3.2. SOLAR AND THERMAL ATMOSPHERIC RADIATION
3.2.1. BASELINE MONITORING ACTIVITIES
The CMDL surface radiation monitoring project began in 1973 with the intent to provide supporting information for baseline climate monitoring activities and to determine trends and variations in the surface radiation budget induced by changing atmospheric composition because of anthropogenic activity. Then, trends predicted in the measured radiation quantities due to anthropogenic sources were near or below the level of detectability for the available instrumentation. However, other sources of variability in the surface radiation budget were also not adequately known or understood; thus the measurements could contribute to the most basic understanding of the natural and changing surface radiation budget. Such contributions included definition of diurnal and annual cycles, effects of cloudiness, variation on daily to decadal time scales, effects of major volcanic eruptions, unexpectedly high concentrations of anthropogenic pollution in the Arctic, effects of constituent variations on narrowband irradiance (e.g., ozone and ultraviolet (UV) changes), and possible anthropogenic modification to cloudiness. In addition to research conducted by CMDL, the surface radiation measurements contribute to several global data bases. Global data bases are needed to evaluate the radiation and energy budget necessary to diagnose the climatic time scale general circulation of the atmosphere. Observations also contribute to satellite-based projects where surface measurements serve to verify spot estimates and to allow features of the intervening atmosphere to be deduced. A major goal of the monitoring program is to obtain a record, as long and complete as possible, of surface radiation parameters which will permit examination of the record for all scales of natural and modified variability. Of particular interest is the determination of the magnitude, representativeness, and possible consequences of any observed changes. To this end, the CMDL radiation group maintains complete and continuous surface radiation budget observations at several globally diverse sites with various ancillary supporting observations. The following describes those projects, particularly recent changes and results.
CMDL Baseline Observatories
The four main CMDL baseline observatories have been involved in the radiation project since the early 1970s. The different environments and conditions among the sites have resulted in different programs evolving at each site. The basic measurements made at all sites include both the downward global and direct components of solar radiation. By late 1995, solar diffuse radiation measurements had been added to each of the sites permitting more accurate determination of global radiation from the sum of vertical direct and diffuse. Broadband thermal infrared measurements were added in the last 10 years. At sites where the surrounding terrain is representative of a larger regional area (SPO and BRW), the upward solar and thermal infrared irradiances are also measured. The solar radiation records acquired at these sites constitute some of the longest known U.S. records of their kind acquired under research conditions. In 1994 a major upgrade to the observing network was accomplished with the conversion to a commercial data-logging system that provides 13-bit accuracy and precision, resistance and voltage measurements, and onsite processing of the data. The raw data are routinely transmitted over phone lines or Internet to the central data processing facility in Boulder where data editing, final calibrations, graphical inspection, and archiving are performed.
The basic measurements currently conducted at each of the four baseline observatories for the past 20 years include normal-direct and downward-broadband solar irradiance, downward solar irradiance in the 0.695 m to 2.8 m band, and wideband spectral direct solar irradiance. Downward broadband thermal irradiance measurements were added at BRW, MLO, and SPO in more recent years as well as upwelling irradiance measurements at SPO and BRW. The wideband spectral direct observations are obtained manually under clear sky conditions while the others are sampled at 1 Hz with 3-minute averages recorded on computer media. Preliminary data from all CMDL radiation sites are generally available on the Internet within a few days of acquisition in the radiation section on the CMDL Home Page (www.cmdl.noaa.gov).
Filter Wheel NIP
The wideband spectral direct solar irradiance measurements are made with a
filter wheel normal incidence pyrheliometer (FWNIP). The data from these observations
are compared to a higher spectral resolution radiative transfer model [Bird
and Riordan, 1986]. The model is based on Beer's law and has only one level
(surface). The aerosol optical depth and precipitable water are adjusted within
the model to obtain a best match with the FWNIP observations. This provides
a low precision, but a relatively stable estimate of mean visible aerosol optical
depth and water vapor at the four baseline observatories. The accuracy of the
method of obtaining aerosol optical depth and water vapor is limited by the
dependence on the absolute values of the extraterrestrial solar spectrum and
instrument calibration, unlike typical applications in sunphotometery. The aerosol
data record from this observational project is shown through 1995 (Figure 3.12).
Fig. 3.12. Monthly average aerosol optical depth as determined from the filter wheel NIP for the four primary observatories. Note that these derived values are not as accurate as determined by some other techniques but are inherently stable and relatively complete over the period of record as compared to other attempts to remotely sense these quantities at these stations.
MLO Apparent Transmission
The transmission of direct broadband solar irradiance through the atmosphere above MLO is monitored using a quantity known as the apparent transmission. This quantity is computed by taking the average of three ratios of direct solar irradiance where each ratio is the quotient of the irradiance at an integer air mass divided by the irradiance at the next smaller integer air mass as first defined by Ellis and Pueschel  and used by Dutton et al.  and others. The apparent transmission is stable over time because it is independent of a radiometer calibration value and also, therefore, quite sensitive to small changes in transmission that can be due to aerosols, ozone, or water vapor. Previous studies [Bodhaine et al., 1981; Dutton et al., 1985] have shown that in monthly averages, aerosols tend to dominate observed changes in the apparent transmission such that the major observed excursions in the record given in Figure 3.13, are because of aerosols. The major observable features in Figure 3.13 are the effects of several volcanoes, particularly Agung in 1963, El Chichón in 1982, Mt. Pinatubo in 1991, and an annual oscillation caused primarily by the springtime transport of Asian aerosol over the site [Bodhaine et al., 1981]. Figure 3.13 is complete through 1995 and most recently shows that the recovery from the eruption of Mt. Pinatubo was not yet complete in 1995. The fact that the MLO apparent transmission record still indicates a Mt. Pinatubo residual is evidence of the sensitivity of the measurement since it is known from other measurements by CMDL and others, that the optical depth of Mt. Pinatubo in 1995 was very low (about 0.005 at 500 nm).
Fig. 3.13. Monthly average apparent solar transmission above Mauna Loa, Hawaii. The effects of major volcanic eruptions and the annual transport of Asian aerosol is most evident.
Boulder Atmospheric Observatory (BAO) Tower
Observations of upwelling and downwelling solar and thermal irradiances at the top of the 300-m BAO tower, located near Erie, Colorado, began in 1985. Nearly continuous observations of these quantities, hourly resolution until 1992 and 3-minute thereafter, have been maintained since 1985. The data provide a unique view of surrounding agricultural land in that the data are more representative than typical surface-based solar radiation budget observations. The data from the site were used in several recent publications [Nemesure et al., 1994; Cess et al., 1995; Dutton and Cox, 1995; Garrett and Prata, 1996; and several earlier papers]. Since 1990, observations of direct solar and downwelling solar irradiances have been made near the base of the tower. This site has contributed data to the World Climate Research Program (WCRP) Baseline Surface Radiation Network (BSRN).
Observations of direct solar, downwelling solar, and thermal IR irradiance began in Kwajalein in 1989. Kwajalein is a small, <4 km2, island in the tropical Pacific. Data obtained at this location are virtually free of any effects of the island and, therefore, are often taken as representative of the open ocean in that region. Data from Kwajalein were used in several recent publications including Dutton , Whitlock et al. , and Bishop et al. . Substantial upgrades to the Kwajalein radiation measurement array are planned for 1996 including spectral direct and diffuse, broadband diffuse, disk-shaded pyrgeometer, UV-B, Photo-synthetically Active Radiation (PAR), and improved solar tracking capability. Data from Kwajalein have been submitted to the BSRN data archive.
Observations of downwelling solar and thermal IR began on the east end of Bermuda in 1990 on the National Aeronautics and Space Administration (NASA) tracking station site. The rather small size and elongated shape of the island in the lower midlatitude westerlies is believed to have a minimal influence on the irradiance measurements, although some clouds of orographic origin are known to exist there in the summer months under certain synoptic meteorological conditions. Data from Bermuda were submitted to the BSRN data archive and have been used by Whitlock et al.  and Bishop et al.  in satellite comparison and validation studies.
3.2.2. SOLAR RADIATION CALIBRATION FACILITY
Calibration support for the four CMDL baseline observatories and the BSRN sites at Kwajalein, Bermuda, and BAO during 1994 and 1995 was carried out by the CMDL Solar Radiation Calibration Facility (SRCF). Calibrations and characterizations of pyranometers and pyrheliometers were performed as needed, and field exchanges of recalibrated instruments were completed. Improved diffuse-sky measurements were implemented at the SRCF with the addition of ventilated tracking disk systems designed for the Eppley automated solar trackers used by CMDL. The improved diffuse measurements, together with automated cavity operation for the collection of solar direct beam data, have enabled more accurate characterizations and calibration procedures requiring accurate determination of solar components (direct beam and diffuse sky).
The CMDL reference cavity radiometers were compared with reference cavities from other organizations during 1994 and 1995. A cavity intercomparison was held at the National Renewable Energy Laboratory in Golden, Colorado, October 810, 1994. In 1995 the CMDL references were taken to the World Radiation Center in Davos, Switzerland, for participation in the WMO-sponsored eighth International Pyrheliometer Comparison (IPCVIII). These comparisons are typically conducted every 5 years and allow reference instruments from all of the WMO regions to document their performance relative to a standard group of instruments maintained at the World Radiation Center. Seventy-seven reference instruments from 37 countries participated in IPCVIII from September 25 to October 13, 1995. Participation in IPCVIII of the CMDL reference cavities (TMI67502 and AHF28553) maintains the historical traceability of the NOAA standards to the World Radiometric Reference maintained in Davos and the World Radiation Center. All solar radiation measurements made by CMDL are thus traceable to the world reference.
Instrument Development Activities
Efforts began during 1995 to add observational capability to the BSRN sites at Kwajalein, Bermuda, and BAO. Dual ventilated shade disk systems were acquired, tested, and deployed to BAO during 1995 with installation at Bermuda and Kwajalein scheduled for early 1996. The dual shade disk systems attach to the Eppley solar trackers and enable a pyranometer and pyrgeometer to be continuously shaded and ventilated. Improvements in solar tracking accuracy for these sites was also achieved by implementing a more accurate solar position algorithm in the tracker control program, precision leveling of the solar tracker during installation and setup, and a solar position detector designed and built by CMDL was added to the solar tracker. Tracking accuracies of better than 0.1 degrees are achievable with these improvements. These improvements are scheduled for installation at the Kwajalein, Bermuda, and BAO sites in 1996. In addition, software was added to the tracker control computer that allows remote access to the tracker control program via modem. This capability, together with the solar position detector data, will allow monitoring of tracker performance at the remote sites and tracker control from Boulder if necessary. Installation of an automated self-calibrating cavity radiometer system in the refurbished MLO solar dome was also completed during the latter part of 1995. A new Eppley automated cavity system was purchased for this application with the goal of incorporating its operation and control in the dome control computer system. When this is completed, continuous direct-solar-beam data will be available from MLO in addition to the NIP data that have been collected since 1958.
In addition to the routine CMDL monitoring support for the four baseline observatories. The SRCF provided support, resources, training, and logistics assistance in other areas such as the World Meteorological Organization/Global Environmental Fund/Global Atmospheric Watch (WMO/GEF/GAW) baseline station network.
3.2.3 AEROSOL OPTICAL DEPTH REMOTE SENSING
Remote sensing of aerosol optical depth is carried out in several projects in CMDL. Derivation of low precision aerosol optical depth from wideband filtered pyrheliometer observations is described in a previous section. Traditional narrow-band sunphotometery measurements are currently made routinely at MLO and Sable Island, Nova Scotia. The CMDL radiation group also maintains a few calibrated handheld sunphotometers of an older but reliable design for use in various short-term field programs. Such field programs have recently included, ACE-1, TOGA CORE, Antarctic dry valley studies, Arctic aircraft flights, and visits to BSRN sites. CMDL will begin to deploy commercial versions of the Multi-Filter Rotating Shadowband Radiometer (MFRSR) from which not only can spectral optical depth be derived but also spectral diffuse and total spectral irradiance fields. As part of a world-wide aerosol optical depth network, the GAW and BSRN programs are awaiting delivery of several multi-channel sunphotometers from the WRC.
The automated solar observatory at MLO, which houses the primary CMDL sunphotometer, was upgraded during 1995. The antiquated computers used for dome control and data acquisition were replaced with a single 486-PC. The dome operates, as before, opening and closing each day with both the internal spar and dome tracking the sun while constantly monitoring precipitation and wind speed to determine shutdown conditions. Instruments on the spar include the CMDL PMOD01 sunphotometer, two dual-channel water vapor meters, an active cavity radiometer, and a backup pyrheliometer. Aerosol optical depth data are obtained at three wavelengths: 380, 500, and 778 nm. One value per day per wavelength is derived from the Langley plot technique. The most recent summary of data from this project was given by Dutton et al. .
3.2.4. MAUNA LOA UV SPECTRORADIOMETER
A researchgrade UV spectroradiometer was installed at MLO in July 1995. Because Mauna Loa (mountain) extends above the marine boundary layer, and because of the diurnal upslopedownslope wind circulation, mornings at MLO often exhibit unusually clear skies, providing an excellent site for solar radiation measurements. The instrument described here was developed and operated by the National Institute for Water and Atmosphere at Lauder, New Zealand [McKenzie et al., 1991, 1992].
The solar radiation measured at the earth's surface depends on the transmission of the atmosphere, the earthsun distance, and the irradiance of the sun. The atmospheric transmission in the UV portion of the spectrum is controlled primarily by total ozone, and, since ozone is affected by anthropogenic influences, solar UV irradiance arriving at the earth's surface is controlled by both natural and anthropogenic effects. Ozone concentration, in turn, is also affected by changes in solar UV. The UVA region of the spectrum (320400 nm) is virtually unaffected by ozone absorption; the UVB (280320 nm) is strongly affected by variations in ozone; and the UVC (<280 nm) is almost entirely absorbed before it reaches the surface. An excellent review of this subject was given by Stamnes .
The data presented here are the first spectroradiometer measurements at MLO. Because of the long Dobson spectrophotometer ozone measurement record at MLO (1957present), a unique opportunity now exists to obtain wellcalibrated UV spectroradiometer measurements and to compare them with the ozone measurements. Past studies show that shortterm variations of UVB irradiance are inversely correlated with variations in total ozone [McKenzie et al., 1991; Hofmann et al., 1996].
A description of the MLO site and instrumentation, and the first 3 months of data are presented. Although this time period was insufficient to observe longterm trends, it is expected that sufficient variation will occur to observe the inverse relation between UV and ozone.
The UV spectroradiometer, built around a commercially available JobinYvon DH10 double monochromator, is interfaced with a computer to provide automatic control and data acquisition [McKenzie et al., 1992]. A 17mm diameter custommade Teflon diffuser, designed to minimize cosine error, is mounted as a horizontal incidence receptor and views the whole sky. A shadow disk may be added in order to separate the diffuse and direct radiative components.
The spectral range of the instrument is 290450 nm, and the bandpass is about 1 nm. The gratings are driven by a stepper motor under computer control, and a complete scan requires about 200 seconds. The irradiance signal is sampled every 0.2 nm using a photomultiplier as a detector. The instrument is mounted in a weatherproof insulated enclosure (painted white) located on a concrete pad at the MLO site. The interior of the enclosure is temperature controlled using a Peltier heater/cooler unit. The computer control and data logging system are located in a small building near the instrument.
Calibration of the spectroradiometer is performed onsite using a standard 1000W FEL quartzhalogen lamp with calibration traceable to the National Institute of Standards. Calibrations are performed at approximately 6month intervals using a precision optical bench. A stability test using a 45W lamp and a wavelength check using a mercury lamp are performed weekly. The expected longterm accuracy of the spectroradiometer system is expected to be better than ±5%. A detailed error analysis for this instrument was given by McKenzie et al. .
The spectroradiometer is programmed to begin measurements at dawn and perform scans at 5° solar zenith angle intervals throughout the day beginning and ending at 95°, except that during the middle of the day the system switches to a scan every 15 minutes. In addition, a scan is performed each midnight to give "dark" values.
A typical clear sky scan is shown in Figure 3.14. The solid curve gives total (direct + diffuse) irradiance for July 16, 1995, at a solar zenith angle of 45°. The longdashed line shows the effective action spectrum accepted for calculating the erythema spectrum used to estimate the effect of UV radiation on human skin [McKinlay and Diffey, 1987]. Note that the effective action spectrum is a dimensionless quantity normalized to 1 for l < 298 nm. The shortdashed line in Figure 3.14 shows the erythemal spectrum for that scan, obtained by multiplying the total irradiance by the effective action spectrum. The erythema can then be calculated by integrating over the erythemal spectrum. In this example the erythema is 17.5 mW cm2. This is the quantity commonly measured by broadband instruments designed to monitor erythema. At smaller solar zenith angle, the irradiances can be much higher, and erythemal irradiances in excess of 45m W cm2 have been measured at MLO.
Fig. 3.14. Example of a Mauna Loa UV spectrum obtained on a clear morning (July 16, 1995) at a solar zenith angle of 45° (solid line). A point is plotted every 0.2 nm. Also shown is the erythema spectrum (shortdashed) for this scan obtained by weighting the UV spectrum by the effective action spectrum (EAS) (longdashed). The EAS is a dimensionless quantity normalized to 1 for l < 298 nm. The erythema for this scan is 17.5 W cm2.
For the following analyses, UV spectroradiometer data for 45° solar zenith angles were chosen for clear mornings at MLO. Clear mornings at MLO were determined by examining other solar radiation records for MLO. For each scan, 1nm averages were formed centered at each 1nm wavelength. In order to quantify changes in UV related to changes in ozone, and to display data as a time series, all spectral irradiance data were adjusted for the eccentricity of the earth's orbit around the sun.
All Dobson spectrophotometer total ozone data were taken directly from the MLO observer notes. Retrieved ozone values were obtained from the UV spectroradiometer data using the method of Stamnes et al. , which uses the irradiance ratio I340/I305 to infer total ozone. Erythemal radiation data were obtained from the spectroradiometer data by applying the effective action spectrum weighting function and integrating over wavelength as discussed previously.
The radiative amplification factor (RAF) is defined as the percent change of UV irradiance divided by the percent change of total ozone, a quantity that was introduced to estimate the effects of ozone depletion on the incident UV radiation. In this work, RAFs were calculated using the power law formulation from Madronich : RAF = -deltaln(I)/deltaln(O3), where I is UV irradiance.
Referring again to Figure 3.14, it is seen that irradiance decreases by 5 orders of magnitude over the wavelength range 290320 nm. All of the variability seen in the data is real and some of it is due to solar structure, such as the obvious calcium lines between 390 and 400 nm. As discussed previously, the 290320 nm range is most strongly influenced by atmospheric ozone.
Figure 3.15 shows a time series over the period DOY 192253, 1995, of 1nm
means of UV irradiance data for a solar zenith angle of 45°over
the 295 to 320nm wavelength band (5nm intervals). Only data for
clear days are shown, giving 27 data points; however, the individual data points
are connected by straight lines for continuity. Although this 3month time
series is not long enough to show a significant trend, significant variations
in both UV irradiance and total ozone occurred. During this time period, stability
tests showed that the instrument was operating well within the expected limits
of calibration uncertainty. The error bars show estimated 2-s
errors that include calibration, noise, and wavelength errors calculated similar
to that shown by McKenzie . The calculated erythema radiation correlates
strongly with irradiance at the shorter wavelengths as expected. Total ozone
values retrieved from the spectroradiometer data correlate well with Dobson
total ozone but show a systematic difference of about 4 Dobson Units. However,
this is better than 2% agreement and could be improved by optimizing the retrieval
algorithm for MLO. RAFs, shown as a function of wavelength in Figure 3.16, are
negligible for wavelengths longer than about 325 nm and increase for shorter
wavelengths. This erythema RAF of about 1.4 for MLO is larger than the values
of 1.11.2 reported for other locations [McKenzie et al., 1991] but
is not significantly different because of the large error bars.
Fig. 3.15. Onenm averages of spectral irradiance on 27 clear sky mornings
at MLO for selected wavelengths at a solar zenith angle of 45° (lower),
corresponding erythema calculated using the EAS (middle), and total column Dobson
ozone compared with ozone retrieved from the spectroradiometer data (top). The
ozone retrieval uses I340/I305 as described by Stamnes
et al. . Error bars shown are 2-s estimates
including calibration, noise, and wavelength errors. Note that DOY 190 = July
Fig. 3.16. Radiative amplification factor (RAF) as a function of wavelength. Note that the erythema RAF is 1.37, equivalent to a wavelength of about 308 nm.
Based on the previous discussion of the MLO UV spectroradiometer program, the following can be concluded: (a) The UV spectroradiometer has operated properly at MLO and is producing excellent data within expected calibration limits. The CMDL program plans to continue these measurements as a longterm project in an effort to detect any possible longterm UV spectral trends and to relate these to ozone trends. (b) UV irradiance variations are strongly correlated (inversely) with Dobson total ozone variations, with the highest correlation coefficients at the shortest wavelengths. Erythema calculated from the spectroradiometer is also strongly correlated with ozone. (c) The RAFs of about 1.4 measured at MLO are higher than those previously measured at other sites but may not be significant because of the large error bars. (d) In this limited data set, no significant UV irradiance trend is evident.
3.2.5. MLO BROADBAND UV
A UV broadband horizontal incidence instrument (Yankee UVB1, SN 950208) was installed at MLO on July 7, 1995. This instrument was interfaced with the station solar radiation data acquisition system to provide 3minute mean data. The UVB1 has a spectral response over the wavelength range 280330 nm and uses a fluorescent phosphor to convert UV light to visible light, which is then detected by a solid-state photodiode. All optical components are thermally stabilized at 45°>/FONT>C using a thermostatically controlled heater. The UVB1 will undergo annual factory calibrations, and its calibration will be checked by comparison with the MLO spectro-radiometer that commenced measurements at the site at about the same time. The performance of the broadband UV radiometer relative to the spectral measurements will be used to assess the information content of broadband UV measurements at other CMDL sites.
CMDL has established an active role in the management of the WCRP BSRN. In addition to supplying data from five CMDL sites to the BSRN archive, CMDL provides the international manager for the program. BSRN is intended to acquire and supply surface radiation data of superior quality for global energy budget and satellite studies. Several instrumentation upgrades are still required at the CMDL observatories to fully comply with BSRN specifications. Recent improvements to the CMDL radiation sites that move in the direction of more complete BSRN compliance are new data loggers and tracking shadow disks for pyranometers and pyrgeometers. In addition, observations of wideband UV-B and PAR as well as spectral diffuse/total irradiance were added at some sites and will be added to more as funding and manpower allow. Considerable effort is put into data processing and analysis for the purpose of passing final data on to the BSRN archive.
3.2.7. WMO GAW STATIONS
The CMDL radiation project participated in an effort to establish surface solar radiation programs at the GAW observatories. This effort involved the development of solar radiation monitoring systems, calibration capability, and personnel training for five sites. By the end of 1995, four of the five GAW sites had operational solar radiation monitoring programs. The four operational sites are located in Algeria, China, Tierra del Fuego, and Indonesia. A fifth site in Brazil is under preparation. Each site is equipped with pyrheliometers and pyranometers to monitor direct solar beam, global horizontal and diffuse-sky radiation, plus an automated cavity radiometer system for calibration. All sensors for the GAW sites were characterized and compared to CMDL standards prior to deployment to each site and site calibrations are performed using an automated cavity radiometer that enables the sites to maintain traceability to the absolute scale and the world radiometric reference. Instrumentation for the site was purchased with funds through the World Bank, Global Environmental Fund. Personnel from each site were trained in Boulder for a period of 1 month and some assistance from Boulder was provided in establishing some of the sites. Three of the four sites have been visited by CMDL personnel (China, Tierra del Fuego, and Indonesia); future collaboration between CMDL and these new monitoring sites is anticipated. Data from the sites is under the control of the individual site scientists and are to be sent to GAW archives and to Boulder for inspection and brief analysis.
3.2.8. VOLCANIC RADIATIVE FORCING AND INDUCED GLOBAL COOLING
The zonal mean global radiative forcing due to the eruptions of El Chichón (1982) and Mt. Pinatubo (1991) was computed based on near global coverage aerosol optical depth estimates made from satellite, [Dutton and Cox, 1995]. These events provide two case studies of the viability of our global observational network and our ability to assess the impact of a major radiation budget altering event. Aerosol optical properties were derived from Mie inverted aerosol size distributions based on surface measured spectral aerosol optical depth. Comparisons between optical properties derived from Mie inversions and those using in situ measured size distributions show little difference between the two in computed volcanic radiative forcing. The computed global zonal mean radiative forcing was used in a simple global thermal mass model to estimate the hemispheric tropospheric cooling, with close agreement to Microwave Sounding Unit (MSU) temperature observations following Mt. Pinatubo, but with poor agreement after El Chichón. The anomalous sea-surface temperature conditions of 1982-1983 are most likely responsible for the thermal model's failure to track observed temperatures. Previously, Dutton and Christy  suggested that the observed volcanic aerosol and radiative forcing following El Chichón and Mt. Pinatubo might be responsible for observed (MSU) and predicted [Hansen et al., 1992] global cooling following these two major eruptions.
3.2.9. BRW SURFACE RADIATION AND METEOROLOGICAL MEASUREMENTS
Measured surface radiation budget components for BRW have been compiled for 1994. From hourly-averaged data, daily and monthly means were produced and merged with ancillary meteorological data for each year, and monthly statistics were computed. These data were published by Stone et al.  which also contains a description of CMDL's monitoring program at BRW and the data processing techniques used. Tables of daily values, monthly statistical summaries, and corresponding plots show annual cycles of several measured and derived radiation variables collated with meteorological data. These data include the four components that constitute the net surface radiation balance, i.e., the upward and downward solar (or shortwave) and the upward and downward thermal infrared (or longwave) irradiances and also direct-beam solar irradiance and surface albedo, derived from the ratio of the reflected to incident radiation. Figure 3.17 is a sample of the data for 1994 showing daily mean time series of the net surface radiation components, the direct beam irradiance, and derived quantities. Daily average total-sky cover, averaged from three hourly National Weather Service (NWS) observations made in Barrow, is also included. Coincident meteorological data (not shown) are displayed similarly in the report to facilitate correlative analyses. In addition to the printed report, the 1992, 1993, and 1994 daily data are accessible digitally through the Internet via connection to the CMDL World Wide Web home page.
Fig. 3.17. Daily average surface irradiance (W m-2) observations and sky cover (cloudiness in tenths of total sky), for Barrow, Alaska, 1994.
Radiation measurements at BRW show a dramatic increase in net irradiance each spring in late May or early June coinciding with the maximum average daily solar gain at the surface. Snow melt typically occurs during the second week in June [Dutton and Endres, 1991] evidenced by a dramatic decrease in surface albedo (Figure 3.17). Monthly mean net radiation generally peaks in July, which tends to be the least cloudy of summer months. The downwelling thermal irradiance reaches a maximum, on average, during August, which is typically the cloudiest summer month and often the warmest. The data acquisition section (1.5) of this Summary Report gives a description of the BRW meteorology program as well as climate summaries for 1994 and 1995.
Year-to-year variations in the net radiation balance at BRW are found to be greatest during the winter months when the longwave components dominate and day-to-day values correlate well with transient weather events. Increased cloudiness, relatively warm temperatures, and westerly winds weaken the surface-based temperature inversion and warm the surface. In fact, clouds tend to radiatively warm the surface most of the year when it is snow covered. Clear periods during winter are coldest and are usually associated with (northeasterly) outflow of air from a quasi-persistent polar anticyclone and relatively calm winds resulting in strong surface-based temperature inversions [Kahl, 1990].
Understanding the links between conditions at BRW and the central Arctic, such as ice distributions in the Beaufort and Chukchi Seas and/or the frequency of cyclones in the central Arctic [e.g., Serreze et al., 1995; Maslanik et al., 1996] is the focus of ongoing research. For instance, an inspection of the 1992-1994 sky-cover record compared with Kahl's  analysis suggests that spring cloudiness has increased significantly in recent years. Because clouds can dramatically influence radiative flux at the surface at this time of year, the net radiation balance may also be affected. A comparison of the 1992-1994 BRW radiation measurements with an earlier record [Maykut and Church, 1973] suggests that this has occurred [Stone et al., 1996]. In turn, these changes may be associated with decreasing sea ice concentrations upwind of Barrow [Maslanik et al., 1996] and/or regionally changing circulation patterns that affect the Arctic hydrologic cycle [e.g., Serreze, et al., 1995]. Only through continuous monitoring of polar processes and analyses of correlative data sets will we begin to understand the complicated feedback mechanisms that determine polar climates; these in turn affect global circulation patterns [Fletcher, 1970].
BRW is strategically situated to investigate Arctic climate interactions because
it is sensitive to processes that occur throughout the region. Useful discussions
on how synoptic-scale systems influence the Barrow climate are given in Halter
and Peterson , Halter and Harris , and Harris and
Kahl . CMDL will continue its monitoring efforts as part of the BSRN
[Wielicki et al., 1995]. In addition, the U.S. Department of Energy (DOE)
is constructing a Cloud and Radiation Testbed (CART) facility nearby as part
of their Atmospheric Radiation Measurement (ARM) Program [Wielicki et al.,
1995; Stokes and Schwartz, 1994], and an ambitious field experiment
to investigate the Surface Heat Budget of the Arctic (SHEBA) is being organized
to take place in the Beaufort Sea. The addition of DOE/ARM remote sensing and
other sophisticated ground-based instruments in the vicinity of BRW will greatly
enhance our ability to assess the unique radiative properties of the Arctic
atmosphere and thus improve parameterizations needed for model studies. In addition,
through comparative analyses of the combined BRW, DOE/ARM, and SHEBA data sets,
critical aspects of Arctic climate will be further investigated.