Kaneohe Bay waters are normally oligotrophic and, especially in the southern basin, are strongly affected by land-derived inputs due to regional geography, proximity to land, and local land use practices (Hoover, 2002; Ringuet and Mackenzie, 2005; Hoover et al., 2006; De Carlo et al., 2007). Past studies in this area have shown that storm events can enhance greatly land and river runoff to Hawaiian coastal waters (Hoover, 2002; Tomlinson and De Carlo, 2003; Ringuet and Mackenzie, 2005; De Carlo et al., 2007). In Kaneohe Bay specifically, intense or prolonged rainfall over the bay and its watershed generates pulses of nutrient- and sediment-rich freshwater runoff to the bay, that subsequently induce phytoplankton blooms and impact carbon system dynamics and cycling (Smith et al., 1981; Laws and Allen, 1996; Kinzie et al., 2001; Ringuet and Mackenzie, 2005; Hoover et al., 2006; DeCarlo et al., 2007; Solomon et al., 2008). The phytoplankton blooms generally persist on timescales of days to weeks (Hoover et al., 2006; De Carlo et al., 2007).


Results of our work in Kaneohe Bay (Fagan and Mackenzie, 2007; Solomon, 2007; Fagan et al., 2008; Solomon et al. 2006, 2008) have shown that the waters of southern Kaneohe Bay, although remaining an overall annual source of CO2 to the atmosphere during the past three (2004-2007) years, shift to being a sink of this important greenhouse gas as a result of biological blooms that follow large inputs of freshwater and nutrients to bay waters. The storm inputs ultimately reduce the overall transfer of CO2 from inner bay waters to the atmosphere but their impacts, if any, on barrier reef sites remain largely unknown. Our data to date also tentatively suggest that climate plays a strong role in the direction and magnitude of air-sea CO2 exchange; bay waters during dry years are a more important source of CO2 to the atmosphere than during wet years. In addition, our work suggests that in coming decades, Kaneohe Bay will shift from being a net annual source of CO2 to the atmosphere to being a sink owing to rising atmospheric CO2, with poorly known effects on pH and coral ecosystems. Many estuarine systems associated with carbonate deposits exhibit the same pattern of present-day air-sea exchange of CO2 as Kaneohe Bay (Borges, 2005), thus highlighting the more global aspects of our work and its application to other systems. 

Fagan and Mackenzie (2007) suggested, on the basis of synoptic sampling during several seasonal cycles, that calcification at more distal and barrier reef sites in Kaneohe Bay is an important process for the generation of CO2 in bay waters and largely tempers any effect of CO2 drawdown in the more bloom-prone and productive inner bay waters. These authors concluded that, because Kaneohe Bay exhibits large changes in the dissolved inorganic carbon (DIC) parameters even on a bimonthly basis, high-resolution time-series data are imperative to characterize accurately the DIC system and the net annual air-sea exchange of CO2. Additionally, estuaries play a critical role in the global cycle of carbon (Borges, 2005) and because those located in subtropical to tropical regions remain rather poorly studied and commonly have coral reef communities, time-series sampling must be emphasized in order to determine accurately the direction and magnitude of air-sea exchange of CO2 in the global shallow subtropical to tropical coastal ocean today and in the future. The slightly less than three years of 4hr sampling interval CO2 and O2 data at our CRIMP-CO2 site (see Figure 2) clearly reveal seasonal and shorter term variations. The latter are often induced by local and rapidly changing forcing, thereby necessitating high temporal resolution studies.

Extensive studies of the Kaneohe Bay barrier reef environment have also been conducted by Atkinson and colleagues for a number of years (e.g., Smith and Atkinson, 1984, Marubini and Atkinson, 1999, Hochberg and Atkinson, 2000; Hearn and Atkinson, 2000; Marubini et al., 2000). These authors have shown the importance of reef flat communities in overall bay productivity as well as calcification. Their excellent work, however, has been limited by the inability to carry out parallel CO2 measurements in air and water over extended periods of time such as those that our group has made recently in southern Kaneohe Bay at the CRIMP-CO2 site (Solomon et al., 2006, 2007, 2008; De Carlo et al., 2007). Thus, based both on the potential impact of ocean acidification on reef communities and the fact that the impact of coastal processes on CO2 exchange between ocean and atmosphere have not been adequately characterized at barrier reef sites worldwide, and certainly not in Kanoehe Bay, it seems logical to shift the emphasis of our CRIMP-CO2 study to the barrier reef of Kaneohe Bay for the next biennium and beyond. To this end we have already moved our buoy to a location adjoining the inside margin of the barrier reef of Kaneohe Bay (see Figure 1), where we believe the seawater conditions reflect changes in seawater properties driven by both organic productivity/respiration and carbonate calcification/dissolution, but are minimally impacted by the land runoff on which we have focused much of our attention to date (e.g., Solomon et al. 2008). The impact of diel cycles of productivity on seawater composition is clearly evident when examining shorter periods in the CRIMP-CO2 record (Figure 3), in which variations of CO2 and O2 are, to a first approximation, inversely related.

Coral Reef Instrumented Monitoring Platform

Ocean Acidification: Impacts on Calcification and Carbonate Mineral Dissolution on the Barrier Reef of Kaneohe Bay, Hawaii

RATIONALE  Continuous anthropogenic emissions of CO2 to the atmosphere and partial uptake of this CO2 by the oceans have resulted, and will continue to result, in an increase in surface seawater acidity (lowering of pH), i.e., the process of ocean acidification, and decreasing calcium carbonate saturation state,  (e.g., Broecker, et al., 1971; Bacastow and Keeling, 1973; Kleypas et al., 1999; Caldeira & Wickett, 2003; Andersson et al., 2003, 2005; Orr et al., 2005). The only way to slow down or prevent continuing ocean acidification is to reduce the emissions of CO2 from human activities to the atmosphere. At this time, because of the current global political and socio-economic situation, a large reduction in CO2 emissions is unlikely to occur in the immediate future. Thus, surface seawater pH will continue to decline with all the ecological implications of such a change in a major Earth surface system carbon reservoir until emissions are reduced and atmospheric CO2 concentration stabilizes (SAP 2.1, 2007).

Ocean acidification raises serious concerns in terms of its effects on marine ecosystems, especially for those organisms generating shells, tests or skeletons out of calcium carbonate (CaCO3). As a consequence of ocean acidification: 1) the ability and rate at which these organisms calcify will decrease (Gattuso et al., 1999; Langdon et al., 2000; Riebesell et al., 2000; Marubini et al., 2003), 2) the physical strength of their skeletons could weaken (a condition similar to osteoporosis), 3) recruitment success of calcifiers and other organisms could be negatively affected (Kuffner et al., 2007), and 4) calcifiers could become less resistant and more vulnerable to environmental stress (Buddemeier et al., 2004). Naturally, because of their exquisite beauty, immense species diversity, significant importance to human economic and esthetic values, and the fact that they are the most “famous” marine calcifier and produce conspicuous structures made of aragonite (orthorhombic CaCO3), the fate of tropical corals and their structures, coral reefs, has received most of the attention in the ocean acidification debate (e.g., Kleypas et al., 1999; 2006; Buddemeier et al., 2004; Royal Society, 2005). Indeed, as coral reefs progressively calcify less efficiently, owing to decreased seawater pH, shifts in community structure towards non-calcifying organisms will probably occur, further impacting these valuable coastal resources upon which other marine organisms and humans depend. In addition, coralline algae, which can be the dominant framework organisms on some reefs, construct a skeleton of metastable high magnesian calcite phases, more soluble and more highly reactive than aragonite or calcite and hence will be the “first responders” via dissolution to declining pH and seawater carbonate saturation state (Morse et al., 2006).  Under the more extreme conditions that can be expected over the next two centuries with the “business as usual” scenario, computer simulations indicate that the lowered seawater pH will produce a corrosive environment in which dissolution of existing carbonate sediments will exceed production and corals will have difficulty accreting at current rates (Andersson et al., 2006).

Work to date by various researchers studying ocean acidification and related problems (e.g., Orr et al., 2005; Feely et al., 2004) has, to a large extent, focused on the open ocean and on special “model” environments that currently reflect conditions expected in the future ocean (Andersson et al., 2007). Other studies have dealt with laboratory and field experiments with calcifying organisms at elevated CO2 conditions (e.g., Gattuso et al., 1999; Seibel and Fabry, 2003; Kuffner et al., 2007; Langdon et al., 2000). To date, however, in-situ studies of ocean acidification, calcification and especially the reverse process, carbonate mineral dissolution, remain sparse for coastal reef environments or have not employed the tools necessary to provide the comprehensive framework required to understand how coastal ecosystems respond to this imminent threat. Estuaries in subtropical/ tropical areas represent a significant fraction of the global estuary system and it is critical to understand how coral and associated calcifying organism calcification and potential dissolution, and periodic storm events in these areas impact air-sea exchange of CO2, the major player in acidification of the oceans.

During the past few years, we have conducted a highly successful, high–temporal resolution (CRIMP-CO2) and synoptic spatial studies of the impact of storm induced (and land runoff) plumes on biological productivity and air-sea exchange of CO2 in Kaneohe Bay. The CRIMP-CO2 continuous monitoring program for CO2 is the longest running coastal ocean time series program in the world. The bay is a semi-enclosed estuary on the (northeast) windward side of the island of Oahu, Hawaii (Figure 1). It hosts many patch and fringing reefs as well as a large barrier reef.  Our efforts have focused primarily on the southern portion of Kaneohe bay, anchored by the CRIMP-CO2 time-series station at 21.428°N, 157.788°W (Figure 1). Kaneohe Bay was chosen as the locus of our current study because it is the largest semi-enclosed estuary in the Hawaiian islands, is relatively protected from waves by a large barrier reef and an intricate system of patch and fringing coral reefs, and the southern portion of the bay is geographically isolated from the rest of the bay and the open ocean (by Coconut Island and the Mokapu Peninsula). Thus, south Kaneohe Bay water has a relatively long residence time compared with water in the central and northern portions of the bay, and is strongly influenced by riverine input. Kaneohe Bay has a long history of impaired water quality due to point source as well as non-point source pollution (e.g., Smith et al. 1981; Laws and Allen, 1996; Tanaka and Mackenzie, 2005), and many oceanographic studies have been conducted there, including the recent synoptic study of processes controlling air-sea exchange of CO2 by Fagan and Mackenzie (2007) and the current work.  In addition, Kaneohe Bay to some extent serves as a model system of the processes operating in the large coral reef ecosystems associated with river inputs found throughout the southwest Pacific.

Figure 1: Aerial view of Kaneohe Bay, Oahu, Hawaii, showing its network of fringing, patch and barrier reefs. The old CRIMP-CO2 (1) and the new (2) locations of the buoy are indicated by yellow squares.

CRIMP-CO2 Measurements:

The time series measurements at the CRIMP-CO2 buoy (Figure 4) include determination of the mole fraction of O2 and CO2 in the air and surface water, as well as barometric pressure, humidity, temperature, and salinity/conductivity. Data are collected at three hour intervals. The seawater-intake pipe draws water from a depth of approximately 0.5 m. The CO2 concentrations are determined by a LI-COR model 820 Non-Dispersive Infrared (NDIR) sensor, which is calibrated using a reference gas with a known concentration of CO2, as well as to CO2-free air, which is created by passing air through a tube containing soda lime. Seawater CO2 concentrations are measured by equilibrating the water sample with a volume of air, and then measuring the CO2 concentration of that air. Concentrations of CO2 can be measured in the range of 0–1000 µatm at an accuracy of < 2.5 % of the measured concentration. Relative humidity and air temperature are measured using a Sensirion model SHT71 humidity sensor. Seawater conductivity and temperature are measured using a Sea-Bird Electronics (SBE-37-SMP) Microcat. Microcat data are transmitted through a communications cable to a data logger. Every three hours, the buoy system is automatically turned on and proceeds through a program that successively pumps seawater for one minute, then reads for 30 seconds each of the two-point calibration gases. Subsequently, the LICOR is turned off while the equilibrator is pumped for 5.5 minutes and is then turned back on to measure the seawater CO2 concentration in the equilibrator. The system then pumps air in for one minute through a filter, and reads the atmospheric CO2 concentration for 30 seconds. Each sample of equilibrated gas is pumped through a silica gel gas dryer before it is measured. The data are logged using PMEL Engineering Development Division designed data logging and transmission electronics, and the data are sent daily to computers at NOAA/PMEL via an Iridium modem. Wind data are obtained from the Hawaii Institute of Marine Biology (HIMB) weather station, located at Coconut Island in Kaneohe Bay, which utilizes a Campbell Scientific RM Young wind monitor, Model 05103.

Figure 4: CRIMP-CO2 buoy showing the platform on which is deployed an SBE 37 SMP microcat  CT and a YSI 6600 multiparameter sonde (not shown)

Buoy CO2 concentrations are corrected for the measured relative humidity to generate a dry value, which is then converted to “moist” (i.e., at 100% saturation) mole fractions according to the assumption that the air right above the water surface is 100% saturated (Weiss and Price, 1980). The solubility of CO2 is computed using an equation from Weiss (1974) using the temperature and salinity. This equation has an estimated accuracy of approximately 0.2 – 0.5 % (Wanninkof, 1992). Calculated fluxes represent the air-sea exchange at the CRIMP-CO2 location. The area-specific net integrated flux is calculated using a trapezoidal integration method.

Buoy O2 data are collected using a Maxtec MAX™-250 Series O2 sensor. This is a galvanic cell type sensor containing a weak acid electrolyte solution rather than the traditionally used potassium hydroxide solution.  The unique design of this sensor allows for measurements that are relatively unaffected by the presence of acidic gases, such as CO2, so is ideally suited for our research. This design also enhances the life-span of the sensor because the acid dissolves troublesome buildup of lead oxide within the cell. The electrical current produced by the reduction of O2 on the (gold) anode (beneath a teflon membrane) is measured by the sensor. The O2 content is calculated by direct proportionality, using molar ratios from the electrochemical reaction.  The sensor range is full-scale (0-100% O2 content) in conditions of 5-95% relative humidity and 5–40 °C, with an accuracy of ± 1–2 % (Maxtec Product Specifications). This range of conditions covers those expected at the CRIMP-CO2 location.  Because both air and water O2 concentrations are measured in the equilibrated air space, use of a sensor that is not specifically designed for the marine environment is possible.

In addition to the above, ancillary data are produced through autonomous deployment of a YSI 6600 Multi-parameter sonde mounted on the platform below the CO2 buoy (Figure 4).  This system measures chlorophyll (chl-a), turbidity, pH, conductivity, temperature, dissolved O2, and depth at 10 minute intervals and is removed (and replaced) for service every 3 weeks for data downloading and re-calibration. The higher frequency water quality data compliment the lower frequency CO2 buoy and synoptic sampling data and permit examination of very short term fluctuations in the biogeochemistry of the water column.

Figure 5: Seawater PCO2 at the CRIMP-CO2 buoy during Season 1, (upper panel); and differences in PCO2 between seawater and atmosphere (lower panel). Positive values indicate that the bay is acting as a source of CO2 to the atmosphere, while negative values indicate a sink behavior.  Note the absence of drawdowns during the second season of the study (i.e., PCO2 values do not drop below the atmospheric value of 383 µatm)

Average atmospheric PCO2 during baseline conditions was 383.4 ± 6 µatm, with a range of 370.8 – 438.1 µatm, and followed an opposite seasonal trend from seawater PCO2. Seasonal peaks occurred in February and lows in September. Variations in monthly averaged baseline values were approximately 11 µatm, and median diel variations were 2.8 µatm. Atmospheric PCO2 was observed to rise during periods of Kona winds, which often accompanied storm events (Figure 6). During such periods, atmospheric PCO2 increased up to 59.7 µatm within a three hour period, with diel variability reaching 60.6 µatm. Average atmospheric PCO2 during periods of southerly winds was 5 µatm greater than the average during northerly winds. Atmospheric PCO2 was also higher during slow southerly winds than during higher velocity winds and, below 2 ms-1, the decrease in air PCO2 is approximately linear with increasing wind speeds. At higher wind speeds, air PCO2 is near marine air averages. Increasing atmospheric CO2 concentration causes a greater difference between air PCO2and water PCO2 during the blooms, when bay water PCO2 is relatively low and facilitates air to sea CO2 flux.

Seasonal variations in atmospheric PCO2 were found to be similar to the seasonal trends in globally averaged marine air CO2 concentrations (2006 seasonal variations of ~ 4 – 5 µatm) and those at Mauna Loa Observatory (~6 µatm in 2006 and 2007) and Cape Kumukahi (~8 in 2005) on the Big Island of Hawaii. Increases in air PCO2 at the CRIMP-CO2 buoy during times of southerly winds are significant with respect to average seasonal variations (~ 11 µatm), as well as to the annual rate of global atmospheric CO2 increase (currently ~1.5–2.5 µatm yr-1 at the Mauna Loa Observatory, Hawaii), and are therefore important with regards to air-sea flux of CO2 in this region. These increases are more pronounced when southerly winds are slow, which is presumably due to greater residence time of air over land, during which there is sufficient time to accumulate CO2. The source of this CO2 may be either respiration of terrestrial organic matter or anthropogenic (i.e., combustion of fossil fuels).  Which was responsible for the accumulation of CO2 observed during this study could not be elucidated.

Figure 5: Seawater PCO2 at the CRIMP-CO2 buoy during Season 1, (upper panel); and differences in PCO2 between seawater and atmosphere (lower panel). Positive values indicate that the bay is acting as a source of CO2 to the atmosphere, while negative values indicate a sink behavior.  Note the absence of drawdowns during the second season of the study (i.e., PCO2 values do not drop below the atmospheric value of 383 µatm)

Physical, biological, and chemical processes combine to produce rapid, short-term variations in air and seawater PCO2, T, and S, which can drastically alter air-sea CO2 flux. Shifts in wind magnitude and direction also occur on very short time scales and affect air-sea CO2 flux. The variability in these parameters confirms the need for high-frequency time series data of the CO2 system at this and other coastal locations. In addition, time series analysis of the long-term, seasonal and inter-annual variability in the CO2 system can only be examined statistically with a much longer time series. The importance of ocean observing systems, such as the CRIMP-CO2 buoy, is twofold. They allow acquisition of high temporal resolution data, and also monitor variability over long periods of time. Both are necessary in order to resolve short-term variability, as well as to monitor long term trends.

Summary of Thesis and Journal Contribution by Ostrander et al. (2007)

Chris Ostrander completed his MS Thesis under the supervision of M. McManus in two years. His work focused on elucidating the physical processes that determine the extent comportment and duration of freshwater plumes in southern Kaneohe Bay. The excerpt below is taken and modified from Ostrander et al. (2007).

The study by Ostrander et al. (2007) described the temporal and spatial variability of freshwater plumes from Kaneohe Stream after storm events in the Kaneohe Bay watershed. Freshwater plumes were examined using a combination of fixed moorings, synoptic shipboard surveys, and lagrangian surface drifters. Data sets were collected over the course of 19 months from August 2005 to March 2007 with particular attention paid to storms during the boreal winters. Stream discharge and duration were found to exert a primary control on plume persistence in the southern Kaneohe Bay system. Time series data showed a strong coherence between wind forcing and surface currents, which, in combination with data derived from shipboard and aerial surveys, indicated that the spatial variability of freshwater plumes is primarily determined by atmospheric forcing.

River discharge is the primary avenue for the transport of terrigenous material and anthropogenic inputs from the continents to the world’s oceans. While the physical variability of large river plumes impacted by large coastal population centers is well documented in the literature; relatively little attention has been paid to describing the physical forcing mechanisms that govern the dynamics of small river plumes discharging into coastal environments. Devlin et al. (2001) and Gaston et al. (2006) have shown that the size and physical variability of small river plumes from coastal systems is primarily governed by wind regimes and river discharge. However, the systems described by these authors, while similar to the scale of discharge in Kaneohe Bay, occurred over the continental shelf of the ocean—not in a semi-enclosed basin. While the systems might respond similarly to physical forcing, a close examination of the possible forcing mechanisms was necessary to substantiate this assumption. The consistent and predictable response of the surface waters of the south bay to the drifter releases over the 19-month duration of this study and the spatial response of the September 2006 freshwater plume to changes in wind forcing are consistent with the results of Kimmerer et al. (1982), who found that the spatial persistence of buoyant effluent plumes from sewage outfalls near Kaneohe Stream in the south bay was governed by the strength and persistence of onshore wind stress. This onshore wind stress is instrumental in creating opposition to the pressure gradient force of the lower-density spreading surface layer and can be the primary agent in preventing the horizontal spreading and mixing of freshwater discharge. In addition to the drifter and plume response, the influence of this wind stress can be seen again in the strong coherence over a range of frequencies between current magnitude and direction and wind magnitude. The lack of coherence between salinity fluctuations and water level at diurnal and semidiurnal frequencies indicates that tidal variations did not have a significant impact on plume dynamics. The lack of a tidal correspondence to salinity and the correlation between wind-driven currents and the persistence of salinity near the stream mouth indicate that spatial dimensions of freshwater plumes emanating from Kaneohe stream are largely controlled by wind forcing. According to the characteristics observed in this study and the theory of plume advection put forth by Yankovsky and Chapman (1997), plumes emanating from Kaneohe stream are most likely “pure surface-advected plumes.” A plume of this nature will respond readily to changes in forcing direction—observations in Kaneohe showed a clear response in surface currents and plume motion to changes in wind forcing. A plume of this nature will also depend heavily on stream discharge to supply it with the low-salinity water that affords its existence. The strong correlation between both stream outflows and the persistence of freshwater in the system implies that the temporal persistence of the plume under consistent physical forcing is largely controlled by the stream discharge.

This study provided an initial examination of the physical forcing mechanisms that govern freshwater plume variability in southern Kaneohe Bay, Hawaii. To understand the full range of physical variability associated with freshwater plumes in this system, additional research is needed to determine the processes leading to the eventual fine-scale mixing of the plume and its advection out of the study area.

department of oceanography

school of ocean earth science and technology

university of hawai’i at manoa

1000 pope road msb 509

honolulu hi 96822


phone 808.956.5924

fax 808.956.7112


Figure 2: Record of carbon dioxide concentrations in air and seawater and sea surface temperature (SST) at the CRIMP-CO2 Lilipuna location in Kaneohe Bay: Note both seasonal trends with most elevated seawater concentrations of CO2 during the summer and the sharp variations (usually drawdowns) that occur on shorter time periods. Instances of sharply reduced CO2 concentrations in seawater correspond to periods immediately following nutrient inputs from storms and attendant phytoplankton blooms.  More comprehensive data from CRIMP-CO2 can be viewed at the NOAA/PMEL website: http://www.pmel.noaa.gov/co2/coastal/kbay/157w_all.htm.

Figure 3: Record of DCO2 and DO2 in seawater at the Lilipuna location of CRIMP-CO2 during thirty days beginning 10 May 2008. Note a general inverse correlation between these two parameters as predicted by the photosynthetic equation. The record demonstrates, to a first approximation, how variations in concentrations of these parameters reflect daily cycles of productivity and respiration. Other important physical parameters that influence the seawater composition are discussed later under “Progress”.

Summary of Thesis and Journal Contribution by Solomon et al. (2008)

Rachel Solomon completed her MS degree in two and a half years. Her work focused on understanding and quantifying the effect of the parameters that control CO2 exchange between the ocean and atmosphere in southern Kaneohe Bay. The excerpt below is taken and modified from Solomon et al. (2008).

The CRIMP-CO2 buoy was deployed on 30 November 2005 in the southern portion of Kaneohe Bay. The highly variable inputs to this part of Kaneohe Bay and their sporadic occurrence required high-frequency, time series investigations of water quality and storm-induced changes to understand their impact in this aquatic system. The CRIMP-CO2 study enabled a better understanding of air-sea CO2 exchange in Kaneohe Bay and, on a broader scale, was useful in the context of sub-tropical, coastal estuaries worldwide, where local climatic variability can have a large impact on the amount of material that enters the coastal ocean.

The winter season of 2005-2006 was marked by a large amount of rainfall whereas the winter season of 2006-2007 was relatively dry. The temporal juxtaposition of years with a wet and a dry winter presented us with an opportunity to compare and contrast the biogeochemical response of bay waters associated with different local climatic conditions. Three major storm events during the study allowed us to observe changes in seawater PCO2 that occur while the system is perturbed by local climatic forcing. The short-term changes observed included varying CO2 concentrations as well as instantaneous air-sea flux rates, which were subsequently used to calculate net annual CO2 exchange with the atmosphere. The storms also affected gas solubility owing to changes in sea surface temperature and salinity. Three large drawdowns of surface water PCO2 were recorded at the CRIMP-CO2 buoy following the onset of the major storm events (Figure 5). In each case, seawater PCO2 temporarily fell below that of the atmosphere. Because the three CO2 drawdowns occurred during Season 1 of the study they lead to a marked difference between seawater PCO2 between Seasons 1 and 2 (Figure 5). Each of these major drawdowns was associated with initial increases in gas solubility. The drawdowns decreased seawater PCO2, the direction of air-sea CO2 flux changed and the bay waters switched from acting as a source of CO2 to the atmosphere to acting as a sink. Storm 3 also produced a drawdown of seawater PCO2, however, it did not drop below atmospheric PCO2 and therefore did not switch the direction of air-sea CO2 flux.

Decreases in surface water temperature and salinity at the CRIMP-CO2 buoy were observed following the onset of each of the major storm events. The T and S changes increased CO2 solubility, facilitating the transfer of CO2 from the atmosphere into the water. This “sink” behavior occurs when ΔPCO2 becomes positive, and may be enhanced by T and S induced solubility changes. Compounding the T and S effects on CO2 solubility is the fact that decreases in seawater T also decrease the molar volume of CO2 and cause a decrease in PCO2.

Average seawater PCO2 during baseline (dry) conditions was 466.8 ± 48.6 µatm. Seawater PCO2 exhibited a pronounced diel cycle (peaking between 2:00 and 8:00 am HST), with a median diel variability of 34.3 µatm, and a much broader range of variability of 7–204 µatm occurring throughout the study period. Seasonal variability in average, baseline seawater PCO2was approximately 112 µatm, with maximum seawater PCO2 observed during September and minimum values seen during February. Under baseline conditions, surface seawater PCO2 at CRIMP-CO2 is high relative to air pCO2 (mean Δ PCO2 of -86 µatm). This area of the bay is therefore a source of CO2 to the atmosphere on annual to longer timescales.

Throughout the study period, instantaneous air-sea CO2 flux at the CRIMP-CO2 buoy ranged from -1.21 to 0.69 mmol C m-2 hr-1 (negative numbers indicate that the bay was a source of CO2 to the atmosphere, positive numbers indicate a sink). Average (and median) air-sea flux was -0.17 (-0.13) ± 0.03 mmol C m-2 hr-1. The uncertainty (of ± 0.03 mmol C m-2 hr-1) was taken from the standard deviation of calculations using three wind speed parameterizations. The range of instantaneous flux estimates was largest when using the WM99 relationship, and smallest when using the NLS00 relationship.

            Southern Kaneohe Bay often became a CO2 sink following storm inputs (of 0.2–0.7 mmol C m-2 hr-1) but remained a net source of CO2 to the atmosphere throughout our study period. The area-specific net annual air-sea flux estimates calculated for seasons 1 and 2 (-1.26 ± 0.15 and -2.25 ± 0.23 mol C m-2 yr-1) are of the same order of magnitude as, and bracket, a flux of -1.45 mol C m-2 yr-1 reported by Fagan and Mackenzie (2007) between Sept. 2003 and Sept. 2004. Season 1, however, was characterized by more annual rainfall than the 2003–2004 study, whereas Season 2 was characterized by less rainfall. These three estimates for net annual area-specific CO2 flux in Kaneohe Bay are similar to estimates from Hog Reef flat in Bermuda and from Yonge Reef in the northern Great Barrier Reef of -1.2 and -1.1 to -1.5 mol C m-2 yr-1, respectively (Frankignoule et al., 1996; Bates et al., 2001).