• Barry, R. G., 1992: Mountain Weather and Climate. 2nd ed. Routledge, 409 pp.

  • Bird, R. E., , and R. L. Hulstrom, 1981: Simplified clear sky model for direct and diffuse insolation on horizontal surfaces. Tech. Rep. SERI/TR-642-761, 38 pp. [Available online at http://rredc.nrel.gov/solar/pubs/PDFs/TR-642-761.pdf.].

  • Blumthaler, M., , W. Ambach, , and M. Salzgeber, 1994: Effects of cloudiness on global and diffuse UV irradiance in a high mountain area. Theor. Appl. Climatol., 50 , 2330.

    • Search Google Scholar
    • Export Citation
  • Calbó, J., , D. Pagès, , and J-A. González, 2005: Empirical studies of cloud effects on UV radiation: A review. Rev. Geophys., 43 .RG2002, doi:10.1029/2004RG000155.

    • Search Google Scholar
    • Export Citation
  • Cede, A., , M. Blumthaler, , E. Luccini, , R. D. Piacentini, , and L. Nuñez, 2002: Effects of clouds on erythemal and total irradiance as derived from data of the Argentine Network. Geophys. Res. Lett., 29 .2223, doi:10.1029/2002GL015708.

    • Search Google Scholar
    • Export Citation
  • Conover, J. H., 1965: Cloud and terrestrial albedo determinations from TIROS satellite pictures. J. Appl. Meteor., 4 , 378386.

  • Emck, P., 2007: A climatology of south Ecuador. Ph.D. dissertation, University of Nürnberg-Erlangen, 272 pp. [Available online at http://www.opus.ub.uni-erlangen.de/opus/frontdoor.php?source_opus=656.].

  • Estupiñan, J. G., , S. Raman, , G. H. Grescenti, , J. J. Streicher, , and W. F. Barnard, 1996: Effects of clouds and haze on UV-B radiation. J. Geophys. Res., 101 , 1680716816.

    • Search Google Scholar
    • Export Citation
  • Foyo-Moreno, I., , I. Alados, , F. J. Olmo, , and L. Aladosaarboledas, 2003: The influence of cloudiness on UV global irradiance (295–385 nm). Agric. For. Meteor., 120 , 101111.

    • Search Google Scholar
    • Export Citation
  • McFarlane, S. A., , and K. F. Evans, 2004: Clouds and shortwave fluxes at Nauru. Part II: Shortwave flux closure. J. Atmos. Sci., 61 , 26022615.

    • Search Google Scholar
    • Export Citation
  • Parisi, A. V., , and N. Downs, 2004: Variation of the enhanced biologically damaging solar UV due to clouds. Photochem. Photobiol. Sci., 3 , 643647.

    • Search Google Scholar
    • Export Citation
  • Petersen, E., 1982: Solstråling og dagslys—målt og beregnet (Solar radiation and daylight—Measured and calculated). Tech. Rep. 34, Lysteknisk Laboratorium, Lyngby, Denmark, 224 pp.

  • Pfister, G., , R. L. Mckenzie, , J. B. Liley, , and A. Thomas, 2003: Cloud coverage based on all-sky imaging and its impact on surface solar irradiance. J. Appl. Meteor., 42 , 14211434.

    • Search Google Scholar
    • Export Citation
  • Sabburg, J., , and J. Wong, 2000: The effect of clouds on enhancing UVB irradiance at the earth’s surface: A one year study. Geophys. Res. Lett., 27 , 33373340.

    • Search Google Scholar
    • Export Citation
  • Schade, N. H., , A. Macke, , H. Sandmann, , and C. Stick, 2007: Enhanced solar global irradiance during cloudy sky conditions. Meteor. Z., 16 , 295304.

    • Search Google Scholar
    • Export Citation
  • Schmidt, D., 1999: Das Extremklima der nordchilenischen Hochatacama unter besonderer Berücksichtigung der Höhengradienten (The extreme climate of the north Chilean High-Atacama with special focus on altitudinal gradients). Dresdener Geographische Beiträge 4, Universität Dresden, 100 pp.

  • Wen, G., , R. F. Cahalan, , T-S. Tsay, , and L. Oreopoulos, 2001: Impact of cumulus cloud spacing on Landsat atmospheric correction and aerosol retrieval. J. Geophys. Res., 106 , 1212912138.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Study area in the southern Ecuadorian Andes. The major ridge, the Cordillera Real, is also the major climate divide, parting the climate into very moist regimes (to the east) and semiarid regimes (to the west).

  • View in gallery

    The average light/cloud conditions at the different sites (order of sites: from left to right and up to down corresponds to from west to east; cf. map in Fig. 1) during the most frequent weather condition (easterly trades; >70% yr−1). Note the 10%–15% chance of 10% enhanced irradiance at the Cajanuma station around noon. Statistics are based on 1-h averages. The source of the clear-sky values is the Bird and Hulstrom (1981) model.

  • View in gallery

    (a) Clear-sky irradiance appropriate for calibration of the model during morning hours. (b) Good compliance of empirical clear-sky global irradiance values with model clear-sky global irradiance.

  • View in gallery

    Selected daily courses of irradiance at the Páramo weather station (3400 m MSL) during different cloud cover conditions. Values indicate the respective day maximum of the variable minimum, mean, and maximum. (a) Clear-sky irradiance. (b) Clouds produce considerably reduced and enhanced G around noon. (c), (d) Peak values of the dataset. (e) Irradiance during a day with heavy overcast and light rain.

  • View in gallery

    Diurnal distribution of instantaneous values of absolute maxima Gmax from time intervals of 1 and 2 h (n > 8000). The thicker, solid curve is the upper-envelope curve of the model clear-sky values Gpot of a year.

  • View in gallery

    Monthly maximum values of G+ tend to be minimal and maximal when maximum Gpot is minimal and maximal.

  • View in gallery

    (a), (b) Enhancement of global irradiance is independent of SE. In addition, the mean amplification factor is the same at all sites, regardless of the altitudes or prevailing humidity (transmissivity) conditions. (c) Averages of all observation sites are summarized. Mean enhancement of current potential radiation is 119.6%; maximum excess occasionally reaches 170%. The general trend of decreasing absolute maxima beyond an SE of 75° must be credited to a decreasing number of observations, not to decreasing excessive irradiance itself. (d) The reason for a decreasing number of observations is that SEs of over 73° are rarest.

  • View in gallery

    Transparent and opaque clouds near the sun produced 1552 W m−2 on one witnessed occasion. During the observation, the values did not drop below 1534 W m−2. The picture was taken from below the pyranometer situated in the center and obscuring the sun. (Model clear-sky values are all horizontal.)

  • View in gallery

    With decreasing cloudiness, mean minimum and maximum minimum irradiance increase until a clear sky is established. Beyond that point, mean minimum and especially maximum minimum irradiance decrease again. At the Páramo station, this effect is more abrupt because the location (3400 m MSL) reaches nearer to or into the clouds. The decrease of maximum minimum irradiance indicates that an increasing cloud cover accompanies increasing enhanced irradiance.

  • View in gallery

    The intersection of the regressions marks the empirical maximum extreme irradiance at the respective weather station. The regressions are retrieved with pairs of maximum measured vs maximum potential (clear sky) irradiance (upper regressions) and with pairs of measured maximum vs. minimum potential (clear sky) irradiance (hence, they are envelope functions of all measured extremes vs. their clear-sky pendant). Stability indices of the inclined regressions are r2 = 0.99 and 1.00. The neighborhood of the two measured extreme values of >1800 W m−2 from Páramo (left; small arrows) values to the intersection shows that these values were “lucky strikes” but are concordant with the phenomenon.

  • View in gallery

    Strong dependency between extrapolated empirical extreme solar irradiance and altitude. The extinction rate for extreme irradiance is high, but the gradient is typical for a decreasing radiation flux with decreasing altitude. The local general moisture conditions (water vapor thickness) at a site seem to be of importance as well: the drier stations tend to higher extreme irradiance (positive residuals); 2000 W m−2 would be found at 4880 m MSL and higher.

  • View in gallery

    The scatter schemes show that appropriate arrangements of an infinitely distant light source (creating a spatially homogenous, parallel light field) and passive diffusers are sufficient to boost diffuse irradiance to considerable magnitudes—in theory even up to 100% of the irradiance Idir coming directly from the source. (a) Two identical extensive, plane-parallel, thin diffusers, with albedo 0.5/transparency 50%, together produce a downward diffuse light flux of 67% × Idir in layer I. If the lower diffuser is replaced by a surface with an albedo of 0.2, diffuse downward flux still amounts to 55.6% × Idir. (b) When increasing total reflectivity of the lower diffusers by increasing their number or, equivalent, by assuming a single, very dense diffuser (aggregated diffusers), total downwelling diffuse shortwave radiation may sum to values of ≥80% × Idir in layer I—(c) in theory, even up to 100%. Note that either a higher or a lower albedo (transparency) of the upper diffuser would result in less downwelling diffuse irradiance in layer I.

  • View in gallery

    Ranked relative diffuse irradiance of all weather stations. Although very rare, Idiff has reached >80% and even >90% of Idir. The midlevel weather stations Cordillera del Consuelo, El Tiro, and, in particular, Cajanuma, apparently most frequently situated between the clouds, are well represented in the top ranking while sunny Vilcabamba’s first appearance is rank 44.

  • View in gallery

    (a) Frequency of 2-h maxima of G+ from instantaneous measurements; (b) share of G+ that exceeds 1400 W m−2; (c) frequency of clear-sky conditions (based on hourly means; Gmeas = Gpot ± 5%).

  • View in gallery

    Histograms of ratios of Idiff and potential (model) Idir of values G+. Mean ratio is remarkably constant; mean ratio across all stations is 34.8% ± 2.6%. Only mid- and upper-level stations have (traces of) ratios of 80% and more.

  • View in gallery

    Control measurements with a luxmeter verified solar irradiance up to 1400 W m−2/130% of clear-sky irradiance.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 51 51 18
PDF Downloads 32 32 14

An Upper Threshold of Enhanced Global Shortwave Irradiance in the Troposphere Derived from Field Measurements in Tropical Mountains

View More View Less
  • 1 Department of Geography, Friedrich-Alexander-University, Erlangen-Nürnberg, Germany
© Get Permissions
Full access

Abstract

Extraordinarily high values of global irradiance (up to 1832 W m−2) incident upon a horizontal surface were observed during a 4-yr meteorological field campaign in the southern Ecuadorian Andes Mountains (4°S). The unexpected magnitude gave rise to a thorough revision of the instrumentation and an assessment of the radiation database. Infrastructure at the sites and software were critically examined, and the sensor and datalogger manufacturers were contacted. The observed enhanced irradiance was quantified with a simple clear-sky model for global radiation. The efforts showed that atmospheric conditions and not artifacts had produced the high values. Cloud radiative effects could be singled out as the exclusive source of the “superirradiance.” Mean (bihourly) maximum enhancement was 119.6% of the potential (clear sky) irradiance; absolute maximum enhancement occasionally reached to over 170%. Thereby, under ideal conditions, the upper threshold of global irradiance is apparently ∼200% of the potential (clear sky) direct radiation [i.e., at the point of observation, downwelling diffuse cloud radiation can (almost) equal the magnitude of its apparent “source”]. The observations were made between altitudes of 1500 and 3400 m MSL in different climates ranging from perhumid to semiarid (i.e., in very cloudy climates and in less cloudy climates). The conditions that were found to explain the magnitude of the extremely enhanced irradiance are not confined to tropical or mountainous environments only, but rather they can occur at any spot in the troposphere where clouds exist. Therefore, the findings appear to be of general validity.

Corresponding author address: Dr. Paul Emck, Department of Geography, Friedrich-Alexander-University, Erlangen-Nürnberg 72119, Germany. Email: pemck@gmx.net

Abstract

Extraordinarily high values of global irradiance (up to 1832 W m−2) incident upon a horizontal surface were observed during a 4-yr meteorological field campaign in the southern Ecuadorian Andes Mountains (4°S). The unexpected magnitude gave rise to a thorough revision of the instrumentation and an assessment of the radiation database. Infrastructure at the sites and software were critically examined, and the sensor and datalogger manufacturers were contacted. The observed enhanced irradiance was quantified with a simple clear-sky model for global radiation. The efforts showed that atmospheric conditions and not artifacts had produced the high values. Cloud radiative effects could be singled out as the exclusive source of the “superirradiance.” Mean (bihourly) maximum enhancement was 119.6% of the potential (clear sky) irradiance; absolute maximum enhancement occasionally reached to over 170%. Thereby, under ideal conditions, the upper threshold of global irradiance is apparently ∼200% of the potential (clear sky) direct radiation [i.e., at the point of observation, downwelling diffuse cloud radiation can (almost) equal the magnitude of its apparent “source”]. The observations were made between altitudes of 1500 and 3400 m MSL in different climates ranging from perhumid to semiarid (i.e., in very cloudy climates and in less cloudy climates). The conditions that were found to explain the magnitude of the extremely enhanced irradiance are not confined to tropical or mountainous environments only, but rather they can occur at any spot in the troposphere where clouds exist. Therefore, the findings appear to be of general validity.

Corresponding author address: Dr. Paul Emck, Department of Geography, Friedrich-Alexander-University, Erlangen-Nürnberg 72119, Germany. Email: pemck@gmx.net

1. Introduction

During a meteorological field campaign in the Andes Mountains of southern Ecuador in 1998–2001 (Emck 2007), very high values of incoming global shortwave solar irradiance from the spectral range of 305–2800 nm incident upon a horizontal surface (henceforth G), of more than 1700 W m−2, and twice of more than 1800 W m−2 (>700 W m−2 over the clear-sky value), appeared in the records. The routine hourly mean values were inconspicuous with regard to anomalous magnitudes, but G larger than (referred to as G+) potential/clear-sky G (Gpot) and well beyond the solar constant (1368 W m−2) frequently showed up among the bihourly sampled maximum records. The measurements had to be questioned, especially when it became clear that the observations of the highest magnitudes appeared to be unprecedented. A review of the relevant literature revealed that although (cloud) enhancement of downwelling irradiance has received increasing attention in the scientific community during the last decade, the magnitude of enhancement is not well established (Calbó et al. 2005). Indeed, to our knowledge, there are only a few works that are supportive of very large enhancements. Wen et al. (2001) presented a theoretical upper bound of cloud-enhanced downwelling solar irradiance of 182% at the earth’s surface (under idealized conditions and a single cloud layer). Applied to situations during very high solar elevations, enhancements of this order would result in values similar to the observed extreme maxima. Schade et al. (2007) recently reported empirical observations of 400 (and once 522) W m−2 over the clear-sky reference value. The conclusion of their study is that cloud radiative effects (CREs) as the cause of the enhancements have been correctly identified before (e.g., Estupiñan et al. 1996; Pfister et al. 2003) but that the potential magnitude has been underestimated so far. Schade et al. made their observations near a northern German seaside resort in June. In terms of the “low” solar elevation (SE) at this latitude, the relative enhancement is similar to the enhancement maxima in Ecuador. However, to what extent the German values have been positively influenced by reflections from the nearby large water body remains unclear. Wen et al.’s theoretical approach and Schade et al.’s work have in common that they state that cumulus clouds of low optical depth and large horizontal extension are responsible for the largest irradiance enhancements. Although enhanced irradiance values from the seaside should not be compared with those of a high mountain area without a closer look at the causes, the suspected cloud conditions can occur in the Ecuadorian Andes as well.

The highest absolute solar irradiance worldwide typically is attributed to very high, snowy mountains and clearness of the atmosphere (e.g., Schmidt 1999). Encountering record-breaking irradiance maxima in relatively low, snow-free, and predominantly humid mountains was not expected. Considered aggravating with respect to plausibility was the fact that the irradiance measurements were not accompanied by any kind of methodical sky observations. Therefore, the extraordinary peaks of G were first regarded as a challenge to the quality of the broadband sensors used and the measuring setup and consequently of the integrity of the entire radiation database. The database was submitted to a systematic validation of potential sources and plausibility of the increased values. Using a simplified clear-sky model for direct and diffuse irradiance (Idir and Idiff, respectively) to find an upper estimate of Gpot, all enhanced irradiance values (exceeding Gpot) from the extreme value database were isolated. The size of the database is 44 000 single records of maximum instant values of G (henceforth Gmax).

2. Study area, instrumentation, and methods

a. Study area

The regional climatological study that produced the exorbitant irradiance values focused on spatial variability of climate along a cross-barrier transect in the southern Ecuadorian Andes (Figs. 1a,b). In the eastern section (Catamayo–Zamora), weather and climate experience a dramatic change along a zonal distance of less than 50 km (Fig. 1b). Because of the low latitude (4°S), two zenithal sun positions occur every year, one in the beginning of March and one in the beginning of October. Between 1998 and 2001, G was measured at low (∼2000 m MSL), middle (2700–2900 m MSL), and upper (3400 m MSL) mountain levels and, with respect to the prevailing moisture and cloud conditions, very contrasting locations. Figure 1b shows the position of the measuring sites.

b. Climate

For more than 70% of the time, the Cordillera Real and easterly trade winds of remarkable constancy of direction (66° ± 9°) and speed (9 ± 3 m s−1) create a pronounced division of the local climate. To the east and at the summits of the climate divide, rainfall (≥0.1 mm h−1) is detected on from 300 (El Libano, 1970 m MSL) to 350 (Páramo, 3400 m MSL) days yr−1 in total quantities of 3000 and 5200 mm yr−1. Downwind of the climate divide, the previous orographic interception of moisture results in predominantly semiarid climates with little rainfall, occurring within 3–6 months and totaling from 300 (Catamayo, 1250 m MSL) to 1300 (Vilcabamba, 1970 m MSL) mm yr−1.

Hence, the most important potential radiation-modifying parameters, cloud cover and air humidity, are very heterogeneously distributed in the investigation area. Cloud cover is perpetual in the upper montane belt of the Cordillera Real (El Tiro, Páramo, and Cordillera del Consuelo), distant and intermittent in the valleys [Vilcabamba and Estación Cientifica San Francisco (ECSF)], and sparse in the far greater lee of the Cordillera Real at Motilón. At Motilón in the very west, clouds are scarce during the first half of the day. Mainly convective clouds have an increasing chance to obscure the sun in the afternoon (Fig. 2). At the upper stations of the climate divide Cordillera Real (El Tiro, Páramo, and Cordillera del Consuelo), permanent clouds allow, on average, just ∼20%–30% of Gpot to be incident at the surface. A distant ceiling is indicated at the near-valley bottom stations Vilcabamba and ECSF.

Median RH during daylight (0800–1800 local standard time) is maximal at the crest stations (Páramo and Cordillera del Consuelo): 99.9%, and minimal at the Vilcabamba weather station: 58%. Temperature extremes, which are potential sources of instrumental errors, are moderate throughout the research area. They range between 0°C (absolute minimum) at the upper mountain levels and 36°C (absolute maximum) at the lower stations.

The permanent orographic precipitation at the eastern flanks of the Cordillera Real produces permanently low optical depth at and downwind of the crest. The turbulent mixing in the lees transports this property into the valleys.

c. Instrumentation

Because monitoring radiation conditions in the Cordillera Real had normal priority within the scope of the climatological study, a simple, second-class (ISO 9060/World Meteorological Organization) Kipp and Zonen “CM3” pyranometer had been selected for measuring G. This radiometer has a glass dome broadband filter that allows solar radiation of 305–2800 nm incident from one-half space to heat a thermopile. The heating generates an electric signal that, through a response function, is referred to the actual G. Errors of the CM3 such as nonlinearity, directional error, zero offset, and some other temperature effects are specified to sum to 2%–5% for instant values (from high to low SE) and to 10% for daily totals if they happen to have the same direction.

The radiometers were leveled horizontally at 2 m AGL. During the deployment of 4 yr, the sensors were not recalibrated. A significant long-term trend due to a changing response to G was not detected. Data sampling was every 5 min (instantaneous values). The arithmetic mean of the 12 values of an hour [hh05–(1 + hh)00, where hh is local hour] was stored as hourly mean. The minimum and maximum values of the instantaneous measurements were stored together with their exact timestamps, in the beginning hourly and later bihourly.

d. Methods

For determining magnitude and characterizing radiation anomalies, the clear-sky reference value was determined using the Bird and Hulstrom model—a simplified clear-sky model for direct and diffuse insolation on horizontal surfaces (Bird and Hulstrom 1981). The model needs only a few atmospheric parameters for calibration but shows excellent agreement with the observed clear-sky values of all the sites at any sun position.

The parameters are ozone (0.3 cm) and water vapor thickness (0.5–3 cm), aerosol optical depth (AOD) at 500 nm (0.05–0.1), AOD at 380 nm (0.05–0.1), forward scattering of incoming radiation (0.85), and surface albedo (0.25–0.4) (the ranges used for calibrations are in parentheses). The true values of the parameters were unknown, so that they had to be guessed until a best fit was achieved between measured and calculated values under “best conditions” (clear sky and low optical depth). The transmission parameters of the model were calibrated once for each weather station with the best conditions so as to have the model always produce the lowest estimate of the irradiance enhancement at a site. The best conditions were found by using only the—as the extreme value sampled—irradiance values of very clear, cloudless days or very clear and cloudless parts of days. The criterion for clear sky was minimum and maximum irradiance of neighboring time intervals being within a small absolute range. This procedure minimizes the influence of clouds that may cause altered best conditions by obscuring the sun or by producing increased Idiff through reflection. Figure 3a shows a situation suitable for calibration. The absolute minimum and maximum G of consecutive time intervals in the morning hours indicate a perfectly stable optical state of the atmosphere. After noon, clouds interfere with clear-sky G, making the observations of that part of the day inappropriate for retrieving calibration parameters. Cloudless periods with reduced clearness of sky, and hence cloudless situations with reduced G, were discarded from the model calibrations. Such periods were rare in the area, however (see below). Figure 3b shows a sample of the model irradiance and the measurements of a clear day. The small differences between the readings at 1500–1800 LT and the model are due to slightly declining clearness of the atmosphere in the afternoon.

The SEs for each site used in the model were calculated with the high-accuracy functions of the National Oceanic and Atmospheric Administration Earth System Research Laboratory (obtained online at http://www.srrb.noaa.gov).

The modeling revealed that optimal clearness of sky is a regular phenomenon at all of the sites. In virtually all cases of tested clear-sky irradiance, the departure between observed and calculated potential radiation stayed within a range of less than ±1% for high sun positions and less than ±2% for low sun positions. Data from the sunshine-endowed Motilón weather station demonstrate the perseverance of this atmospheric state in the region. A time series of maximum irradiance values of 12 consecutive days of cloudlessness (96 measurements) in August of 2001 deviate, on average, by only 1.8% from the potential values and even from the potential radiation total by only 0.7%. Remarkable is that the departures in the given example and in general do not show any significant time dependency that would indicate deteriorating transmission conditions in the atmosphere during the day or over the period of cloudlessness. The reason for this is the indicated stable, dynamic easterly trades that perpetually replace the body of air with washed out and partially dry air from the climate divide.

A similar agreement between potential and measured G is observed at the other weather stations as well and, significant, under most weather conditions (large cloud gaps or cloudlessness provided). Solely clear-sky values measured at the weather stations ECSF and El Libano show that at low altitudes in the east there are less constant optimal optical conditions. This is due to considerable fluctuations in water vapor content of the ground layer at these locations. A sample calculation for the ECSF station shows that high vapor thickness as compared with low vapor thickness of the atmosphere can reduce clear-sky global radiation by ∼6% (8%) at high (low) SEs. Measured clear-sky radiation therefore can fall behind potential clear-sky radiation by 6% (8%) in isolated cases of very high humidity and coincident sunshine. However, most of the values of measured clear-sky radiation at the ECSF stay within a margin of 1%–2% from the potential (calculated) value, which shows that measured clear-sky radiation and potential radiation also here usually converge to a large extent.

Hence, the proposed approach seems to yield fairly exact benchmark values of maximum Gpot for any SE and sun–Earth distance. The uncertainty about the weighting of the transmissivity parameters AOD and vapor thickness (larger values of AOD compensate smaller vapor thickness and vice versa) does not affect the results with respect to the magnitude of Gpot. It has influence, however, on the calculated proportions of Idir and Idiff. For example, using larger values of AOD and a smaller vapor thickness, the model produces an increased share of Idir (for a fixed value of Gpot). Using pairwise moderately changed AOD and vapor thickness (±0.02 and ±0.05), the model returns a small uncertainty for Idir of <3% (0800–1700 LT) and <2% (0900–1600 LT).

Note that comparisons of model and measured values in this study are always “true” in the sense that measured instantaneous values are compared with model instantaneous values and empirical hourly averages are compared with averaged 5-minutely retrieved model values.

3. Results and discussion

This section starts with the demonstration of some selected daily courses of G and a brief interpretation with regard to cloud conditions. It then continues with the examination of the Gmax dataset and the substantiating of these interpretations.

Figure 4 shows measurements of irradiance of a perfectly clear day (Fig. 4a), of three days with intermittent cloud cover (Figs. 4b–d), and of one day of heavy overcast (Fig. 4e). (Figures 4a–c are based on hourly values of minimum, mean, and maximum measurements.) In Fig. 4a, the alignment of minimum, mean, and maximum values along a single curve indicate a very stable optical state of the atmosphere during the entire day. No temporary reduction or enhancement of irradiance is detected. Figure 4b shows an incident of typical irradiance enhancement. A variable, likely single, scattered (2/8) to broken (6/8) cloud layer with transparent and opaque components generated reduced and enhanced values of G around noon. In Figs. 4c,d, more complex vertical cloud structures, probably a double cloud layer, produced the record-breaking absolute values of irradiance. Figure 4e displays irradiance during heavy, permanently closed cloud cover and light precipitation. The maximum minimum irradiance of that day was 63 W m−2/6% of Gpot (1335 LT).

The examination of G+ was done with all the values of Gmax of all weather stations. The findings will exemplarily be shown with the results from the climatologically contrasting sites Páramo (predominantly cloudy, cool, wet mountain top, 3400 m MSL) and Vilcabamba [predominantly scattered (2/8) to broken (6/8) cloud cover, warm, dry, near valley bottom, 1970 m MSL].

a. Validation of radiation measurements under clear-sky conditions

Under the even radiation conditions of a clear sky, climatic influences other than radiation itself that potentially affect the measurements or their recording are minimized. Measured high irradiance and predicted values must agree. An improper processing of high radiative signals by sensor and logger at high sun positions, eventual effects through high temperatures, or the influence of nearby radio transmitters should become evident. Figure 3b shows that on cloudless days recording of high irradiance values is flawless and the equipment operates well.

b. Observed G+ under all weather conditions

If singular weather conditions or the measuring system produce erratic errors of G, then faulty positive values may have arbitrary amplitudes or appear in nighttime data. In Fig. 5 it is shown that G+ does not have arbitrary magnitudes and stays within a defined range over Gpot. (The vertical patterns in Fig. 5 have a simple metrological reason: the probability of an absolute maximum of irradiance to occur within a fixed time interval is highest toward solar noon.) The values Gmax at the cloud-infested Páramo site tend to concentrate within the value spectrum of diffuse light conditions (<500 W m−2), whereas maximum values of noon hours in Vilcabamba concentrate in super potential radiation magnitudes (between 1000 and 1500 LT, >50% of Gmax). In Fig. 5a, both absolute maxima of over 1800 W m−2 measured at the Páramo station are outliers with respect to the daytime; their relative value (∼160%), however, has frequently been equaled and surpassed at other daytimes. Positive signals did not appear in the nocturnal records of any weather station (the radiation channel of the logger could not be programmed to skip the nighttime).

Considering the radiation conditions at both selected sites, a relationship of extreme irradiance with clouds becomes conceivable. Icing of or water on the optical surfaces of the equipment, for obvious reasons interfering with radiation measurements, can be excluded. Temperatures at both stations always stay above zero, and, at least at the Vilcabamba site, it is highly improbable that water on the sensor (condensation, rain, etc.) and high insolation at noon will occur at the same time. On clear nights, a significant longwave cooling of air and sensors happens at the Páramo site, but it has never been observed to result in below-freezing air temperatures.

c. Extreme irradiance and solar energy

If extreme irradiance has a direct solar background, then there must be a causal relationship between potential supply of solar energy and magnitude of absolute maximum G+. Maximum available solar energy at 4°S latitude has a bimodal annual distribution with a major peak in March and a slightly smaller peak in October. The superposition of a minimum maximum solar elevation and a maximum sun–Earth distance during austral winter is responsible for a pronounced low of solar energy supply in June–August; the opposite effect during austral summer results in only a moderate decline in available solar energy between October and March (Fig. 6, shaded area). The seasonal fluctuations of monthly maximums of G+ in Fig. 6 (thick curve) show the expected correlation. Absolute monthly maximum G+ and maximum supply of solar irradiance coincide for the greater part. The same is true for annual minima of absolute monthly maxima and the minimum availability of solar energy. A continuous recording would likely have resulted in even better agreement between energy supply and maximum G+. Why in austral summer 2000/01 the clear trend of the previous years seems interrupted has not been investigated in detail. Prevailing anomalous cloud conditions are the most plausible cause (November 2000, e.g., had extraordinarily extended cloudless periods).

d. Extreme irradiance and solar elevation

If clouds are the source of enhanced irradiance, then this radiative effect should be verifiable at any SE. Enhanced irradiance from a low sun may appear not to be high in absolute terms but there is no apparent reason why it would not be high in relative terms. Moreover, because all components of G [Idir, Idiff (clear sky), and CREs] from any section of the sky and regardless of the distance to the point of observation (PO) are (more or less) subject to the same extinction rate, their relative proportions and thus their relative contribution to G is approximately conservative. In the real sky dome where optical properties of the atmosphere may be heterogeneously distributed, this applies only to radiation coming from the same section of the sky (e.g., near the sun); however, because this part of the sky contributes the greatest share of radiation to the total of G, the proportionality remains authoritative for the total incidence. This means that if radiation enhancement occurs passively and comes from a section of the sky close to the sun then the relative enhancement of G must be, first, the same for all SEs and, second, independent of the total transmissivity of the atmosphere over a site. If this is true, it must be verifiable through examination of the relative values of G+ and their assigned SEs. Figure 7 shows that this is possible.

In Fig. 7, the assumptions are met to the greatest extent. When the clear-sky value is exceeded, CREs amplify G on average to 119.6% (±2%) for any SE (>10°) (calculation: 100G+/Gpot). Mean maximum enhancement of G is larger than 140% and occasionally reaches over 170% of G (Cajanuma).

The highest relative extremes seem to have a bias toward higher stations (Fig. 7c). As will be shown later in greater detail, this is due to the higher probability of favorable cloud conditions at higher altitudes. To form a closed, white screen, (dense) clouds need only to envelop the space around the pyranometer within a short range and do not have to be optimally arranged over a large section of the sky dome like in the largely fogless valleys, which is only an improbable, not an impossible, scenario.

e. Observed cloud conditions responsible for extreme irradiance

The average enhancement of irradiance through clouds of about 119.6% or ∼240 W m−2 at noon does not qualify as “extreme.” The surplus has a small standard deviation for any SE and location, indicating that the cause for most excessive irradiance incidences is common and hence likely originates from the same kind of radiative effect. It is obvious that this radiative effect is simple overhead cloud reflection (meaning that white clouds are brighter than the blue sky they hide; e.g., Pfister et al. 2003).

Studies dealing with diffuse solar radiation are abundant, but the quantitative aspects of the cloud radiative surplus at the earth’s surface in tropical high mountains probably have never been closely investigated. One study performed on the tropical Pacific Ocean island of Nauru (0°, <30 m MSL) (McFarlane and Evans 2004) shows that under fair weather conditions the CRE accounts for an average of 116% of Gpot. Data of G from the National Renewable Energy Laboratory (NREL) in Golden, Colorado (39.7°N, 1829 m MSL), recorded in June and July of 2002 allow the calculation of a positive CRE of similar average magnitude of ∼117% for these months. Both figures differ from the observations in the Cordillera Real but not decisively. Small departures like these can arise from the instrumentation, from measuring methods, or from the method of determining the clear-sky reference value Gpot. If the latter is based on a higher clearness of the sky (e.g., through lower vapor thickness) than is generally present during cloudy conditions, the relative enhancement of G is still the same but appears to be smaller relative to Gpot.

Cede et al. (2002) stated in a synopsis of a very large dataset (n > 200 000) of G from sites between tropical (22°) and almost Arctic (64°) latitudes in Argentina that there is no meridional effect for cloud enhancement of solar irradiance. A further conclusion of this study points in the same direction: namely, that the mere presence of cloud cover is more important for moderate G+ than is specific cloud form. Cloud forms that cause the highest instantaneous values of G+ can vary between types of cumulus and cirrus, but coverage has to be at least 4/8–7/8. Maximum enhancement for instantaneous values in this dataset is 135%; the 98th percentile of G+ for 60-min doses is 113%.

Of the results of this study, in particular the observation that, independent of any spatial parameters, a high percentage of cloud cover seems to be the prerequisite for the largest values of enhanced solar radiation is of interest. A positive correlation between cloud cover and enhancement was known before [e.g., Wen et al. (2001), Pfister et al. (2003), and Schade et al. (2007) for shortwave radiation and Blumthaler et al. (1994), Estupiñan et al. (1996), Sabburg and Wong (2000), and Foyo-Moreno et al. (2003) for UV global radiation (to name just a few)]. However, every one of these studies focused on observations at a single (or at maximum a few) PO in the midlatitudes. Cede et al. (2002) showed this independency for an entire subcontinent that spans a wide range of latitudes, altitudes, and climates.

Our own observations as well hint at an increased presence of clouds being the mainspring of enhanced G. A very high incident of G+ (1552 W m−2 or 142% of Gpot) was witnessed and photographed in Vilcabamba (Fig. 8). The picture leaves no doubt that a higher coverage of transparent clouds next to the sun would have resulted in higher G+.

The database of absolute minimum and maximum values of G as well supports the idea of significant cloud presence being essential for high and extreme G+. When high and extreme irradiance increase with increasing cloud cover, then there must be evidence of a counterbalancing attenuation of the sun within limited time intervals of the incident. The graphs in Fig. 9 show that this assumption is right. The minimum extreme values of G that were recorded next to the maxima indicate that the average attenuation tends to be inversely proportional to the magnitude of the absolute extreme irradiance incident. As stated, the more completely that the half space is covered with reflecting media, the greater their contribution to G+ can be. As a consequence, the larger the participating cloud compounds are, the smaller is the probability of high minimum G before and after the incident. This effect causes the steeply declining slope of maximum minimum irradiance above 1200 W m−2 in Fig. 9.

f. Maximum extreme irradiance

The questions that remain are what exactly do the cloud properties have to be and how are the clouds distributed in the sky to produce extreme G+ of >150%? Can conclusions be drawn as to the upper-limit at which clouds can enhance solar irradiance? In this article, an empirical maximum value and a theoretic approach are briefly discussed.

The empirical and site-specific answer is given with an extrapolation of the dependency of measured maximum maxima of G+ from Gpot. Examples for the Páramo and Vilcabamba weather stations are presented in Fig. 10. The results are not surprising: the highest absolute values of extreme irradiance can be expected in the summit area of the Cordillera Real (1825 W m−2 at 3400 m MSL) and the lowest can be expected in the valleys to the east (≤1673 W m−2). The resulting empirical maxima seem to be dependent on prevailing site-specific optical properties of the atmosphere and prevailing cloud conditions. Altitude and empirical maximum possible irradiance correlate strongly and approximately linearly between 1900 and 3400 m MSL (Fig. 11). A technical artifact would most likely not be sensitive to this effect.

Theoretical considerations can be used to try to answer the question of how diffusers in a homogeneous, parallel light field have to be arranged in a space to produce a maximum of total irradiance at a given PO. According to the previous conclusion that near-overcast conditions (i.e., a large bright white screen) produce the highest values of G+, the starting point is an extended, thin, plane-parallel diffuser over the PO. To increase total downwelling irradiance through multiple reflection, the albedo below the horizon of the PO is lifted by assuming additional layers of diffusers (Fig. 12).

The scatter schemes in Fig. 12 confirm that already simple arrangements of a semitransparent stratiform cloud cover (e.g., altocumulus stratiformis translucidus) considerably enhance total downwelling Idiff. In the case of a double-cloud-layer situation, downwelling Idiff between layers may total to 67% × Idir. Given a small cloud gap in the upper layer through which the sun shines, resulting G+ rises to 167% × Idir or up to ∼157% of Gpot, respectively. When the lower cloud layer is replaced with a natural surface with an albedo of 0.2, downwelling Idiff still mounts to 56% × Idir and G+ potentially to ∼156% × Idir. This would correspond to up to ∼146% of Gpot. If the number of low-level cloud layers is increased (Fig. 12b) or, more realistic, is replaced by a single but extremely dense cloud layer [densest clouds (e.g., Barry 1992) and fresh snow cover (e.g., Conover 1965) can have albedos >90%], G+ reaches 190% × Idir.

It is clear that a cloud arrangement as in the scatter scheme is highly unlikely to occur in the sky. However, a small modification of it results in a potentially large variety of more realistic cloud arrangements suitable for maximally enhanced irradiance. If the upper, semitransparent layer is lowered and the lower, dense (aggregated) layer is given an increased vertical extension with the exception of a close range near the PO, total downwelling Idiff would reach the PO with nearly unchanged intensity as before. The resulting small(er)-scale pocket of clear atmosphere in dense clouds and/or fog is more likely to form than a certain cloud arrangement over the largest part of the sky. At exposed mountain ridges or over snow-covered terrain, for example, where low-level clouds and/or fog are very frequent, the described small-scale clearances may even be fairly common [though hard to detect because of limited size, instability (short lived), and relative complexity].

Because water clouds are near perfect diffusers of shortwave radiation (generating an isotropic field of radiation while absorbing very little radiant energy), the scatter schemes are reasonable abstractions of cloud–radiation interaction—in particular, if examined at short ranges. Also, atmospheric effects (aerosol, molecular scattering, and absorption) or spectral dependencies that are positively correlated with the pathlength of a ray through the air and that have been ignored so far are minimized under this restriction. Consequences are that in absolute terms attenuation of Idir is very small between entering the clearance and being incident at the PO. Because average pathlength of scattered light at close range is greatly reduced as well, the same is true for Idiff. Although arbitrarily reduced ranges are unrealistic in this context because clouds never have infinite optical depth, the ratio of Idiff × I−1dir at the PO will still be considerably higher within the cloud pockets in question than in all-sky situations of enhanced irradiance. The theoretic upper bound of G+ (Idiff + Idir → 200% × Idir) as proposed in Fig. 12c applies here also; the real upper bound will largely depend on the maximum possible optical depth that the low-level clouds and/or fog can reach.

An exploration of the database for values IdiffIdir (at SE > 20°) shows that incidents of the dimensions Idiff > 80% × Idir and even Idiff > 90% × Idir have been recorded. To see where the highest values occurred the most frequently, they were ranked (Fig. 13). The order of the sites in the resulting list reveals a significant conformity with the consequences of the diffuser model. Elevated stations at which multiple cloud layers or cloud pockets are more frequent and paths of reflected irradiance are generally shorter are overrepresented among the very high values.

Still, very high values of over 80% requiring a high conformity with the scatter schemes must be considered to be rare (0.1% of Gmax). Moderate SEs between 25° and 35° appear to favor relatively extreme diffuse irradiance (a very high value of 94% was measured at a high SE of 57°). Cajanuma is particularly often represented among the high values given its overall small representation in the database (<7%). This is due to its exceptional location in the principal clearing zone behind the climate divide where a mixture of clouds and sunshine is the norm. If the parameters needed for IdiffIdir coincide with a very high SE, maximum G can reach ∼2000 W m−2 in the entire Cordillera Real.

g. Frequency of enhanced irradiance

Because of the sampling method, convincing conclusions on true temporal frequency of G+ cannot be drawn. It should be stressed once again that the presented frequencies are the maxima of relatively large time intervals (2 h). This causes the absolute frequency of G+ to be underestimated and their mean magnitude to be overestimated.

1) Frequency of extremely enhanced irradiance

In the records, the extreme irradiance incidents (>150% of Gpot) are rare. Their frequency, however, is certainly larger than is evident from the measurements. The majority of short-lived incidents must have been missed by the recording mode (measurement every 5 min) and because the used sensor has a response time of 18 s (95%).

2) Spatial and seasonal frequency of G+

Figures 14a and b show that throughout the Cordillera Real lifting of G over Gpot is very common, even in the humid sector to the east. At El Libano (1970 m MSL), every fifth recorded Gmax is larger then Gpot. Very high excess of absolute irradiance is more regular at the midlevels between 2600 and 3000 m MSL on the crest and westward (Fig. 14b). As a reference, Fig. 14c displays the frequency of clear skies. The values are low in the Cordillera Real, indicating a very high frequency of cloudy and partially cloudy conditions.

The effect that cloud gaps (of equal size) produce longer periods of direct radiation at a high SE than at a low SE (Emck 2007) leads to the circumstance that most incidences of lifted irradiance occur at noon hours when absolute irradiance is high anyway. At all stations, ∼65% of the daily irradiance values exceeding clear-sky irradiance fall in the 6 h around noon (1000–1600 LT). Hence, potential radiative stress of the biosphere that may emanate from very high absolute irradiance is aggravated through this.

Because of the largely uniform weather, the seasonal distribution of irradiance enhancement in the investigation area is mainly determined by just the celestial mechanical variables SE and sun–Earth distance. The seasonal distribution of relative enhancement magnitude depends on SE alone, while the magnitude of absolute enhancement depends on SE and the sun–Earth distance.

3) Frequency of magnitude of cloud-enhanced irradiance

Following the results of the theoretical considerations in section 3f, which indicate that Idir is probably a more direct measure then Gpot for quantifying the CRE, the magnitude frequency of CREs is presented using the ratio Idiff × I−1dir. In Fig. 15, the distribution of this variable is given for G+ (SE > 20°) of all measuring sites. The distribution of the CREs reflects the average cloud situations at the POs and validates some previous assumptions. At locations at which overcast is dominant or clouds are scarce (Páramo, Cordillera del Consuelo, or Motilón), the maximum frequency of CREs is slightly shifted to the lower values relative to the other stations. It is obvious that too many or too few clouds spoil the frequency of optimized conditions suitable for larger positive CREs. Vilcabamba has the steepest decline toward high values, which is due to the fact that low-level clouds and enhanced multiple reflection are the least probable at this site. ECSF has a similar distribution characteristic as Vilcabamba. The site is also situated behind a mountain ridge that shields it from the main wind direction and thus frequently enjoys improved insolation conditions and a distant ceiling. This apparently results in a distinct local maximum of the values similar to that for Vilcabamba. Contrary to the Vilcabamba site, however, fog is more frequent and so are the relative frequencies of the highest CREs (≥60%). The third low PO, El Libano, experiences excessive fog and/or low-level clouds during most of the time. The frequent obscuration results in a less pronounced maximum peak but brings forward the elevated relative frequency of CREs (≥65%) relative to Vilcabamba. An outstanding PO with respect to high magnitudes of CREs is Cajanuma. The reason for this is the standing cap cloud and the lee-wave cloud. Their relative position to the PO and the sun and their shallow vertical extension that makes them transparent or appear white often coincide with direct insolation. As a consequence, even mean hourly values of G mount to 110% of Gpot (see Fig. 2). At all higher measuring sites there is evidence of extreme CREs ≥ 80%. Except for Motilón, one could speak of a small, second local maximum of relative frequency (Cajanuma, El Tiro, Páramo, Cordillera del Consuelo). This may be interpreted as a hint for the existence of a separate class of short-lived, small-scale, and vertically complex cloud configurations that was found potentially responsible for the maximum of extreme CRE.

Despite the well-defined local nuances in the CREs, Fig. 15 shows that bulk CREs are the same at all sites. This results in the strikingly equal measures of central tendency (mean and median) among the locations.

4. Technical audit

a. Zero offset

In the opinion of the manufacturer, the only effect with a sensory background that could provoke an extra signal of several hundred watts per meter squared in the CM3 pyranometer is a temporary zero offset through a rapid change of temperature. This effect is specified with a signal of <4 W m−2 at 5 K h−1. It means that if the reference temperature of the sensor decreases by 0.14 K s−1 an extra signal of maximally +400 W m−2 could be induced. The most likely constellation for such a rapid cooling under (Ecuadorian) outdoor conditions is high wind speed and a large temperature leap dT. To find the required magnitude for these exogenic parameters, the heat transfer rates from the pyranometer to a ventilated air environment were determined. The calculations took into account the specific properties of all involved materials of the pyranometer (material, mass, and surface area of outer shell) and the relevant laws of thermal transition resistance. Any influence of moisture (cooling through evaporation) was factored out because excessive irradiance is also observed under dry conditions. The results are unambiguous: either a drastic temperature leap of dT = 40 K and a reasonable 5 m s−1 wind speed or a reasonable temperature leap of dT = 2 K and a drastic wind speed of 130 m s−1 would induce a temperature gradient of the required order. Both scenarios are very unrealistic, even at the Páramo weather station.

b. Sensor sensitivity

Sensitivity range of the CM3 is 0–2000 W m−2. The sensor of the used type is constructed to measure incoming global solar radiation of the spectrum 305–2800 nm. The manufacturer’s modeling and experiments under standard conditions have shown a correct response of the CM3 to this parameter. Each pyranometer is individually tested and calibrated before it is released. The (individual) calibration certificates are delivered along with the sensors. The pyranometers of the later-equipped midlevel weather stations were from a different production lot so that an (undetected) manufacturer error is unlikely. No complaints of other customers about sensors generating unexpectedly high values are known to the manufacturer so far.

c. Logger

On inquiry, the logger manufacturer Thies revised logger and software for faults that would explain incorrect processing of electric signals or storage of data that originate from a pyranometer exposed to shortwave radiant flux densities of up to 2000 W m−2, without result. A systematic manufacturer error of one or more production lots would have manifested also with other customers. All loggers were operated with identical programs; the correctness of the programmed calibration factors for the pyranometers was validated.

d. Infrastructure

Logger and weather stations were well grounded with a copper rod in (predominantly) wet soils. No latent electrical problem or static electricity issues (logger or any weather sensor) ever manifested. The sensor bodies were thermally insulated from the metal infrastructure with a plastic plate to minimize heat conduction, despite the improbability of a thermally induced zero offset.

The evaluation of the potential errors in equipment and setup was done with the kind collaboration of the suppliers (pyranometer: Kipp and Zonen, the Netherlands; datalogger: Thies Clima, Germany).

e. Verification with an independent system

Probably the best argument for a proper functioning of the setup is the verification with an independent system. Comparative measurements in parallel were done with a luxmeter. A sensor of this type is not the ideal radiometer for double checking a broadband radiometer because the silicon photodiode has a very selective spectral response to sky light (similar to that of the human eye). However, with known photometrical equivalences (103 ± 13 lm W−1 for direct sunlight; Petersen 1982), a relationship between global irradiance and lux can be approximated. A different cosine response of both radiometer types could be ignored, because the in-parallel measurements were carried out between 1142 and 1245 LT at a high SE. The location was Vilcabamba; the maximum SE was 73° at 1204 LT. The results are shown in Fig. 16.

5. Summary and conclusions

In high mountains at low latitudes, shortwave global irradiance incident upon a horizontal surface G can reach values of over 1800 W m−2. Considerations that have led to this conclusion suggest that total downwelling diffuse irradiance Idiff can surge to >90% of the actual direct irradiance Idir as observed at the surface. This phenomenon means that G at the earth’s surface can reach as much as 2000 W m−2 and more where Idir exceeds 1050 W m−2.

The theoretical upper threshold to which G could ever become enhanced at all in an environment with solely diffuse reflectors seems to be defined by the extent to which IdiffIdir (the value Idiff converges to the value of Idir) can be realized in the atmosphere. The presumed necessary cloud conditions can potentially occur at any spot in the troposphere where clouds exist.

It has been shown that the often loosely quoted illuminated sides of towering cumulus congestus and snow and ice accumulations are not stringent prerequisites for very high, enhanced irradiance. Snow and ice do not exist because of the high temperatures, and deep convection and therefore cumulus congestus and cumulus nimbus are very rare in the investigation area (because of the influence of the southeastern Pacific high pressure system; Emck 2007). Of significance is that enhanced irradiance is observed on a daily basis. The cloud conditions that lead to the largest documented values of enhanced irradiance were not directly observed but were deduced from the radiation database itself or other in situ weather observations. Favorable cloud conditions are near overcast, with stratiform, thin clouds preferably in combination with dense, low-level clouds or fog. A maximal albedo below the horizon of the point of observation provided by low-level clouds or fog (snow cover would meet this purpose as well) is necessary to increase total downwelling Idiff to the maximum extent possible. The smaller the scale at which the required reflections take place is, the smaller is the extinction of Idiff relative to Idir through atmospheric effects and hence the higher can be the resulting values Idiff × I−1dir and G, respectively.

The maximum extreme G+ that was measured is 173% of Gpot; the maximum CRE is >90% of potential Idir. Values of this magnitude have very rarely been observed, however, and only at the exposed, cloud-enveloped mountain ridges. The manifold parameters that have to be optimal at the same time make events of maximal CREs extremely transient and spatially limited and thus almost impossible to detect.

The measurements reveal that the frequency of enhanced G depends on the predominant cloud frequency at the PO (negatively correlated), but the value for mean relative enhancement is approximately constant and hence independent of solar elevation (daytime/season) and dominant weather type. Mean measures for this enhancement (averaged over all sites) are 119% ± 2.0% of clear-sky irradiance Gpot and 34% ± 2.6% for CREs. It must be mentioned that because of the large recording intervals of the extreme values the derived means of the corresponding database are biased. Absolute frequency of enhanced irradiance will be higher than observed, and its true mean value will be lower. A continuous, direct (i.e., not integrating) recording with fast-responding sensors to trace short-lived peaks should be a requisite for further research on enhanced irradiance.

Because of the short-lived nature and limited spatial occurrence of extremely enhanced irradiance, its impact on weather and climate should be insignificant. Mean enhanced irradiance, however, may have a small effect on the climate at locations such as Vilcabamba where >50% of the discrete extreme recordings are higher than Gpot (1100–1500 LT). Although the presence of enhanced irradiance does not change the size of an established radiation budget at a site, the quantitative contribution to it may often be underestimated. However, one should keep in mind that clouds that enhance irradiance also attenuate irradiance, usually to an extent that the total CRE over a longer period results in a reduction of G. Exceptions are areas in which fixed cloud patterns exist, like at mountain ridges. At the site Cajanuma, for example, the CRE of standing clouds regularly enhances solar irradiance by ∼10% for as long as several consecutive hours (Fig. 2). Here, the positive CRE will have a measurable effect on the microclimate.

Whether mean or extreme enhanced irradiance has manifested in specific properties of life forms in the equatorial Andes—in particular, vegetation—is an interesting question. Although very high SEs exist anywhere between about 30°N and 30°S, only the mountains at the lowest latitudes are lushly grown up to 4000 m MSL (i.e, up into regions where the highest absolute values of G+ occur). Tightly related with this issue because of cell-damaging and mutagen potential would be the extent to which UV radiation is prone to extreme enhancement as well. Parisi and Downs (2004) have reported similar magnitudes of CRE-induced mean enhancements for specific biologically active (harmful) spectra of UVB: 121%–125% for snow blindness (photokeratitis), cataracts, general plant damage, and fish melanoma “action spectra.” For DNA action spectra (in particular, shortwave), even an outlier value of 140% was determined. Enhancements of this action spectrum lasted up to 85 min; average duration was 11.3 min (measuring period: January–June 2003; site 27.3°S, 700 m MSL; Queensland, Australia). The authors reason that the observed enhanced doses of UVB and explicitly the intermittent exposure to (enhanced) UV radiation are likely to contribute to the risk of skin carcinoma in humans.

Last, an important consequence of enhanced irradiance incident at the earth’s surface is the potential misinterpretation of surface reflectance retrievals and aerosol optical thickness from high-resolution satellite images (e.g., if ignored or not well quantified; Wen et al. 2001). In a case study on a field of fair-weather cumulus with heterogeneously sized clear patches in it, the authors detected considerable “cloud adjacency” effects at the earth’s surface, kilometers away from the clouds. The effects decrease exponentially within a 0.5-km distance of the cloud borders but at 2–5 km of mean cloud-free distance still stay at a statistically significant, elevated level. This finding is more confirmation of the assumption that all-sky CREs, and not only CREs from clouds near the sun, contribute to enhanced irradiance.

Acknowledgments

We thank Prof. D. P. Häder, University of Erlangen, and Asst. Prof. A. Brenning, University of Waterloo, for assistance with the script; Dr. F. Vignola, University of Oregon/Solar Radiation Monitoring Laboratory, Sr. Sci. D. Myers, NREL, and G. Pelletier for initial discussions and encouraging the investigation; R. Schrader of Thies Clima and Ir. L. van Wely of Kipp and Zonen for technical support; Dipl. Geogr. R. Beger for fieldwork; and the German Council of Research (DFG) for funding the project.

REFERENCES

  • Barry, R. G., 1992: Mountain Weather and Climate. 2nd ed. Routledge, 409 pp.

  • Bird, R. E., , and R. L. Hulstrom, 1981: Simplified clear sky model for direct and diffuse insolation on horizontal surfaces. Tech. Rep. SERI/TR-642-761, 38 pp. [Available online at http://rredc.nrel.gov/solar/pubs/PDFs/TR-642-761.pdf.].

  • Blumthaler, M., , W. Ambach, , and M. Salzgeber, 1994: Effects of cloudiness on global and diffuse UV irradiance in a high mountain area. Theor. Appl. Climatol., 50 , 2330.

    • Search Google Scholar
    • Export Citation
  • Calbó, J., , D. Pagès, , and J-A. González, 2005: Empirical studies of cloud effects on UV radiation: A review. Rev. Geophys., 43 .RG2002, doi:10.1029/2004RG000155.

    • Search Google Scholar
    • Export Citation
  • Cede, A., , M. Blumthaler, , E. Luccini, , R. D. Piacentini, , and L. Nuñez, 2002: Effects of clouds on erythemal and total irradiance as derived from data of the Argentine Network. Geophys. Res. Lett., 29 .2223, doi:10.1029/2002GL015708.

    • Search Google Scholar
    • Export Citation
  • Conover, J. H., 1965: Cloud and terrestrial albedo determinations from TIROS satellite pictures. J. Appl. Meteor., 4 , 378386.

  • Emck, P., 2007: A climatology of south Ecuador. Ph.D. dissertation, University of Nürnberg-Erlangen, 272 pp. [Available online at http://www.opus.ub.uni-erlangen.de/opus/frontdoor.php?source_opus=656.].

  • Estupiñan, J. G., , S. Raman, , G. H. Grescenti, , J. J. Streicher, , and W. F. Barnard, 1996: Effects of clouds and haze on UV-B radiation. J. Geophys. Res., 101 , 1680716816.

    • Search Google Scholar
    • Export Citation
  • Foyo-Moreno, I., , I. Alados, , F. J. Olmo, , and L. Aladosaarboledas, 2003: The influence of cloudiness on UV global irradiance (295–385 nm). Agric. For. Meteor., 120 , 101111.

    • Search Google Scholar
    • Export Citation
  • McFarlane, S. A., , and K. F. Evans, 2004: Clouds and shortwave fluxes at Nauru. Part II: Shortwave flux closure. J. Atmos. Sci., 61 , 26022615.

    • Search Google Scholar
    • Export Citation
  • Parisi, A. V., , and N. Downs, 2004: Variation of the enhanced biologically damaging solar UV due to clouds. Photochem. Photobiol. Sci., 3 , 643647.

    • Search Google Scholar
    • Export Citation
  • Petersen, E., 1982: Solstråling og dagslys—målt og beregnet (Solar radiation and daylight—Measured and calculated). Tech. Rep. 34, Lysteknisk Laboratorium, Lyngby, Denmark, 224 pp.

  • Pfister, G., , R. L. Mckenzie, , J. B. Liley, , and A. Thomas, 2003: Cloud coverage based on all-sky imaging and its impact on surface solar irradiance. J. Appl. Meteor., 42 , 14211434.

    • Search Google Scholar
    • Export Citation
  • Sabburg, J., , and J. Wong, 2000: The effect of clouds on enhancing UVB irradiance at the earth’s surface: A one year study. Geophys. Res. Lett., 27 , 33373340.

    • Search Google Scholar
    • Export Citation
  • Schade, N. H., , A. Macke, , H. Sandmann, , and C. Stick, 2007: Enhanced solar global irradiance during cloudy sky conditions. Meteor. Z., 16 , 295304.

    • Search Google Scholar
    • Export Citation
  • Schmidt, D., 1999: Das Extremklima der nordchilenischen Hochatacama unter besonderer Berücksichtigung der Höhengradienten (The extreme climate of the north Chilean High-Atacama with special focus on altitudinal gradients). Dresdener Geographische Beiträge 4, Universität Dresden, 100 pp.

  • Wen, G., , R. F. Cahalan, , T-S. Tsay, , and L. Oreopoulos, 2001: Impact of cumulus cloud spacing on Landsat atmospheric correction and aerosol retrieval. J. Geophys. Res., 106 , 1212912138.

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

Study area in the southern Ecuadorian Andes. The major ridge, the Cordillera Real, is also the major climate divide, parting the climate into very moist regimes (to the east) and semiarid regimes (to the west).

Citation: Journal of Applied Meteorology and Climatology 47, 11; 10.1175/2008JAMC1861.1

Fig. 2.
Fig. 2.

The average light/cloud conditions at the different sites (order of sites: from left to right and up to down corresponds to from west to east; cf. map in Fig. 1) during the most frequent weather condition (easterly trades; >70% yr−1). Note the 10%–15% chance of 10% enhanced irradiance at the Cajanuma station around noon. Statistics are based on 1-h averages. The source of the clear-sky values is the Bird and Hulstrom (1981) model.

Citation: Journal of Applied Meteorology and Climatology 47, 11; 10.1175/2008JAMC1861.1

Fig. 3.
Fig. 3.

(a) Clear-sky irradiance appropriate for calibration of the model during morning hours. (b) Good compliance of empirical clear-sky global irradiance values with model clear-sky global irradiance.

Citation: Journal of Applied Meteorology and Climatology 47, 11; 10.1175/2008JAMC1861.1

Fig. 4.
Fig. 4.

Selected daily courses of irradiance at the Páramo weather station (3400 m MSL) during different cloud cover conditions. Values indicate the respective day maximum of the variable minimum, mean, and maximum. (a) Clear-sky irradiance. (b) Clouds produce considerably reduced and enhanced G around noon. (c), (d) Peak values of the dataset. (e) Irradiance during a day with heavy overcast and light rain.

Citation: Journal of Applied Meteorology and Climatology 47, 11; 10.1175/2008JAMC1861.1

Fig. 5.
Fig. 5.

Diurnal distribution of instantaneous values of absolute maxima Gmax from time intervals of 1 and 2 h (n > 8000). The thicker, solid curve is the upper-envelope curve of the model clear-sky values Gpot of a year.

Citation: Journal of Applied Meteorology and Climatology 47, 11; 10.1175/2008JAMC1861.1

Fig. 6.
Fig. 6.

Monthly maximum values of G+ tend to be minimal and maximal when maximum Gpot is minimal and maximal.

Citation: Journal of Applied Meteorology and Climatology 47, 11; 10.1175/2008JAMC1861.1

Fig. 7.
Fig. 7.

(a), (b) Enhancement of global irradiance is independent of SE. In addition, the mean amplification factor is the same at all sites, regardless of the altitudes or prevailing humidity (transmissivity) conditions. (c) Averages of all observation sites are summarized. Mean enhancement of current potential radiation is 119.6%; maximum excess occasionally reaches 170%. The general trend of decreasing absolute maxima beyond an SE of 75° must be credited to a decreasing number of observations, not to decreasing excessive irradiance itself. (d) The reason for a decreasing number of observations is that SEs of over 73° are rarest.

Citation: Journal of Applied Meteorology and Climatology 47, 11; 10.1175/2008JAMC1861.1

Fig. 8.
Fig. 8.

Transparent and opaque clouds near the sun produced 1552 W m−2 on one witnessed occasion. During the observation, the values did not drop below 1534 W m−2. The picture was taken from below the pyranometer situated in the center and obscuring the sun. (Model clear-sky values are all horizontal.)

Citation: Journal of Applied Meteorology and Climatology 47, 11; 10.1175/2008JAMC1861.1

Fig. 9.
Fig. 9.

With decreasing cloudiness, mean minimum and maximum minimum irradiance increase until a clear sky is established. Beyond that point, mean minimum and especially maximum minimum irradiance decrease again. At the Páramo station, this effect is more abrupt because the location (3400 m MSL) reaches nearer to or into the clouds. The decrease of maximum minimum irradiance indicates that an increasing cloud cover accompanies increasing enhanced irradiance.

Citation: Journal of Applied Meteorology and Climatology 47, 11; 10.1175/2008JAMC1861.1

Fig. 10.
Fig. 10.

The intersection of the regressions marks the empirical maximum extreme irradiance at the respective weather station. The regressions are retrieved with pairs of maximum measured vs maximum potential (clear sky) irradiance (upper regressions) and with pairs of measured maximum vs. minimum potential (clear sky) irradiance (hence, they are envelope functions of all measured extremes vs. their clear-sky pendant). Stability indices of the inclined regressions are r2 = 0.99 and 1.00. The neighborhood of the two measured extreme values of >1800 W m−2 from Páramo (left; small arrows) values to the intersection shows that these values were “lucky strikes” but are concordant with the phenomenon.

Citation: Journal of Applied Meteorology and Climatology 47, 11; 10.1175/2008JAMC1861.1

Fig. 11.
Fig. 11.

Strong dependency between extrapolated empirical extreme solar irradiance and altitude. The extinction rate for extreme irradiance is high, but the gradient is typical for a decreasing radiation flux with decreasing altitude. The local general moisture conditions (water vapor thickness) at a site seem to be of importance as well: the drier stations tend to higher extreme irradiance (positive residuals); 2000 W m−2 would be found at 4880 m MSL and higher.

Citation: Journal of Applied Meteorology and Climatology 47, 11; 10.1175/2008JAMC1861.1

Fig. 12.
Fig. 12.

The scatter schemes show that appropriate arrangements of an infinitely distant light source (creating a spatially homogenous, parallel light field) and passive diffusers are sufficient to boost diffuse irradiance to considerable magnitudes—in theory even up to 100% of the irradiance Idir coming directly from the source. (a) Two identical extensive, plane-parallel, thin diffusers, with albedo 0.5/transparency 50%, together produce a downward diffuse light flux of 67% × Idir in layer I. If the lower diffuser is replaced by a surface with an albedo of 0.2, diffuse downward flux still amounts to 55.6% × Idir. (b) When increasing total reflectivity of the lower diffusers by increasing their number or, equivalent, by assuming a single, very dense diffuser (aggregated diffusers), total downwelling diffuse shortwave radiation may sum to values of ≥80% × Idir in layer I—(c) in theory, even up to 100%. Note that either a higher or a lower albedo (transparency) of the upper diffuser would result in less downwelling diffuse irradiance in layer I.

Citation: Journal of Applied Meteorology and Climatology 47, 11; 10.1175/2008JAMC1861.1

Fig. 13.
Fig. 13.

Ranked relative diffuse irradiance of all weather stations. Although very rare, Idiff has reached >80% and even >90% of Idir. The midlevel weather stations Cordillera del Consuelo, El Tiro, and, in particular, Cajanuma, apparently most frequently situated between the clouds, are well represented in the top ranking while sunny Vilcabamba’s first appearance is rank 44.

Citation: Journal of Applied Meteorology and Climatology 47, 11; 10.1175/2008JAMC1861.1

Fig. 14.
Fig. 14.

(a) Frequency of 2-h maxima of G+ from instantaneous measurements; (b) share of G+ that exceeds 1400 W m−2; (c) frequency of clear-sky conditions (based on hourly means; Gmeas = Gpot ± 5%).

Citation: Journal of Applied Meteorology and Climatology 47, 11; 10.1175/2008JAMC1861.1

Fig. 15.
Fig. 15.

Histograms of ratios of Idiff and potential (model) Idir of values G+. Mean ratio is remarkably constant; mean ratio across all stations is 34.8% ± 2.6%. Only mid- and upper-level stations have (traces of) ratios of 80% and more.

Citation: Journal of Applied Meteorology and Climatology 47, 11; 10.1175/2008JAMC1861.1

Fig. 16.
Fig. 16.

Control measurements with a luxmeter verified solar irradiance up to 1400 W m−2/130% of clear-sky irradiance.

Citation: Journal of Applied Meteorology and Climatology 47, 11; 10.1175/2008JAMC1861.1

Save