The Blue Hill Meteorological Observatory occupies a unique place in the history of the American Meteorological Society and the development of atmospheric science. Through its 129-yr history, the observatory has been operated by founder Abbott Lawrence Rotch (1861–1912), Harvard University, and the National Weather Service, and it is presently run by the nonprofit Blue Hill Observatory Science Center. While daily temperature and precipitation records are available through the National Climatic Data Center, they do not include the full record of sunshine duration data that were measured using a Campbell–Stokes sunshine recorder. We have recently digitized the observatory's original daily sunshine archives, and now present the first full collection and analysis of sunshine records extending from 1889 to the present. This dataset is unique and salient to modern climate research because the collection represents the earliest and longest continuous measurements of insolation outside of western Europe. This record provides an unprecedented glimpse into regional climate features as well as important links between global phenomena and regional climate. Analysis reveals long-term fluctuations of cloud cover and solar radiation, including signals of regional industrialization, global dimming, volcanic eruptions, and the 11-yr solar cycle. Shorter-period fluctuations include evidence of an intricate annual pattern of sunshine duration and correlations with the Arctic Oscillation, the North Atlantic Oscillation, and galactic cosmic rays.

Analysis of daily sunshine duration reveals regional and global climate patterns in a newly digitized 125-year dataset from the Blue Hill Observatory.

The Blue Hill Meteorological Observatory is located on the 635-ft (~194 m) peak of Great Blue Hill in Milton, Massachusetts, approximately 10 miles south of downtown Boston and within the Blue Hills Reservation, a 7,000-acre (~283 hectares) park and recreation area operated by the Massachusetts Department of Conservation and Recreation. Occupying the highest elevation along the immediate Atlantic coast from southern Maine to Florida, Great Blue Hill provides views of the Atlantic coast to the east; Providence, Rhode Island, to the southwest; Mount Wachusett to the west-northwest; and Grand Monadnock in southern New Hampshire to the northwest. The nonprofit Blue Hill Observatory Science Center, which currently operates the observatory, continues its dedication to careful meteorological records as well as extensive atmospheric science education and outreach activities (www.bluehill.org).

Upon being conceived and built by Abbott Lawrence Rotch in 1884/85, regular observations began on 1 February 1885, and the observatory (Fig. 1) soon became renowned for its pioneering studies of the vertical structure of the lower troposphere (Conover 1990). Rotch was appointed as the first professor of meteorology at Harvard University, established precision observing standards at Blue Hill, collected a suite of the best new meteorological instruments, and corresponded with the world's pioneering atmospheric scientists, including Vilhelm Bjerknes and Lèon Teisserenc de Bort. Rotch died abruptly in 1912, but his founding legacy left the observatory in good stead. Through directorships of visionary scientists, including American Meteorological Society (AMS) and BAMS founder Charles Franklin Brooks (director from 1931 to 1957), the observatory garnered a well-earned reputation for strong leadership, careful observation, cutting-edge research, and promotion of the enterprise of atmospheric science (Conover 1990).

Fig. 1.

Northwestern corner of Blue Hill Observatory, postcard circa 1907.

Fig. 1.

Northwestern corner of Blue Hill Observatory, postcard circa 1907.

Many of the instruments originally installed and operated by Rotch during the early days of the observatory continue to be used to gather weather observations today. One such instrument is the Campbell–Stokes sunshine recorder (Fig. 2), an instrument developed in the late nineteenth century in the United Kingdom to measure the duration of bright sunshine in a day (Brooks and Brooks 1947; Helmes and Jaenicke 1984). Sunshine recorders of this type consist of a solid glass sphere mounted concentrically above a rounded metal frame. The recorder focuses the sun's rays onto a calibrated paper card held by the frame at the focal length of the glass, thus scorching a small hole. As the sun travels across the daytime sky, the hole becomes a trace, the length of which can be translated to the number of minutes of bright sunshine that occurs. Any occultation of the sun by thick clouds will interrupt the scorch trace. On a clear day, the minimum radiation needed to scorch the specially treated paper card is reached a few minutes after sunrise and before sunset (at a threshold of approximately 120 W m−2) and is subject to slight variation based on seasonal solar geometry and clarity of the atmosphere (Harrison et al. 2008).

Fig. 2.

Photographs of (right) the Campbell–Stokes sunshine recorder in use since 1993 and (left) burned trace cards.

Fig. 2.

Photographs of (right) the Campbell–Stokes sunshine recorder in use since 1993 and (left) burned trace cards.

Despite the advent of improved measurement technology, the importance of the Blue Hill Campbell–Stokes sunshine record stems from the unique continuity of the measurements. Several similarly long sunshine records have been published from Europe but none from the Western Hemisphere (Pallé and Butler 2001; Butler et al. 2007; Wulfmeyer and Henning-Müller 2006; Sanchez-Lorenzo and Wild 2012). The original instrument remained in use from 1886 through 1993, when it was stolen from the observatory tower. When news spread of the loss of the instrument, the penitent party returned the instrument but not before an identical replacement recorder had been procured from Casella Measurement in London, United Kingdom. The speedy replacement resulted in a data gap of just 12 days. Measurements from 1993 to present have been made on identical Casella burn cards using the replacement sphere, while the original recorder is on display in the Observatory Science Center.

ARCHIVAL AND ANALYSIS METHODS.

Meticulously detailed, handwritten archival records dating back to 1889 are catalogued in bound annual volumes in the observatory library. For each day of the year, the observers measured the hourly scorch length on the burn card to record the duration of bright sunshine with a precision of 0.1 h (6 min). Further descriptions of instrument operation, siting, observatory procedures, and historical anecdotes are interspersed in the record and have been collected and published by John Conover, a long-time Blue Hill observer, researcher, and director (Conover 1990). A nearly complete daily climate record of precipitation and minimum and maximum temperatures are already digitally archived at the National Climatic Data Center (NCDC). The daily bright sunshine measurements made by the Campbell–Stokes sunshine recorder since 1965 are available from NCDC; however, the roughly 75 years of daily sunshine data prior to 1965 had never been digitally archived. Some of the other unique data elements from the early climate record, including cloud observations and visibility measurements, have been the subject of earlier analysis (Husar et al. 1981).

Each page of the handwritten volumes was photographed by digital camera. Typically, a full month of daily sun duration records (one record per day) and related observations are preserved on each page, resulting in approximately 900 high-resolution data images. The observations were mostly recorded in script pencil marks, the character of which varied by the individual recorder and resisted digitization attempts by modern optical character recognition software. We have now manually digitized 42,777 daily sunshine observations and aligned the full daily sunshine record with corresponding observations of maximum temperature, minimum temperature, and precipitation taken at Blue Hill. This full daily sunshine dataset is now available online (at www.bluehill.org), and curation of the data series is ongoing and further discovery of missing data or data corrections will be updated on the site too.

Missing data and corrections.

The handwritten archival records for daily sunshine are approximately 95% complete over the 125-yr period of record, with a total of 81 months of data missing, mostly between 1916 and 1929 (Table 1). Between 1889 and 1916 and again after 1929, there are no missing data other than several individual months and several scattered days. Other daily measurement records for these missing dates are present in the original archive books, and the observatory does possess monthly bright sunshine totals for the missing months, allowing representative reconstruction of missing data.

Table 1.

Missing daily sun duration data in the Blue Hill Observatory Campbell–Stokes record. Ancillary data for these dates are not missing.

Missing daily sun duration data in the Blue Hill Observatory Campbell–Stokes record. Ancillary data for these dates are not missing.
Missing daily sun duration data in the Blue Hill Observatory Campbell–Stokes record. Ancillary data for these dates are not missing.

The missing daily data in the 1910s and 1920s present a challenge for some methods of time series analysis. Fortunately, the 95% completeness of the original record assures that tests of long-term trends, spectral variability, and correlations with other data series are not strongly affected by the missing data. Furthermore, the monthly sunshine data record and ancillary daily cloud observations have allowed us to reconstruct data representative of the missing periods (Schneider 2001). The reconstructed daily data for missing periods were created such that the number of estimated sunshine hours is correlated with the mean of the observer's hourly daytime cloud cover estimates and adds to the recorded monthly total sunshine hours. The data presented in figures and text herein have all been subjected to analysis with and without reconstructed data for the missing dates. The presented figures include the reconstructed data, but the interpretations of the analysis and strength of cited correlations are not dependent on the reconstructions because of the small fractional extent of the reconstruction and the aggregation of so many data points. The primary result of the reconstruction is to strengthen confidence in the persistence of identified sunshine trends and variations through the period between 1916 and 1929, where original daily sunshine data coverage was previously only about 60%.

In addition to the reconstruction of the original dataset, we have also curated the record for data quality and obvious inconsistencies. Blue Hill's observers have historically considered both corrected and uncorrected measurements of sunshine duration. The corrected sunshine duration was intended to account for the minutes after sunrise and before sunset when the sun is above the horizon but not intense enough to provide the approximately 120 W m−2 necessary to burn a trace in the card (Stanhill 2003). Because the corrected sunshine duration was applied inconsistently over the record, the unadjusted sunshine is considered the most consistent long-term record of sunshine duration, and for the purposes of this study, the uncorrected minutes were used as starting points for all calculations and analyses. To prevent human-induced variability in the observations, standard procedures for reading the scorched cards and deriving bright sunshine minutes from burn lengths have been in place since the beginning of the record. For the past several decades, even greater effort has been made to ensure the homogeneity of the sunshine observations by assigning one senior observer to read and interpret the scorched cards. However, these additional measures were not in place in the early part of the record, with on-duty observers solely responsible for reading and recording the burn length. No further homogenization has been undertaken, and the daily sun data represent the unaltered original records of sun duration read from the burn cards.

Data analysis.

Here we present a preliminary analysis of the Blue Hill sunshine data using a variety of standard tools. The intent of this analysis is to illustrate some of the important trends and features that are detectable in a record of this length and to provide a foundation for future climate studies that might benefit from this dataset. These tools and analyses aim to establish a relationship between measured sunshine duration and solar irradiance and to identify long-term trends and recurring features within the dataset.

Long-term trends and seasonality

Both raw daily sunshine hours and computed solar fractions (relative to the daily maximum bright sunshine in excess of 120 W m−2) are examined in terms of their mean, variability, and distribution. The calculations employ moving-average data filters with 1- and 10-yr windows and linear fits to highlight variations over multiple time scales. The data are also binned by sunshine fraction, day of the year, month, and day of the week in order to test the time distribution of trends in the data. These trends are then considered in the context of associated global and regional observations of volcanic activity and solar radiation as well as potential anthropogenic influences, including regional and global aerosol loading, air quality measures, and contrail-induced cirrus.

Spectral variability and correlations

Robust statistical testing of sunshine relationships to natural cycles affecting the atmosphere is an integral component in the analysis of such an extensive record of daily sunshine observations. The daily resolution of the observations enables us to test shorter timescale relationships and strengthens confidence in the characterization of longer-period variations. We used a combination of wavelet and correlation analyses in order to detect potential periodic signals within the dataset. The wavelet analysis follows the well-established methods of Torrence and Compo (1998), which includes tests of significance based on theoretical wavelet spectra for white and red noise. Cross correlations were tested between the daily sun fraction measurements and the daily teleconnection indices [including Arctic Oscillation (AO), North Atlantic Oscillation (NAO), and El Niño–Southern Oscillation (ENSO)], which are available since 1950 and maintained by the Climate and Weather Linkage program of the Climate Prediction Center Zhou et al. 2001; Climate Prediction Center 2013). Correlation was also tested between daily sun fractions and sunspot numbers and between daily sun fractions and galactic cosmic ray detections [available online through the National Geophysical Data Center (2013)]. The significance at p = 0.05 of correlation r for many independent observations is given by

 
formula

where the number of independent samples N* is reduced to account for first-order autocorrelation in the two series (Chatfield 2004), x and y, by

 
formula

RESULTS AND DISCUSSION.

Seasonal sunshine patterns and long-term trends.

The data clearly reveal the seasonal cycle of solar irradiance expected in the Northern Hemisphere—the mean number of hours of monthly sunshine peaks in July at 255.8 total hours, which represents 57.5% of the possible bright sunshine, and this decreases each month to the December minimum of 126.8 h, or 46.2% of the possible bright sunshine. The monthly analysis of sunshine and sun fraction does not reveal any surprising features, but when the sunshine fraction is examined as a function of the day of the year, some of the nuanced features of sunshine climatology in eastern Massachusetts begin to emerge. Figure 3 displays a clear peak in annual sunshine fraction near 1 September and a clear minimum around 1 January. The sunshine fraction decreases fairly steadily through the fall until the early winter minimum, but the spring increases are less monotonic. For example, a marked increase in sunniness occurs through February, only to be followed by an early March decline. March ends with more sunshine, but once again falls in early April, before beginning a more consistent climb toward high summer sun fractions.

Fig. 3.

Sun fraction dataset plotted as a function of day of the year, averaged over the full period of record. The red line indicates a 2-week running mean, and the blue line indicates mean daily sun fraction.

Fig. 3.

Sun fraction dataset plotted as a function of day of the year, averaged over the full period of record. The red line indicates a 2-week running mean, and the blue line indicates mean daily sun fraction.

With respect to long-term trends in sunshine fraction (Fig. 4), the mean sunshine fraction has remained relatively steady, with long-term means varying by just a few percent, averaging 51.0% in the first half of the record and 52.6% in the second half. Despite the quasi-stationary appearance of mean sun fraction, a careful examination of moving averages and the number of sunny days (>92% sun fraction, shown as the purple line in Fig. 4) reveals some of the natural and anthropogenic factors that are closely associated with regional and global climate change and variability.

Fig. 4.

Long-term measurements and trends in sun fraction and clear days (>92% sunny): 1-yr (red) and 10-yr (black) moving averages of moving weekly sun fraction measurements, overlaid on raw daily sun fraction observations (blue points). Right axis displays the number of clear days (purple) in a moving 365-day interval.

Fig. 4.

Long-term measurements and trends in sun fraction and clear days (>92% sunny): 1-yr (red) and 10-yr (black) moving averages of moving weekly sun fraction measurements, overlaid on raw daily sun fraction observations (blue points). Right axis displays the number of clear days (purple) in a moving 365-day interval.

Modeled estimates of volcanic aerosol effects on mean optical depth through the twentieth century (Stothers et al. 1986; Zielinski 2000) show an excellent match between the largest eruptive reductions in optical depth and the five most distinct episodes of abrupt reductions in sunny days in the Blue Hill record (annotation; Fig. 4). Indeed, following the Santa Maria eruption in 1902 and the Novarupta eruption of 1912, completely sunny days were nearly absent from the Blue Hill record for more than a year. Interestingly, the modeled optical effects of El Chichón (1982) and the Pinatubo eruption (1991) are as large as the early century impacts and the amplitude of the reduction in annual clear days is approximately 30 days in each case, but that reduction begins from a baseline of around 35 clear days per year in 1900–15 compared to about 75 clear days per year from 1980 to 1995.

Perhaps most notably, we observe an apparent effect of regional industrial smoke/haze on the sunshine record in the first half of the century—the fraction of clear days is significantly suppressed from 1900 to 1950 compared to the first decade of the record and the last 60 years. The decline and subsequent recovery in the annual count of clear days is fairly steady and is correlated with the industrialization and population growth of the Northeast United States, as well as regional trends in coal-generated sulfur dioxide emissions and visibility observations from Blue Hill (Husar et al. 1981). The effect of reduced economic activity around the Great Depression also appears to have temporarily resulted in reduced atmospheric turbidity and, therefore, enhanced counts of sunny days. The number of clear days continues climbing into the mid-1960s as manufacturing intensity and emissions declined in the Northeast. It is interesting that the midcentury recovery in the number of annual sunny days is much more pronounced than the small rises in mean sun fraction: from minimums of about 25 in the 1920s back to counts exceeding 80 fully sunny days per year by the mid-1960s. At first, it seems surprising that the 1960s peak in sun fraction and sunny days would occur just prior to the effective implementation of the U.S. Clean Air Act (1970) and continued emissions reductions in the Northeast. However, the sunny years from 1963 to 1966 also correspond to a long period of dry weather at Blue Hill (including the driest year on record, 1965), and thus other factors may have influenced this sunny period. Indeed, regional emissions did continue to fall, but the regional improvements appear to have been offset by an increasing global aerosol load, including black carbon and sulfates, leading to global dimming from the 1960s through about 1990 (Streets et al. 2006). The effect of increasing contrail-induced cirrus may also be apparent in the decline in mean sunshine fraction and clear days from mid-1960s peaks. While regional and global aerosol loads are now below 1960 levels, the sunshine fraction and sunny day counts have not recovered their 1960s peaks. Steady increases in contrail-induced cirrus have been well-documented (Minnis et al. 2004) in the Northeast United States, and they appear to have offset some of the recovery of brightness associated with aerosol trends.

Fitted solar radiation trends (see the  appendix) are nearly identical to those observed in the sun fraction analysis. Nevertheless, the application of the Ångström–Prescott equation permits a more quantitative comparison between Blue Hill records and global radiation measurements of the global dimming phenomenon observed between the late 1950s and early 1990s. In general, the extrapolated solar radiation data and long-term sun fraction trends from Blue Hill (Fig. 4) reflect the same kinds of changes that took place around the world during the latter half of the twentieth century, including large regions of Africa, Asia, and Europe. Blue Hill observed a radiative decrease of approximately 7.9% between 1964 and 1985, while a global decrease of 9 W m−2 or 5.3% occurred between 1958 and 1985. The degree of global dimming varied in different latitudinal and longitudinal areas. Locally, annual totals of global radiation in Moscow, Russia, decreased 7.1% between 1958 and 1993, while irradiance was reduced 0.36 Wm−2 yr−1 in the Arctic Circle and 0.28 W m−2 yr−1 in the Antarctic. Moreover, additional recent studies demonstrate a similar rebound in solar radiation since the minimum was reached in the late 1980s (Wild et al. 2005). Surface radiation measurements taken between 1985 and 2000 by the Baseline Surface Radiation Network of the World Climate Research Programme show that more than 4 times as many observation sites have seen increasing averages of solar radiation than decreasing.

Spectral variability and correlations.

Wavelet analysis (Fig. 5) of daily sunshine at Blue Hill shows fluctuations in periodicity across the record from the subseasonal through the multidecadal scale. The wavelet power spectrum in Fig. 5 is calculated for sunny days only (greater than 92% of possible sunshine). The wavelet power is most pronounced between 8- and 16-yr periodicity, centered at 11.5 years, with especially strong power during the 1900–20 and 1960–80 time frames. Mean power over the full period (Fig. 5, right) of record demonstrates statistically significant periodicity at the p = .05 level between 10 and 13 years, peaking at 11.5 years. Wavelet power spectra and autocorrelations of average daily sun fraction also reveal the 11-yr periodicity, but the strength of the annual cycle is much stronger (as in Fig. 3), and the 11-yr periodicity is somewhat weaker than for sunny days alone. The 11-yr signal in the sunshine data is strongly suggestive of a relationship with the well-known 11-yr solar magnetic activity cycle (Gnevyshev 2007).

Fig. 5.

Clear days (>92% bright sun) wavelet power spectrum as a function of (left) period and year and (right) mean power over the full period of record. Areas inside the solid black lines in (left) and to the right of the dashed blue line in (right) are significant at the p = .05 level.

Fig. 5.

Clear days (>92% bright sun) wavelet power spectrum as a function of (left) period and year and (right) mean power over the full period of record. Areas inside the solid black lines in (left) and to the right of the dashed blue line in (right) are significant at the p = .05 level.

During the cycle, multiple solar features fluctuate in intensity, resulting in a variety of observable radiative and atmospheric changes on Earth (e.g., Friis-Christensen and Lassen 1991; Haigh 1996; Shindell et al. 1999; Coughlin and Tung 2004; Wang et al. 2012). Total solar irradiance varies in phase with the activity cycle, with an amplitude of approximately 0.1% of the 1366 W m−2 mean solar constant. In addition, high solar particle fluxes during active periods are efficient at scattering galactic cosmic rays, thus causing modulation in the cosmic ray flux reaching Earth that is strongly anticorrelated with the solar activity cycle (Pallé and Butler 2001). The intensity of cosmic rays varies globally by about 15% over one solar cycle owing to changes in the strength of the solar wind (Carslaw et al. 2002) and can also change by similar fractions over the course of several days during magnetic storms. Both the direct changes to irradiance and the potential indirect effects of cosmic ray–induced cloud condensation nucleation have been much debated as potentially significant factors in global climate (e.g., Svensmark 1998; Carslaw 2009; Kirkby et al. 2011; Magee and Kavic 2012). The influence of these potential effects upon the long sunshine record at Blue Hill should be detectable, and both would be expected to affect insolation in similar ways. An increase in irradiance during activity maxima would suggest longer late evening and early morning burns and more numerous sunnier days. Shielding of cosmic rays during activity maxima would imply reduced cosmic ray–induced cloud particle nucleation and somewhat less cloud cover. Cross correlations of daily sun fractions and measurements of cosmic ray fluxes at Climax, Colorado (data available from 1953 through 2005), show the correlation of sunshine 12 days lagging cosmic ray flux (Table 2). This is a positive correlation at a p = 0.03 confidence level. Contrary to the hypothesized anticorrelation between cosmic rays and sunshine, we observe that sunshine is actually elevated in the days following high cosmic ray counts and depressed slightly following low counts. Daily sunshine fractions show no statistically significant correlation with sunspot numbers. The finding of a positive cosmic ray count–sunshine correlation and the absence of correlation between sunspot number and sunshine adds to the evidence against a climatologically significant enhancement of cloud nucleation by cosmic rays.

Table 2.

Correlation of Blue Hill sun fraction with teleconnection indices and cosmic ray counts. Significance levels were tested using the autocorrelation-reduced estimate of independent sample length described by Eq. (A2).

Correlation of Blue Hill sun fraction with teleconnection indices and cosmic ray counts. Significance levels were tested using the autocorrelation-reduced estimate of independent sample length described by Eq. (A2).
Correlation of Blue Hill sun fraction with teleconnection indices and cosmic ray counts. Significance levels were tested using the autocorrelation-reduced estimate of independent sample length described by Eq. (A2).

Cross correlation functions with several data series have also been tested to discover any clear links between teleconnection modes and sunshine at Blue Hill. The Arctic Oscillation and the North Atlantic Oscillation are closely related modes of atmospheric variability that have been associated with disparate patterns of temperature and precipitation in the Northeast United States (Climate Prediction Center 2013; www.cpc.ncep.noaa.gov/products/precip/CWlink/daily_ao_index/teleconnections.shtml). The daily index values of each of these modes are significantly positively correlated with sunshine fractions at Blue Hill. For both comparisons, the correlation is strongest with sun fraction lagging behind the oscillation indices by 2 days. This does appear to provide a reasonable suggestion that positive NAO and AO index values are associated with subsequent sunnier days at Blue Hill and that negative index values are associated with cloudier days following. The correlation with the AO is especially strong (Table 2). Testing of correlations between monthly sun fraction and the multivariate ENSO index hint at a possible positive correlation at about 8-month lag, but it is only significant at the p = 0.25 level.

CONCLUSIONS.

The continuing record of daily sunshine measurements at Blue Hill Observatory, spanning 42,777 daily observations between 1889 and 2012, establishes a unique link between the history of meteorology in the United States and modern climate science. The observatory's emphasis on preserving original weather instrumentation and observing techniques is a boon to modern climate science, resulting in reliable weather records of unprecedented stability and continuity that are increasingly rare in an age of automated observing networks.

In this particular study, daily measurements of sunshine duration and fraction from the Campbell–Stokes instrument were examined using long-term-trend and correlation analyses. The length and consistency of the Blue Hill record enables the exposition of subtle influences and climate signals that are not usually accessible. Indeed, we observe that the sunshine fractions at the observatory are linked with a wide variety of natural and anthropogenic processes. These linkages are sometimes intuitive, as in the case of the correlation of sunshine with industrial activity and regional air quality, and sometimes less so, as with relationships to teleconnection indices and comic ray incidence. With regard to natural oscillations in the atmosphere—NAO, AO, and ENSO—we have established that each of these oscillations may play minor roles among the many influences on Boston's cloud climatology. We also find that periodicity in sunny days at the observatory is associated with the 11-yr solar cycle, but we find no significant correlation with the daily sun fraction. In addition, the finding that subsequent sunshine fraction is positively correlated with cosmic ray incidence differs from the result that would be expected if cosmic ray–induced cloud nucleation played a significant climatological role in cloud formation.

Great potential remains for further research and for the use of this dataset. For example, the sunshine record has not been tested against the multitude of other long-term continuous records kept at Blue Hill or long climate records from nearby stations. Moreover, the combination of local datasets of this extent with emerging satellite-based cloud climatologies (e.g., International Satellite Cloud Climatology Project; Rossow and Dueñas 2004) provide crucial sources of data for understanding how the Earth system responds to a changing climate. The link between careful records of past and modern climate science underscores the importance of maintaining historical climate sites and promoting access to unique data. The full record of daily Blue Hill sunshine observations is now available online (at www.bluehill.org).

ACKNOWLEDGMENTS

We thank countless past weather observers and Blue Hill foundation members for their contributions, which have made the continuing maintenance of the Blue Hill climate record possible. Current chief observer Robert Skilling and Observatory Program Director Don McCasland are gratefully acknowledged for their integral roles in hosting volunteers, keeping scrupulous weather records, and providing the general public a valuable glimpse into operational meteorology and AMS history.

APPENDIX: IRRADIANCE MODELING.

Over the past century, there have been several attempts at establishing a relationship between sunshine duration and solar irradiance [e.g., reviews by Suehrcke (2000), Hinssen and Knap (2007), and Wang et al. (2012)]. The first such empirical correlation was proposed by Anders Ångström in 1929, which was later modified by Prescott (1940) and Page (1964) to become what is now known as the Ångström–Prescott equation:

 
formula

In this equation, n/N is the recorded sunshine percentage, E0 (W m−2) is the incoming total possible irradiance entering Earth's atmosphere determined by date and latitude, and Eg ↓ (W m−2) is the calculated irradiance at the ground surface. Constants a and b are fitted parameters dependent on location and instrumental function and are discussed in more detail below. The incoming solar irradiance E0 is computed by

 
formula

In Eq. (A2), S represents the satellite-measured solar constant, taken as 1366 W m−2; Z represents the zenith angle, which is dependent on latitude, solar declination angle, and time of day; and D represents the Earth–sun distance adjustment measured in astronomical units. It is well established that S fluctuates weakly with the 11-yr cycle solar activity cycle, but we did not build this variability into Eq. (A2) because we wished to test whether evidence of the cycle's impact on sunshine can be extracted from the data.

Meanwhile a and b are fitted constant coefficients such that b represents the sensitivity of normalized global irradiance to normalized sunshine percentage, while a represents global irradiance under overcast conditions when direct radiation is below the burning threshold of the sunshine recorder (approximately 120 W m−2). Together, a + b signifies global irradiance under clear-sky conditions. To determine the best constant coefficients, given the location and operation of the Blue Hill Campbell–Stokes instrument, several years (2005–07) of data from a collocated Davis Vantage Pro2 photodiode radiation sensor were cross-referenced with the Campbell–Stokes sunshine measurements. The cross-referenced monthly data were used to solve for the values of a and b that produced the best relationship between sunshine fraction and inferred irradiance. It was determined that a = 0.14 ± 0.05 and b = 0.59 ± 0.05 by percentage error analysis of the calculated Eg ↓ and the actual Eg ↓ from the time span.

REFERENCES

REFERENCES
Brooks
,
C. F.
, and
E. S.
Brooks
,
1947
:
Sunshine recorders: A comparative study of the burning-glass and thermometric systems
.
J. Atmos. Sci.
,
4
,
106
115
,
doi:10.1175/1520-0469(1947)004<0106:SRACSO>2 .0.CO;2
.
Butler
,
C. J.
,
A.
García-Suárez
, and
E.
Pallé
,
2007
:
Trend and cycles in long Irish meteorological series
.
Biol. Environ.: Proc. Roy. Ir. Acad.
,
107B
,
157
165
.
Carslaw
,
K.
,
2009
:
Atmospheric physics: Cosmic rays, clouds and climate
.
Nature
,
460
,
332
333
,
doi:10.1038/460332a
.
Carslaw
,
K.
,
R. G.
Harrison
, and
J.
Kirkby
,
2002
:
Cosmic rays clouds, and climate
.
Science
,
298
,
1732
1737
,
doi:10.1126/science.1076964
.
Chatfield
,
C.
,
2004
:
The Analysis of Time Series: An Introduction
.
Texts in Statistical Science, Vol. 59, Chapman & Hall/CRC
,
333
pp
.
Climate Prediction Center
,
cited 2013
:
Teleconnections
.
Conover
,
J. H.
,
1990
:
Blue Hill Meteorological Observatory: The First 100 Years, 1885–1985
.
Amer. Meteor. Soc.
,
514
pp
.
Coughlin
,
K.
, and
K. K.
Tung
,
2004
:
Eleven-year solar cycle signal throughout the lower atmosphere
.
J. Geophys. Res.
,
109
,
D21105
,
doi:10.1029/2004JD004873
.
Friis-Christensen
,
E.
, and
K.
Lassen
,
1991
:
Length of the solar cycle: An indicator of solar activity closely associated with climate
.
Science
,
254
,
698
700
,
doi:10.1126/science.254.5032.698
.
Gnevyshev
,
M. N.
,
1977
:
Essential features of the 11-year solar cycle
.
Sol. Phys.
,
51
,
175
183
,
doi:10.1007/BF00240455
.
Haigh
,
J. D.
,
1996
:
The impact of solar variability on climate
.
Science
,
272
,
981
984
,
doi:10.1126/science.272.5264.981
.
Harrison
,
R. G.
,
N.
Chalmers
, and
R. J.
Hogan
,
2008
:
Retrospective cloud determinations from surface solar radiation measurements
.
Atmos. Res.
,
90
,
54
62
,
doi:10.1016/j.atmosres.2008.04.001
.
Helmes
,
L.
, and
R.
Jaenicke
,
1984
:
Experimental verification of the determination of atmospheric turbidity from sunshine recorders
.
J. Climate Appl. Meteor.
,
23
,
1350
1353
,
doi:10.1175/1520-0450(1984)0232.0.CO;2
.
Hinssen
,
Y. B. L.
, and
W. H.
Knap
,
2007
:
Comparison of pyranometric and pyrheliometric methods for the determination of sunshine duration
.
J. Atmos. Oceanic Technol.
,
24
,
835
846
,
doi:10.1175/JTECH2013.1
.
Husar
,
R. B.
,
J. M.
Holloway
,
D. E.
Patterson
, and
W. E.
Wilson
,
1981
:
Spatial and temporal pattern of eastern U.S. haziness: A summary
.
Atmos. Environ.
,
15
,
1919
1928
,
doi:10.1016/0004-6981(81)90226-2
.
Kirkby
,
J.
, and
Coauthors
,
2011
:
Role of sulphuric acid, ammonia and galactic cosmic rays in atmospheric aerosol nucleation
.
Nature
,
476
,
429
433
,
doi:10.1038/nature10343
.
Magee
,
N. B.
, and
M. J.
Kavic
,
2012
:
Probing the climatological impact of a cosmic ray-cloud connection through low-frequency radio observations
.
J. Atmos. Sol. Terr. Phys.
,
74
,
224
231
,
doi:10.1016/j .jastp.2011.10.003
.
Minnis
,
P.
,
J. K.
Ayers
,
R.
Palikonda
, and
D.
Phan
,
2004
:
Contrails, cirrus trends, and climate
.
J. Climate
,
17
,
1671
1685
,
doi:10.1175/1520-0442(2004)0172.0.CO;2
.
National Geophysical Data Center
,
cited 2013
:
Cosmic rays
.
Page
,
J. K.
,
1964
:
The estimation of monthly mean values of daily total short wave radiation on vertical and inclined surfaces from sunshine records for latitudes 40°N–40°S
.
The United Nations Conference on the New Sources of Energy
,
Vol. 4
,
United Nations
,
378
390
.
Pallé
,
E.
, and
C. J.
Butler
,
2001
:
Sunshine records from Ireland: Cloud factors and possible links to solar activity and cosmic rays
.
Int. J. Climatol.
,
21
,
709
729
,
doi:10.1002/joc.657
.
Prescott
,
J. A.
,
1940
:
Evaporation from a water surface in relation to solar radiation
.
Trans. Roy. Soc. South Aust.
,
64
,
114
118
.
Rossow
,
W. B.
, and
E. N.
Dueñas
,
2004
:
The International Satellite Cloud Climatology Project (ISCCP) web site: An online resource for research
.
Bull. Amer. Meteor. Soc.
,
85
,
167
172
.
Sanchez-Lorenzo
,
A.
, and
M.
Wild
,
2012
:
Decadal variations in estimated surface solar radiation over Switzerland since the late 19th century
.
Atmos. Chem. Phys.
,
12
,
8635
8644
,
doi:10.5194/acp-12-8635-2012
.
Schneider
,
T.
,
2001
:
Analysis of incomplete climate data: Estimation of mean values and covariance matrices and imputation of missing values
.
J. Climate
,
14
,
853
871
,
doi:10.1175/1520-0442(2001)0142.0.CO;2
.
Shindell
,
D.
,
D.
Rind
,
N.
Balachandran
,
J.
Lean
, and
P.
Lonergan
,
1999
:
Solar cycle variability, ozone, and climate
.
Science
,
284
,
305
308
,
doi:10.1126/science.284.5412.305
.
Stanhill
,
G.
,
2003
:
Through a glass brightly: Some new light on the Campbell–Stokes sunshine recorder
.
Weather
,
58
,
3
11
,
doi:10.1256/wea.278.01
.
Stothers
,
R. B.
,
J. A.
Wolff
,
S.
Self
, and
M. R.
Rampino
,
1986
:
Basaltic fissure eruptions, plume heights, and atmospheric aerosols
.
Geophys. Res. Lett.
,
13
,
725
728
,
doi:10.1029/GL013i008p00725
.
Streets
,
D. G.
,
Y.
Wu
, and
M.
Chin
,
2006
:
Two-decadal aerosol trends as a likely explanation of the global dimming/brightening transition
.
Geophys. Res. Lett.
,
33
,
L15806
,
doi:10.1029/2006GL026471
.
Suehrcke
,
H.
,
2000
:
On the relationship between duration of sunshine solar radiation on the Earth's surface: Ångstrom's equation revisited
.
Sol. Energy
,
68
,
417
425
,
doi:10.1016/S0038-092X(00)00004-9
.
Svensmark
,
H.
,
1998
:
Influence of cosmic rays and Earth's climate
.
Phys. Rev. Lett.
,
81
,
5027
5030
,
doi:10.1103/PhysRevLett.81.5027
.
Torrence
,
C.
, and
G. P.
Compo
,
1998
:
A practical guide to wavelet analysis
.
Bull. Amer. Meteor. Soc.
,
79
,
61
78
,
doi:10.1175/1520-0477(1998)0792.0.CO;2
.
Wang
,
K.
,
R. E.
Dickinson
,
M.
Wild
, and
S.
Liang
,
2012
:
Atmospheric impacts on climatic variability of surface incident solar radiation
.
Atmos. Chem. Phys. Discuss.
,
12
,
9581
9592
,
doi:10.5194/acp-12-9581-2012
.
Wild
,
M.
, and
Coauthors
,
2005
:
From dimming to brightening: Decadal changes in solar radiation at Earth's surface
.
Science
,
308
,
847
850
,
doi:10.1126/science.1103215
.
Wulfmeyer
,
V.
, and
I.
Henning-Müller
,
2006
:
The climate station of the University of Hohenheim: Analyses of air temperature and precipitation time series since 1878
.
Int. J. Climatol.
,
26
,
113
138
,
doi:10.1002/joc.1240
.
Zhou
,
S.
,
A. J.
Miller
,
J.
Wang
, and
J. K.
Angell
,
2001
:
Trends of NAO and AO and their associations with stratospheric processes
.
Geophys. Res. Lett.
,
28
,
4107
4110
,
doi:10.1029/2001GL013660
.
Zielinski
,
G. A.
,
2000
:
Use of paleo-records in determining variability within the volcanism–climate system
.
Quat. Sci. Rev.
,
19
,
417
438
,
doi:10.1016/S0277-3791(99)00073-6
.