. However, data from the central corridor of Cascadia Basin between 45.5°N and 47.5°N latitudes represent an anomaly, where the overlying sediment cover of basement appears to be complete and relatively thick at the present time, but heat flow values remain much lower than predicted over the entire crustal age range. This heat flow deficit below the conduction-only curve continues for 8 Ma as the crust ages, until the deformation front of the subduction zone is actually reached. Landward of the deformation front within the CSZ, a different set of subduction-related processes begin, including substantial fluid migration and sediment compaction [Carson et al., 1991, 1994].
 The differences observed between heat flow versus age data and conduction-only predicted values for the central corridor of Cascadia Basin may be due to the presence of exposed basement ventilators during much of the 8 Myr history of this section of the plate. Fluid circulation through these outcrops cooled the igneous basement and upper mantle, similar to the present environment of the southern corridor. Turbidites covered these ventilators in the Late Pleistocene and then basement temperatures in the central corridor began to undergo thermal rebound. However, heat flow values in the older crustal section still lie well below the theoretical conductive-only values.
 With the present data compilation and physical property data from DSDP, ODP, and IODP drill holes in Cascadia Basin, it is possible to make rough back-of-the-envelope estimates of latitudinal variation in basement temperature, as the Juan de Fuca plate enters the CSZ, similar to those previously estimated by Hyndman and Wang . If we assume that the physical properties of the uppermost sediment column along the entire eastern Cascadia Basin are comparable to those determined by Davis et al.  from Leg 168 in the north, we can assign a cumulative thermal resistance of 400 m2 K/W for the top 600 meters of the sediment column. If the thermal conductivity of the sediments below the top 600 meters is approximately 1.5 W/°K m [Hyndman and Wang, 1993, 1995], we can use the measured heat flow values for the northern, central and southern corridors of the basin to estimate basement temperatures at an equivalent distance (30 km west) from the deformation front.
 Most of the sediment fill along the CSZ resulted from large episodic turbidite flows at the end of the Pleistocene [Underwood et al., 2005]. Correcting the heat flow values used for these estimates for sedimentation could also increase these heat flow values by as much as 15% [Davis et al., 1999], but uncertainties regarding the impact of Pleistocene and Holocene sedimentation preclude refining these estimates any further. In addition, sediment thickening and fluid expulsion near the deformation zone tends to balance the effect of recent sedimentation [Wang et al., 1993].
 Examining the basement temperatures of the north/central/south corridors just west of the CSZ deformation zone individually, and using heat flow values uncorrected for sedimentation of 125 mW/m2 (north corridor) [Davis et al., 1989, 1990; Hyndman and Wang, 1993, 1995], 110 mW/m2 (central corridor) [Moran and Lister, 1987], and 150 mW/m2 (south corridor) (this paper), with sediment thicknesses from ODP Site 888 and DSDP Site 174, we calculate approximate basement temperatures of 234°C in the north at crustal age of 6 Ma just west of Vancouver Island (consistent with Hyndman and Wang [1993, 1995]), 170°C at 8 Ma in the central corridor at 47° N, and 106°C at an age of 8 Ma in the southern portion of the basin off the Oregon coast (Table 1). The values for the central corridor are not substantially different from 200°C for basement temperatures off the Olympic Peninsula estimated by Hyndman and Wang , obtained using a different compilation. However that study used the data from Shi et al. , most of which are from stations within the deformation zone (their Figure 1) where fluid expulsion has been shown to occur [Carson et al., 1991, 1994], and to artificially elevate heat flow values [Wang et al., 1993].
Table 1. Estimated Temperatures of the Sediment/Basement Interface Just West of the CSZ Deformation Front for the Three Corridors of Cascadia Basin
Hyndman and Wang  previously estimated basement temperatures of 200°C at the deformation front of the subduction zone offshore Oregon. This is a considerably higher estimate than our basement temperatures of 100°C for the southern corridor, which are based largely on our new multicorer heat flow data from those stations distant from Nubbin Seamount. From the previous discussion of possible corrections to the heat flow data from sedimentation processes, it is important not to place too much confidence in these estimates, which have large uncertainties. Because of the uncertainties, it would be more conservative to conclude that sediment/basement interface temperatures along the 6 to 8 Ma isochrons in Cascadia Basin can be estimated as >250°C in the north, between 100° and 200°C for the central corridor, and ∼100°C in the southern corridor respectively.
 These temperatures are within the critical range for determining the landward extent of the observed updip limits for the aseismic section of the CSZ, which has been associated with clay dehydration that occurs near 100°C [Oleskevich et al., 1999; Hyndman and Wang, 1993, 1995]. The rough estimates based on our heat flow compilation support the conclusion from previous thermal models that suggested that the ∼100°C updip limit of aseismic subduction faults would be in younger crust (6 Ma) and at shallower depths near Vancouver Island in the north than near the Oregon margin in the south [Oleskevich et al., 1999]. In any case, it appears that the entire northern corridor of Cascadia Basin has higher average geothermal flux than the southern section, as suggested by bottom water temperature data of Hautala et al. . Further, temperatures of igneous basement as the Juan de Fuca plate as it enters the CSZ are also significantly higher in the north than in the south, and this temperature difference may be higher than that previously estimated [Hyndman and Wang, 1995].
 Interpretation of compiled archive geophysical data, obtained over several decades with a variety of instruments and processing techniques, carries substantial risk. Data quality and processing techniques within the compilation are not consistent, and spatial coverage is general nonuniform and may contain errors and biases inherited from the initial studies. Requiring internal consistency of data obtained from multiple sources is the only realistically available tool, but is generally insufficient. However, eventual confirmation from systematic geographical coverage, using modern data acquisition and analytical techniques, is the ultimate remedy for this uncertainty. Prior to these modern surveys, advances in understanding oceanographic processes must sometimes be cantilevered out beyond the current optimum standards for data, and the compilation of imperfect archive data is sometimes the only possible mechanism for the generation and initial testing of new hypotheses.
 With these caveats, the present compilation allows us to draw several conclusions, in descending order of confidence. First, the inference from a water column heat budget [Hautala et al., 2005, 2009] that geothermal flux through the northern part of Cascadia Basin seafloor is significantly higher than the southern section is supported by direct heat flow probe measurements. Previous published heat flow data indicate a midbasin average value above 200 mW/m2 for the northern corridor [Davis et al., 1999; Thomson et al., 1995], while the present compilation suggests the southern and middle corridors have average midplate heat flow values between 100 and 150 mW/m2. The compiled heat flow versus age profile for the southern corridor lies well below any predicted sediment-corrected, conduction-only reference value. In the southern corridor, the presence of abundant still visible basement outcrops provide pathways for fluid and heat exchange between seawater and the crustal reservoir at all crustal ages. Newly acquired heat flow and water column data from near the Nubbin basement outcrop showing venting of warm hydrothermal fluid, at a crustal age of 7 Ma and only 50 km from the deformation front near the Oregon margin, strongly supports this interpretation for the southern corridor.
 In the central corridor of Cascadia Basin, the compiled heat flow versus age profile shows similar reduced values as the southern corridor, being substantially below the reference conduction-only prediction at all crustal ages, even when corrections for sedimentation are applied. The lack of exposed basement outcrops at the present time for crustal ages older than 2 Myr within this central corridor suggests that the corridor was cooled by exposed basement ventilators for much of the 8 Myr history, similar to the southern corridor, but these exposed basement outcrops have been recently covered by the Late Pleistocene turbidites derived from the Cordilleran Ice Sheet. The turbidite and pelagic sediment cover in the last 15,000 years would have blocked any further exchange between active crustal hydrothermal reservoir and seawater and the central corridor should be presently undergoing reheating due to thermal rebound as predicted by the Hutnak and Fisher  model.
 Finally, using the assumption that sediments along the 6 to 8 Ma crustal age isochrons adjacent to the North American margin are similar in physical properties, extrapolation of our compiled heat flow data supports the previous suggestion by Hyndman and Wang  that the sediment/basement interface for the Juan de Fuca plate at the deformation front just prior to subduction is substantially warmer (more than 150°C higher) in the north than in the south. It is not clear if the temperature contrast between the north and southern portions of the igneous portion of the Juan de Fuca plate continues eastward into the subduction zone, or how significant this temperature gradient is in controlling the subduction process.
 This study illustrates that although progress has been made regarding our understanding of fluid circulation within the upper crustal reservoir, areas such as the central corridor of Cascadia Basin may have a complex thermal history that is in part controlled by sedimentation from the adjacent continents. If temperature of the incoming slab is a primary factor in positioning the locked portion of the subduction zone thrust fault, the central corridor may be the first observed example of Late Pleistocene turbidites impacting that process. Finally, the strong south-to-north geothermal heat flux gradient, combined with a consistent topographic slope rising to the NE, appears to be a major driving force for the general bottom water flow within the basin. This suggests a potentially important role for this small basin in better understanding the driving forces of global bottom water circulation.
 This research was supported by NSF grants OCE-0452565 and OCE-1037870 and Washington Sea Grant R/NP-6 to HPJ. N. Pisias contributed to the modifications to the OSU multicorer. D. Hasterok, M. Hutnak, K. Homola, and an anonymous reviewer contributed significantly to the improvement of the original manuscript.