• Cardone, V. J., , J. G. Greenwood, , and M. A. Cane, 1990: On trends in historical marine wind data. J. Climate, 3 , 113127.

  • Carter, D. J. T., 1982: Prediction of wave height and period for a constant wind velocity using the JONSWAP results. Ocean Eng., 9 , 1733.

    • Search Google Scholar
    • Export Citation
  • Carter, D. J. T., , and L. Draper, 1988: Has the north-east Atlantic become rougher? Nature, 332 , 494.

  • Cherniawsky, J. Y., , and W. R. Crawford, 1996: Comparison between weather buoy and Comprehensive Ocean-Atmosphere Data Set wind data for the west coast of Canada. J. Geophys. Res., 101 , 1837718389.

    • Search Google Scholar
    • Export Citation
  • Freeland, H. J., 1990: Sea surface temperatures along the coast of British Columbia: Regional evidence for a warming trend. Can. J. Fish. Aquat. Sci., 47 , 346350.

    • Search Google Scholar
    • Export Citation
  • Gower, J. F. R., 1996: Intercalibration of wave and wind data from TOPEX/Poseidon and moored buoys off the west coast of Canada. J. Geophys. Res., 101 , 38173829.

    • Search Google Scholar
    • Export Citation
  • Gulev, S. K., , and L. Hasse, 1998: North Atlantic wind waves and wind stress fields from voluntary observing ship data. J. Phys. Oceanogr., 28 , 11071130.

    • Search Google Scholar
    • Export Citation
  • Hamilton, G. D., 1980: NOAA Data Buoy Office programs. Bull. Amer. Meteor. Soc., 61 , 10121017.

  • Mantua, N. J., , S. R. Hare, , Y. Zhang, , J. M. Wallace, , and R. C. Francis, 1997: A Pacific interdecadal climate oscillation with impacts on salmon production. Bull. Amer. Meteor. Soc., 78 , 10691079.

    • Search Google Scholar
    • Export Citation
  • Neu, N. H. A., 1984: Interannual variations and longer term changes in the sea state of the North Atlantic from 1970 to 1982. J. Geophys. Res., 89 , 63976402.

    • Search Google Scholar
    • Export Citation
  • Reynolds, R. W., , and D. C. Marsico, 1993: An improved real-time global sea surface temperature analysis. J. Climate, 6 , 114119.

  • Smith, S. D., 1988: Coefficients for sea surface wind stress, heat flux and wind profiles as a function of wind speed and temperature. J. Geophys. Res., 93 , 1546715472.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Map showing the locations of the buoys whose data are discussed here, and of weather station P, indicated by P at 50°N, 145°W

  • View in gallery

    Variation of mean SST with lat for all buoys, showing (solid line) the avg decrease of SST with lat, and (dashed line) the warmer, offset relation for coastal buoys not influenced by upwelling. Solid diamonds indicate NOMAD buoys

  • View in gallery

    Sea surface temperature data from buoy 46004 (combining U.S. and Canadian data) showing (top plot) the original time series and (lower plot) the values with the annual cycle (first and second harmonics) removed. A linear regression indicates a strong warming trend (0.064°C yr−1)

  • View in gallery

    Wind speed time series for buoy 46004 (legend as in Fig. 3) showing the effect of the change of anemometer height in Jun 1983. This causes an apparent negative trend with time (−0.06 m s−1 yr−1) as shown by the heavy line

  • View in gallery

    Wind speed time series for buoy 46004 (legend as in Fig. 3) multiplying wind speeds measured before Jun 1983 by a factor of 0.9 to compensate for the change of anemometer height. This reduces the apparent negative trend with time to −0.015 m s−1 yr−1 (as listed in Table 3), but does not remove the apparent step near 1982

  • View in gallery

    Monthly mean wind speeds measured at buoy 46132 (legend as in Fig. 3). This is an example of a short time series, showing a large (0.26 m s−1 yr−1) and statistically significant trend over the 5 yr 1994–99. See Table 3

  • View in gallery

    Monthly average significant wave height time series from buoy 46005, showing apparent increase by 0.46 m (17%) in 22 yr. Legend as in Fig. 3

  • View in gallery

    Comparison of computed trends (fractional changes per year, trends divided by mean values) of wind speed and wave height for all buoys. Buoys giving more than 20 yr of data are shown as solid diamonds. Seven buoys showing relatively large trends are identified for comparison with Table 3

  • View in gallery

    Global trends in SST over the period 1982–99 derived from the Reynolds ship and satellite blend 1° dataset (Reynolds and Marsico 1993). Warming rates from (dark blue) negative, or cooling, to (red), the most rapid warming, up to 0.1°C yr−1, are indicated by the color sequence blue, green, yellow, red. Average global warming shown by these data for the period is about 0.01°C yr−1. The west coast of Canada is identified as an area of relatively rapid warming at up to 0.07°C yr−1. Apparent rapid warming off Greenland and Labrador is in regions where data accuracy is reduced by ice cover

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 2 2 2
PDF Downloads 2 2 2

Temperature, Wind and Wave Climatologies, and Trends from Marine Meteorological Buoys in the Northeast Pacific

View More View Less
  • 1 Institute of Ocean Sciences, Sidney, British Columbia, Canada
© Get Permissions
Full access

Abstract

Time series of sea surface temperature, wind speed, and significant wave height from meteorological buoys off the west coast of Canada and the adjacent United States are long enough and of sufficient quality to be useful for studying interannual variability and trends. Long-term averages of data provide a precise climatology of surface temperature, wind speed, and wave height for locations along- and offshore. Data from many of the buoys suggest a warming trend, but only three buoys show statistical significance, and sheltered buoys show no increases. Both the wind speed and the wave height data show an increasing trend with more statistical significance. Future data from these buoys should benefit from better calibration and a wider variety of sensors, as well as from longer time series.

Corresponding author address: Dr. J. F. R. Gower, Institute of Ocean Sciences, P.O. Box 6000, Sidney, BC V8L 4B2, Canada. Email: gowerj@pac.dfo-mpo.gc.ca

Abstract

Time series of sea surface temperature, wind speed, and significant wave height from meteorological buoys off the west coast of Canada and the adjacent United States are long enough and of sufficient quality to be useful for studying interannual variability and trends. Long-term averages of data provide a precise climatology of surface temperature, wind speed, and wave height for locations along- and offshore. Data from many of the buoys suggest a warming trend, but only three buoys show statistical significance, and sheltered buoys show no increases. Both the wind speed and the wave height data show an increasing trend with more statistical significance. Future data from these buoys should benefit from better calibration and a wider variety of sensors, as well as from longer time series.

Corresponding author address: Dr. J. F. R. Gower, Institute of Ocean Sciences, P.O. Box 6000, Sidney, BC V8L 4B2, Canada. Email: gowerj@pac.dfo-mpo.gc.ca

1. Introduction

An array of 26 meteorological buoys off the west coast of the United States and Canada provides weather and surface ocean data for coastal and offshore waters in the Gulf of Alaska north of 45°N and east of 160°W (Fig. 1). The first buoys were 12- and 10-m discus buoys installed in the 1970s by the U.S. National Data Buoy Center. These are now replaced by an expanded network of 6-m Navy Oceanographic Meteorological Automatic Device (NOMAD) buoys offshore, and 3-m discus buoys near shore and in sheltered waters, deployed by the United States and by the environment and fisheries departments of the Canadian federal government. The first U.S. buoys were deployed in 1972–77 at six offshore locations numbered 46001 to 46006. Four U.S. nearshore buoys were added in 1984–91. The first Canadian buoys were installed in 1987, at which time location 46004 was transferred to Canadian responsibility. The Canadian network was brought up to its full strength of 16 operational buoys in 1993. An additional experimental buoy was added in 1998 for testing sensors for biological and optical variables. The buoys measure wind speed and direction, wave height and spectrum, surface water and air temperature, and atmospheric pressure. All data are transmitted in real time at hourly intervals.

Although the operational purpose of the buoys is short-term weather forecasting, the time series of data from the buoys are long enough (6–27 yr) to be a useful indicator of long-term trends, and to show the signature of short-term “climatic” effects such as El Niño. Any apparent trends in temperature, wind speed, and wave height need to be evaluated as a check on the quality of the buoy data. Wind and wave data from the Canadian buoys have been used in validation of satellite sensors such as the TOPEX/Poseidon altimeter (Gower 1996) and of the Comprehensive Ocean–Atmosphere Data Set (COADS; Cherniawsky and Crawford 1996).

This paper is based on monthly average values of sea surface temperature, wind speed, and significant wave height (SWH, the height exceeded by one-third of the waves, H1/3, computed as 4 times the standard deviation of the measured height values) in the time period 1972–99. Recent Canadian data for the period May 1990 to May 1999 were derived from a new compilation provided by Environment Canada for this study. Data from the U.S. buoys are available on the Web at http://www.ndbc.noaa.gov.

2. Buoy data

The positions of the 26 buoys in the west coast U.S. and Canadian networks are shown in Fig. 1. Offshore buoys were originally 10- and 12-m discus buoys, but these were replaced in 1982–92 by NOMAD shiplike hulls measuring 6 m × 4 m. All other buoys have 3-m discus hulls. The offshore buoys are located about 400 km away from the coast, with the exception of 46006, which is positioned about 1000 km offshore to give greater warning of weather systems coming from the west. The weather ships at Ocean Weather Station Papa (position indicated by the letter P in Fig. 1) provided this same warning for Canada from 1949 to 1981.

The offshore buoys are listed in tables in this paper in an anticlockwise order starting at 46003. Data from the line of nearshore buoys, moored in exposed positions within 100 km of the coast, are listed in a north-to-south order from 46205 off the north end of the Queen Charlotte Islands south to 46027 off northern California. A further eight buoys are located in more sheltered Canadian coastal waters, behind the Queen Charlotte Islands and Vancouver Island. Two of these buoys (46131 and 46146) are located in the Strait of Georgia. A further two (46181 and 46134) are located in narrow coastal inlets. Buoy 46134 was installed only recently and its data are not considered here.

Nominal measurement accuracy for the Canadian buoy data is ±0.5°C for sea surface temperature, ±0.3 m s−1 for wind speed, and 10% for significant wave height (Axys Environmental Consulting Ltd., 2001, personal communication). Water temperatures are measured by thermistor at a depth of 80 cm. The wind speed accuracy is that specified for the Young Wind Monitor Model 05103, and is assumed to apply to all deployed instruments. An additional source of error at low wind speeds is the threshold wind speed required to turn the rotor, which is specified as 1 m s−1. Wind speed accuracy for U.S. buoys is quoted as ±1 m s−1 or 10%, whichever is greater (Hamilton 1980). Wind speeds are measured by two propeller anemometers mounted at 3.7 and 4.7 m above the water. The higher anemometer is used unless its data are suspect. On the older 10- and 12-m discus buoys, wind speed was measured at a height of 10 m, and water temperature at a depth of 1.5 m. Significant wave height accuracy is that maintained by calibration on shore prior to deployment. Sensors are rejected if errors measured in predeployment tests are outside the 10% range.

Buoys are rotated among locations on the annual service cruises, so that a reasonable error model is to assume independent random errors of the above-specified magnitudes that remain fixed for each year. On this basis, calibration errors will limit the accuracy with which average values and trends in the data can be determined to the values listed in Table 1 for time series covering 5, 10, and 20 years.

Both winds and temperatures are averages over 10-min periods computed once each hour. Winds are vector averages. The wave heights are deduced from vertical “heave” accelerometers mounted inside the buoys. In the NOMAD hulls, the sensors are gimballed to remain vertical when the buoy tilts. In the 3-m hulls, the accelerometers are strapped down, and buoy tilt is a possible source of error. Buoy wave measurements are averages of 37 min of data collection.

This study is based on monthly mean values of sea surface temperatures (SST), wind speeds, and significant wave heights computed from the hourly buoy measurements up to and including May 1999. Means are computed for all months for which at least 300 hourly values (approximately 40% of the maximum possible) are collected. For computing anomalies from the monthly values, a mean seasonal cycle is computed from all available data for each buoy. The cycle is assumed to consist of an annual sinusoid and a second harmonic (2 cycles per year) whose amplitudes and phases minimize the standard deviation of the residual monthly values, obtained by subtracting these cycles from the data.

Sections 3 and 4 of the paper discuss first the climatologies and then the trends in sea surface temperature, wind speed, and wave height. Response of the buoys to the major El Niño events in 1982/83 and 1997/98 are discussed in section 5. Temperature trends measured by the buoys are also compared to trends derived from the global Reynolds dataset (Reynolds and Marisca 1993) in section 6.

3. Climatologies of sea surface temperature, wind speed, and wave height

The mean value of SST, wind speed, and wave height and the amplitude of the seasonal cycles deduced for each buoy are summarized in Table 2. The order of the buoys (northwest to southeast in three series) reflects their locations as shown in Fig. 1. Consistent patterns in the values in the table suggest that relative calibration is accurate to the levels suggested in Table 1.

The mean temperatures are about 0.5°C lower for each degree of latitude northward (Fig. 2). The values at most NOMAD buoys (black diamonds) fall within 0.1°C of the linear relation (solid line) SST (°C) = 10.25° − 0.475° (latitude − 50°), which applies for the latitude range 40°–55°. Buoy 46003 is about 2.5°C colder than this relation, and buoy 46184 is 0.7°C warmer. These two differences can be explained by greater than average horizontal heat flux at these locations, the Alaska Stream bringing colder water from the northeast at 46003, and the Alaska gyre bringing warmer water from the southwest at 46184.

All Canadian coastal 3-m buoys except 46206 follow an offset relation (dashed), which is about 1.3°C warmer than the NOMAD relation. Such an increase would be expected due to a combination of more hours of sunshine at locations in the rain shadows of the Vancouver and Queen Charlotte Islands, and northward water flow near shore due to the buoyancy-driven Haida current. Southern coastal buoys will be cooled by the effects of coastal upwelling. Buoys 46041 to 46050 lie close to the NOMAD relation, while 46027 is about 3°C cooler. This buoy is closer to the shore (11 km) than the other U.S. buoys, in an area of intense upwelling.

The annual cycle of SST has an amplitude of about 4°C offshore, but is less near the coast where upwelling tends to reduce summer temperatures. Coastal upwelling reduces the amplitude of the annual cycles at lower latitudes from 3.8°C off the northern Queen Charlotte Islands to 2°C at buoy 46050 off central Oregon and 0.6°C (less than the residual variation in monthly mean values) at buoy 46027 off northern California. It is highest (5°C) in the more stratified waters of the Georgia Strait and coastal inlets where a shallow surface layer is strongly warmed in summer.

The mean wind speed is highest at 7.9 m s−1 for buoy 46003. The three Canadian offshore buoys give very similar averages (7.47 ± 0.03 m s−1), while the U.S. buoy to the north and the three to the south all show mean speeds 7.0 ± 0.10 m s−1. This difference is an order of magnitude larger than the 0.03 m s−1 error expected from random calibration errors (Table 1) but observed wave heights are in the same ratio (see later), suggesting a real difference with position. Consistency between several groups of buoys in Table 2 is about ±0.1 m s−1, comparable to the values in Table 1.

The amplitude of the annual cycle of wind speed for most buoys is about 1.5 m s−1, or 0.22 ± 0.05 of the mean wind speed. Buoy 46204 shows the highest annual-cycle amplitude (2.4 m s−1). The exposed buoys 46208, 46132, and 46029 show low values near 1.1 m s−1. The southernmost buoys (46050 and 46027) form a triangle of lower values (0.6 to 1.0 m s−1) with buoy 46002. The three buoys in sheltered waters also have low annual-cycle amplitudes in this range.

The mean significant wave heights are highest (2.7–3.0 m) for the offshore NOMAD buoys, with four (46003 and the Canadian NOMADs) in the range 2.9–3.0, and the remaining U.S. NOMADs (46001 and those to the south) giving values near 2.7 m. These are the same two groups noted earlier as showing different mean wind speeds, suggesting that the differences in mean wind speeds and in mean wave heights are both real. Northern exposed 3-m buoys give values in this same range (2.6–2.8 m), while buoys to the south on the continental shelf show lower SWH (2.1–2.4 m). The three buoys in sheltered waters show a relatively low annual-cycle amplitude, even allowing for the lower mean wind speed, as would be expected from the reduced fetches at these sites.

It is interesting to note that the measured long-term-mean SWH values for all 18 exposed buoys are within 5% of values computed using the measured long-term-mean wind speeds in the Joint North Sea Wave Project (JONSWAP) fully developed relation (Carter 1982), with buoy wind speeds increased by 10% to convert from 5- to 10-m measurement height (see later), and an arbitrary 1.25 m added as contribution from swell. This suggests an absence of calibration errors larger than about 5%. The ratio of the annual-cycle amplitude to the mean wave height is about double the equivalent ratio found for the wind speeds, as would be expected from the square-law relation between wind speed and fully developed wave height (Carter 1982).

4. Trends in sea surface temperature, wind speed, and wave height

The trends of SST, wind speed, and wave height with time are shown in Table 3. Most trends in SST suggest warming, but only three of these have statistical significance. An apparent warm trend would be increased by the effect of the recent 1997/98 El Niño, which caused a significant rise in sea surface temperatures near the end of the time period (see next section). For buoys deployed in 1982 or earlier, a compensating effect would be expected from the 1982/83 El Niño. Data for the periods affected by these two events (October 1982–May 1983 and August 1997–March 1998, see next section) are omitted from the trend analyses. The statistical significance is calculated assuming that SST anomalies are correlated over 3 months, since autocorrelations for SST give values of about 0.7, 0.5, 0.3 for the first 3 months of lag. Wind speed and wave height anomalies were found to give autocorrelations below 0.2 for all lags, so statistical significance is computed assuming months are independent.

The eight offshore buoys (first block of data in Table 3) all suggest positive trends, though five show low statistical significance. Those deployed by the United States (sequentially numbered 46001 to 46006) have the longest data records (22–26 yr) and therefore give the best indication of real, long-term trends. Three neighboring buoys (46001, 46184, 46004) show trends that are relatively large (0.03°, 0.1°, 0.06°C yr−1) and statistically significant (0.02 chance of random occurrence, or less). Figure 3 shows the time series for buoy 46004, which gives the most significant trend in Table 3. This and buoy 46184 are the only buoys that show a trend above the rates listed in Table 1.

One possible contribution to a long-term temperature increase for some offshore buoys will be the reduction in sensor depth from 1.5 to 1 m, when the buoys were converted from large discus to NOMAD. This might contribute to a positive trend since the shallower sensor would sense slightly warmer temperatures in summer when the water is stratified, and about the same temperature in winter when the water would be better mixed by storms. The average effect is expected to be small, but could contribute at all offshore locations except 46184 and 46036, at which NOMAD buoys were deployed from the start in 1987 and 1988.

For buoy 46004 (Fig. 3), the conversion from large discus to NOMAD was made in June 1983. If the data before June 1983 are omitted, then the warming rate drops to 0.041°C yr−1. No step in the data is evident, though the early data are on average about 1°C cooler than the average after June 1983. If this change is due to the change in hull type, then it should show a seasonal signal, but Table 4 shows that there is no sign of this. The increase is comparable in the spring and summer when stratification is expected (0.80°, 1.09°C) and in the fall and winter when it is not (0.93°, 0.82°C).

Trend analysis in wind speed is complicated for some buoys by the changeover from large discus to NOMAD hulls, with a reduction of anemometer height from 10 to 5 m. Figure 4 shows the time series for buoy 46004, which was changed over in June 1983. The best-fit estimate of the apparent reductions in measured wind speed at the change of anemometer height for the six buoys 46001 to 46006 are 0.915, 0.95, 0.862, 0.87, 0.925, 0.895. The factor 0.87 applies to buoy 46004 as plotted in Fig. 4. Since the same change in height occurred on all buoys, an average factor of 0.9 was applied to all 10-m data (see Fig. 5 for buoy 46004). This is the average of the empirically determined factors and represents a slightly larger change than the factor 0.94 proposed by Smith (1988) for the ranges of wind speeds and air–sea temperature differences measured by the buoys. The computed trends are then as listed in Table 3.

Table 3 shows that there is an apparent trend of increasing wind speed for most buoys, though among the long time series NOMAD buoys (those covering more than a 260-month time interval), three show positive and three show negative trends. However, positive trends are larger and more statistically significant. The probabilities of random occurrence for wind speed and wave height are calculated assuming that monthly anomaly values are independent, as noted earlier. The data plot for 46004 (Fig. 5) suggests that the negative trend for this buoy is largely due to an apparent step (calibration error?) in early data before 1981. Buoy 46132 shows the fastest apparent increase of 0.25 m s−1 yr−1 (Fig. 6), though data from this buoy cover the relatively short period of 66 months (5.5 yr). All of the 10 buoys showing statistically significant trends (probability of random occurrence equal to or less than 1%) show increases, with an average value of 0.09 m s−1 yr−1, or about 1 m s−1 over 10 yr. Trends for the buoys with the longer time series are smaller and are comparable to error values shown in Table 1, suggesting that the larger observed trends will not continue over long periods.

Trends in significant wave height are once again mostly positive, and by roughly the same percentages as for wind speed, but the computed rates of change are again comparable to expected errors due to calibration shown in Table 1. The highest and most significant trend listed is for buoy 46005 (Fig. 7). This plot suggests that most of the apparent trend is due to the low early (before 1981) and high recent values, with a long static period in between. Winds for this buoy show no corresponding relative low before 1981, though there is evidence of a recent increase after 1994. The anemometer height at this location was changed from 10 to 5 m in 1986. Among the 3-m buoys, the computed rate of increase is again highest for buoy 46132 (15% in a 5-yr period).

Real long-term changes in SWH should correlate with the trends in wind speed, though there is a major contribution to SWH at exposed buoys from swell due to remote winds as noted earlier. Figure 8 shows a comparison of trends expressed as fractions (trend divided by mean value). Buoys with long data records (solid diamonds) show small trends. Buoys 46050, 46131, 46132, 46147, and 46206 show positive trends in both wind speed and wave height, suggesting these trends may be real, if only short term. Buoy 46132 shows the most rapid increase in both wind speed and wave height. The two buoys in Georgia Strait (46146 and 46131) show an increase in SWH, though only buoy 46131 shows a significant increase in wind speed.

5. Response to El Niño events in 1982/83 and 1997/98

Two major El Niño events occurred in the time period covered by these buoy observations, resulting in significant warming along the equatorial east Pacific in the periods October 1982–May 1983 and July 1997–February 1998. Table 5 shows the mean anomalies in sea surface temperature, wind speed, and wave height, calculated over these two 8-month periods. Two 4-month “non–El Niño” periods ending and starting two months before and after the 8-month periods, were used as a baseline to remove effects of long-term trends.

The table shows a positive response to the 1997/98 El Niño in temperature among coastal buoys, and a much weaker response in wind speed and wave height. The offshore NOMAD buoys show a response to both events that appears positive in the north and negative in the south. All 3-m buoys were deployed after the 1982/83 event. During the 1997/98 event, the exposed coastal buoys from 46205 in the north to 46050 in the south warmed by an average of 1.4° ± 0.5°C. The more southerly buoys show the strongest warming, and it is a pity that data from buoys 46041 and 46027 are missing. Data from the four sheltered buoys round the Queen Charlotte Islands show similar warming (1.3° ± 0.14°C). The two buoys in Georgia Strait (46131 and 46146) show a very small response.

The apparent delay in response of local sea surface temperature to warming on the equator was calculated as 1 ± 1 month, using the average response of buoys 46205 to 46204 in Table 5. This delay was used in selecting the time periods for computing the results in Table 5 and for omitting periods of data from the trend analysis (Table 3).

Data from west coast Canadian coastal stations (Freeland 1990) also show the positive temperature response to the El Niños. These stations started recording much earlier (one before 1920, one in the 1920s, eight in the 1930s, one in the 1950s, five in the 1960s, and three in the 1970s for a total of 19) but are affected by greater variability due to their coastal locations, and the fact that data are collected only once each day on the daylight high tide. Five of these stations are in the Strait of Georgia. For the 1997/98 event these stations showed a small negative response (−0.13° ± 0.32°C) compared to stations on the open coast (0.97° ± 0.41°C), similar to the responses observed for the buoys for these two groups. Responses for the 1982/83 event were similar (0.26° ± 0.21°C and 1.00° ± 0.40°C), and showed the same difference between the two groups.

6. The observed trend in SST

Taken together, the offshore buoys suggest a warming trend of about 0.04°C yr−1 for the period 1977–1999, with individual buoys showing trends between 0.01 and 0.1°C yr−1. Data from the five southern-exposed coastal buoys (46206 to 46027) suggest comparable trends, though with low statistical significance. The observed trends suggest the possibility of dividing the buoys into two groups. The eight offshore buoys and the five southern-exposed coastal buoys together show an average warming trend of 0.04°C yr−1 with rms scatter of 0.025°C yr−1 in individual buoy trends about this mean. All remaining buoys taken together show an average cooling trend of 0.015°C yr−1 with rms scatter of about 0.04°C yr−1. The scatter in the observed trends is comparable to the errors suggested by Table 1, indicating a need for improved calibration.

Data from west coast Canadian coastal stations (Freeland 1990) also show a warming trend. Freeland (1990) computed an average trend from the full time series of 0.004°C yr−1 for exposed stations and 0.003°C yr−1 for sheltered ones including those in the Strait of Georgia, using data to 1989. More recent warm temperatures increase these warming trends to 0.010° and 0.007°C yr−1, respectively. The data show a more rapid warming (average for all stations) of 0.044° ± 0.016°C yr−1 since 1970, with a tendency for cooling since 1990. Stations in the Strait of Georgia show the strongest cooling since 1990 (−0.055° ± 0.045°C yr−1), compared to the average for exposed buoys (0.00° ± 0.04°C yr−1), a similar result to the buoys. Southern-exposed stations suggest cooling (−0.03° ± 0.05°C yr−1) compared to northern stations (0.01° ± 0.02°C yr−1), the reverse of the pattern suggested by the buoys. These errors are rms scatter in the individual values for four stations in the Strait of Georgia, and six northern- and four southern-exposed stations.

Reynolds and Marsico (1993) present a compilation of global monthly mean SST values derived from ships and satellite measurements on a 1° degree grid. This started at the beginning of 1982 and continues to be updated monthly. Figure 9 shows the global distribution of temperature trends for January 1982–June 1999, computed from these data. The trends are shown as an image in which more rapidly warming areas are shown as yellow and red. The mean trend computed from these data for the global ocean is about 0.01°C yr−1. The image shows many “hot spots” of more rapid warming, among them the area of the buoy array plotted in Fig. 1. Here, the Reynolds dataset shows average warming close to the coast of about 0.07°C yr−1, with less warming (0.02°–0.03°C yr−1) at the locations of the offshore buoys. Average warming trend of the NOMAD buoys is about 0.04°C yr−1 in rough agreement with the Reynolds data. The low spatial resolution of the Reynolds data makes comparison with coastal buoys more problematic. Table 3 shows average warming of 0.05°C yr−1 at the southern coastal stations (46206 and southward) and no significant warming farther north (buoys 46205–46132 and 46145–46204).

7. Conclusions

Time series from the buoys are now long enough to start to show apparent long-term trends in temperature, wind speed, and wave height, as well as signatures from short-term climatic effects such as El Niño. These trends need to be evaluated both as a check on calibration of the buoy data and to investigate whether they represent effects of long-term climatic change. Data quality from the buoys, as indicated by the consistency of the long-term buoy climatologies, appears to be good. Further improvements are planned in sensor calibration accuracy initially for SST, and in the variety of sensors deployed and variables measured.

The buoy data suggest trends of increasing values in all three variables investigated. These trends apply only to the period 1977–99 for the NOMAD buoys, and roughly 1990–99 for the nearshore buoys. In the case of SST, the data affected by the 1997/98 El Niño event were omitted from the trend analysis. The trends may not continue, and may represent part of a more local cyclic variation, such as the Pacific interdecadal oscillation (Mantua et al. 1997). The average trend of about 0.04°C yr−1 is larger than the 0.01°C yr−1 which represents the present global-mean rate of warming. It is also larger than the long-term rate derived from shore stations for this area by Freeland (1990), but is comparable to the rate shown by these shore stations since 1970.

Apparent trends in both wind speed and SWH have also been discussed for other areas of the world (Gulev and Hasse 1998; Cardone et al. 1990; Carter and Draper 1988; Neu 1984). Highest reported trends, of the order of 0.1–0.5 m s−1 decade−1 in wind speed, and 0.1–0.4 m decade−1 in SWH, are found in the North Atlantic. Many are at levels comparable to the accuracy of the data, and require longer data records for confirmation, as is the case for the buoy data presented here.

Many buoys showed increased SST during the 1997/98 El Niño event, with about 1-month delay between the warming at the equator and that shown by the buoys. Warming was strongest along the coast, but was not detected in the Georgia Strait. Offshore buoys showed a warming in the north, and a cooling in the south, both for the 1997/98 event and for the earlier 1982/83 event.

Buoy 46134 (Fig. 1) was deployed in November 1998 as part of a program to provide additional ocean data important for understanding marine ecosystems. In addition to the meteorological sensors, this buoy is equipped with an insolation sensor measuring the photosynthetically active radiation (PAR), and with in-water sensors measuring chlorophyll fluorescence, water color, underwater PAR and salinity. An acoustic profiler provides data on the distribution of zooplankton with depth. Data can be viewed online at http://www.pac.dfo-mpo.gc.ca/sci/ecobuoys/. If problems of calibration and optical fouling can be overcome, the installation of optical instruments will be made more permanent and will be extended to other buoys in the west coast network, making a start to extending the variety of time series beyond those shown here.

Acknowledgments

This work was undertaken with funding from the Ocean Climate Program of Fisheries and Oceans Canada, using data provided by the Meteorological Service of Canada, and by the National Data Buoy Center of the United States.

REFERENCES

  • Cardone, V. J., , J. G. Greenwood, , and M. A. Cane, 1990: On trends in historical marine wind data. J. Climate, 3 , 113127.

  • Carter, D. J. T., 1982: Prediction of wave height and period for a constant wind velocity using the JONSWAP results. Ocean Eng., 9 , 1733.

    • Search Google Scholar
    • Export Citation
  • Carter, D. J. T., , and L. Draper, 1988: Has the north-east Atlantic become rougher? Nature, 332 , 494.

  • Cherniawsky, J. Y., , and W. R. Crawford, 1996: Comparison between weather buoy and Comprehensive Ocean-Atmosphere Data Set wind data for the west coast of Canada. J. Geophys. Res., 101 , 1837718389.

    • Search Google Scholar
    • Export Citation
  • Freeland, H. J., 1990: Sea surface temperatures along the coast of British Columbia: Regional evidence for a warming trend. Can. J. Fish. Aquat. Sci., 47 , 346350.

    • Search Google Scholar
    • Export Citation
  • Gower, J. F. R., 1996: Intercalibration of wave and wind data from TOPEX/Poseidon and moored buoys off the west coast of Canada. J. Geophys. Res., 101 , 38173829.

    • Search Google Scholar
    • Export Citation
  • Gulev, S. K., , and L. Hasse, 1998: North Atlantic wind waves and wind stress fields from voluntary observing ship data. J. Phys. Oceanogr., 28 , 11071130.

    • Search Google Scholar
    • Export Citation
  • Hamilton, G. D., 1980: NOAA Data Buoy Office programs. Bull. Amer. Meteor. Soc., 61 , 10121017.

  • Mantua, N. J., , S. R. Hare, , Y. Zhang, , J. M. Wallace, , and R. C. Francis, 1997: A Pacific interdecadal climate oscillation with impacts on salmon production. Bull. Amer. Meteor. Soc., 78 , 10691079.

    • Search Google Scholar
    • Export Citation
  • Neu, N. H. A., 1984: Interannual variations and longer term changes in the sea state of the North Atlantic from 1970 to 1982. J. Geophys. Res., 89 , 63976402.

    • Search Google Scholar
    • Export Citation
  • Reynolds, R. W., , and D. C. Marsico, 1993: An improved real-time global sea surface temperature analysis. J. Climate, 6 , 114119.

  • Smith, S. D., 1988: Coefficients for sea surface wind stress, heat flux and wind profiles as a function of wind speed and temperature. J. Geophys. Res., 93 , 1546715472.

    • Search Google Scholar
    • Export Citation
Fig. 1.
Fig. 1.

Map showing the locations of the buoys whose data are discussed here, and of weather station P, indicated by P at 50°N, 145°W

Citation: Journal of Climate 15, 24; 10.1175/1520-0442(2002)015<3709:TWAWCA>2.0.CO;2

Fig. 2.
Fig. 2.

Variation of mean SST with lat for all buoys, showing (solid line) the avg decrease of SST with lat, and (dashed line) the warmer, offset relation for coastal buoys not influenced by upwelling. Solid diamonds indicate NOMAD buoys

Citation: Journal of Climate 15, 24; 10.1175/1520-0442(2002)015<3709:TWAWCA>2.0.CO;2

Fig. 3.
Fig. 3.

Sea surface temperature data from buoy 46004 (combining U.S. and Canadian data) showing (top plot) the original time series and (lower plot) the values with the annual cycle (first and second harmonics) removed. A linear regression indicates a strong warming trend (0.064°C yr−1)

Citation: Journal of Climate 15, 24; 10.1175/1520-0442(2002)015<3709:TWAWCA>2.0.CO;2

Fig. 4.
Fig. 4.

Wind speed time series for buoy 46004 (legend as in Fig. 3) showing the effect of the change of anemometer height in Jun 1983. This causes an apparent negative trend with time (−0.06 m s−1 yr−1) as shown by the heavy line

Citation: Journal of Climate 15, 24; 10.1175/1520-0442(2002)015<3709:TWAWCA>2.0.CO;2

Fig. 5.
Fig. 5.

Wind speed time series for buoy 46004 (legend as in Fig. 3) multiplying wind speeds measured before Jun 1983 by a factor of 0.9 to compensate for the change of anemometer height. This reduces the apparent negative trend with time to −0.015 m s−1 yr−1 (as listed in Table 3), but does not remove the apparent step near 1982

Citation: Journal of Climate 15, 24; 10.1175/1520-0442(2002)015<3709:TWAWCA>2.0.CO;2

Fig. 6.
Fig. 6.

Monthly mean wind speeds measured at buoy 46132 (legend as in Fig. 3). This is an example of a short time series, showing a large (0.26 m s−1 yr−1) and statistically significant trend over the 5 yr 1994–99. See Table 3

Citation: Journal of Climate 15, 24; 10.1175/1520-0442(2002)015<3709:TWAWCA>2.0.CO;2

Fig. 7.
Fig. 7.

Monthly average significant wave height time series from buoy 46005, showing apparent increase by 0.46 m (17%) in 22 yr. Legend as in Fig. 3

Citation: Journal of Climate 15, 24; 10.1175/1520-0442(2002)015<3709:TWAWCA>2.0.CO;2

Fig. 8.
Fig. 8.

Comparison of computed trends (fractional changes per year, trends divided by mean values) of wind speed and wave height for all buoys. Buoys giving more than 20 yr of data are shown as solid diamonds. Seven buoys showing relatively large trends are identified for comparison with Table 3

Citation: Journal of Climate 15, 24; 10.1175/1520-0442(2002)015<3709:TWAWCA>2.0.CO;2

Fig. 9.
Fig. 9.

Global trends in SST over the period 1982–99 derived from the Reynolds ship and satellite blend 1° dataset (Reynolds and Marsico 1993). Warming rates from (dark blue) negative, or cooling, to (red), the most rapid warming, up to 0.1°C yr−1, are indicated by the color sequence blue, green, yellow, red. Average global warming shown by these data for the period is about 0.01°C yr−1. The west coast of Canada is identified as an area of relatively rapid warming at up to 0.07°C yr−1. Apparent rapid warming off Greenland and Labrador is in regions where data accuracy is reduced by ice cover

Citation: Journal of Climate 15, 24; 10.1175/1520-0442(2002)015<3709:TWAWCA>2.0.CO;2

Table 1.

Errors in long-term trends introduced by calibration uncertainties discussed in the text, for an avg of 1 yr of data, and for time series of 5, 10, and 20 yr

Table 1.
Table 2.

Buoy climatologies for SST, wind speed, and wave height, showing mean values and annual-cycle amplitudes. Numbers of months with sufficient data are shown for SST and are roughly the same for all three types of data from each buoy

Table 2.
Table 3.

Measured trends in SST, wind speed, and wave height showing in each case the probability that the observed r2 value would occur by chance in normally distributed data with 3-month autocorrelation time. The length of the data record (months between and including first and last values), is shown in the last column

Table 3.
Table 4.

Comparison of mean SST anomaly (°C) for buoy 46004 before and after Jun 1983 (hull type change 10/12-m discus to 6-m NOMAD) by 3-month season. JFM refers to Jan, Feb, Mar, etc. Numbers of monthly averages and std dev (SD) of the individual monthly averages about the means are also shown

Table 4.
Table 5.

Response to 1982/83 and 1997/98 El Niño events in SST, wind speed, and SWH. For the 1982/83 event, measurements are averaged over the 8 months of Oct 1982 to May 1983, and the average over the two 4-month periods Apr to Jul 1982 and Aug to Nov 1983 was subtracted. For the 1997/98 event the equivalent periods were Aug 1997 to Mar 1998, Feb to May 1997, and Jun to Sep 1998. Wind data for Sep 1996 to May 1997 are missing for buoy 46006, so only a single “non–El Niño” period is used. The lower, separate section of the table shows average values for buoys 46003 to 46006 (eight NOMAD buoys), 46205 to 46027 (eleven 3-m buoys on the exposed coast), 46145 to 46204 (four 3-m buoys north, east, and southeast of the Queen Charlotte Islands), and 46131, 46146, and 46181 [three 3-m buoys, two in the Georgia Strait, one in a northern inlet (Fig. 1)]. Ranges of values for these groups of buoys are indicated

Table 5.
Save