Record high solar irradiance in Western Europe during first COVID-19 lockdown largely due to unusual weather
Chiel C. van Heerwaarden, Wouter B. Mol, Menno A. Veerman, Imme B. Benedict, Bert G. Heusinkveld, Wouter H. Knap, Stelios Kazadzis, Natalia Kouremeti, Stephanie Fiedler
CCOVID-19 lockdown contribution to spring surface solar irradiancerecord in Western Europe
Chiel C. van Heerwaarden ([email protected]) , Wouter B. Mol , Menno A.Veerman , Imme B. Benedict , Bert G. Heusinkveld , Wouter H. Knap , Stelios Kazadzis ,Natalia Kouremeti , and Stephanie Fiedler Meteorology and Air Quality Group, Wageningen University, Wageningen, The Netherlands Royal Netherlands Meteorological Institute, De Bilt, The Netherlands Physikalisch-Meteorologisches Observatorium Davos, World Radiation Center (PMOD-WRC), Davos, Switzerland University of Cologne, Institute of Geophysics and Meteorology, Cologne, Germany
August 4, 2020
Spring 2020 broke sun duration records across western Europe. The Netherlands recorded the highest surfaceirradiance since 1928, exceeding the previous extreme by 13 %, and the diffuse fraction of the total irradiancemeasured a record low percentage (38 %). The coinciding irradiance extreme and a reduction in anthropogenicpollution due to the COVID-19 measures triggered the hypothesis that cleaner-than-usual air contributed to therecord. Based on analyses of ground-based and satellite observations and experiments with a radiative transfermodel, we estimate a 1.3 % (2.3 W m − ) increase in surface irradiance with respect to the 2010-2019 meandue to fewer aerosols in the atmosphere, and a 17.6 % (30.7 W m − ) increase due to the exceptional dry andcloud-free weather conditions. The noticeable signal of cleaner air in the observed irradiance in western Europesuggests that larger COVID-19-related irradiance increases can be expected for more polluted regions. A large part of western Europe (Fig. 1a, hatched area) experienced exceptionally sunny and dry weather fromMarch 23 to the end of May. Sun duration extremes were reported in the United Kingdom, Belgium, Germany,and The Netherlands [2, 3, 4, 5] paired with exceptionally deep blue skies [6]. The period had recurring weatherpatterns favorable for sunshine, with persistent north- to easterly flow over Western Europe or weak winds in thecentre of high pressure systems. Clouds reduced daily solar irradiance on average by 22 % in 2020 with respectto clear-sky conditions, 3 σ lower than the 2004-2020 mean reduction of 36 %. This resulted in a time-integrated1 a r X i v : . [ phy s i c s . a o - ph ] A ug )60 ◦ N50 ◦ N40 ◦ N 15 ◦ W 0 ◦ ◦ E M e a n s u r f a ce s o l a r i rr a d i a n ce ( W m − ) b ) Direct Diffuse Total1981-2010 mean % % %
165 16553 % % % % % % − − − − h P a g e o p o t . h e i g h t a n o m a l y ( d a m ) Figure 1:
Surface irradiance in spring 2020 relative to earlier years.
Figure shows a)
500 hPa geopo-tential height anomaly of 2020 spring (March, April, May) with respect to the 1981–2010 climatology, based onERA5 reanalysis [1], hatched area indicate locations where the total irradiance in ERA5 exceeds the 1979–2019maximum by more than 1%, and b) top-10 years of daily mean integrated global horizontal irradiance (GHI)since 1928 for the Veenkampen station from March 1 until May 31, partitioned into direct and diffuse. Thepercentages show the diffuse portion of GHI where such measurements are available. Veenkampen station wasmoved over a distance of 2 km in 2012, 1989 is missing from the record.surface solar irradiance for spring (March, April, May) that was the largest ever observed at the Veenkampenstation (The Netherlands) since 1928 (Fig. 1b). The daily mean irradiance sum of 206 W m − exceeded theprevious record of 2011 by 25 W m − . The diffuse radiation reaching the surface was only 38 % of the totalsolar irradiance in the period, compared to 49–58 % in the other top-ten springs with high irradiance (Fig. 1).These records all happened amid the first European wave of the COVID-19 pandemic [7, 8], during whichmany countries went into lockdown, leading to a reduction in anthropogenic pollution. Less traffic and industrialactivity led to a 30 % loss in NO x and a 20 % loss in SO emissions [9, 10]. The large leap with which theirradiance records were broken made us hypothesize that the reduction in anthropogenic aerosols and contrailsrelated to the COVID-19 lockdown are an additional driving force behind the observed irradiance extremes nextto the mainly cloud-free weather. To test the hypothesis, we analysed data [13] from the Baseline Surface Radiation Network’s (BSRN [14])measurement station in Cabauw, The Netherlands. This station is located in the center of the regions thatreported sun duration records, and has already available observations of irradiance and aerosol optical depth(AOD) for spring 2020 (Fig. 2). The onset of the prolonged time period of fair weather on March 21 coincides2 eb 16 Mar 01 Mar 16 Apr 01 Apr 16 May 01 May 16 May 31050100150200250300350400 D a il y m e a n s u r f a ce s o l a r i rr a d i a n ce ( W m − ) Direct irradianceDiffuse irradiance Clear-sky irradianceAOD (499 nm) PWATFlight departures 0.00.20.40.60.81.01.21.41.6 A e r o s o l o p t i c a l d e p t h ( - ) / F li g h t d e p a r t u r e s ( x ) P r ec i p i t a b l e w a t e r - U T C a v e r a g e ( mm ) Figure 2:
Time series of relevant variables in spring 2020 in Cabauw, NL . These include time series ofmeasured direct and diffuse irradiance (BSRN Cabauw, NL), clear-sky global horizontal irradiance (CopernicusAtm. Monitoring Service (CAMS) McClear dataset for Cabauw), 499 nm aerosol optical depth (AOD) atCabauw measured using a precision filter radiometer and processed at PMOD/WRC [11], daytime precipitablewater (PWAT) at the grid point closest to Cabauw based on ERA5 reanalysis [1] and the weekly moving averageof flight departures at (major) western Europe airports (OpenSkyNetwork COVID-19 dataset [12]). Error barsindicate daily variability ( ± σ ).with the strong drop in flight activity that marked the onset of the COVID-19 lockdown in many Europeancountries (Fig. 2). The fair weather is reflected by the large amounts of total irradiance, i.e., direct and diffuseirradiance taken together in the observations (Fig. 2), and the large contribution of direct solar irradiancetherein. Until May 31, there were only three overcast days. The surface irradiance is gradually increasing overtime towards the end of May, hence the sunny days later in the period weigh more heavily in the mean shownin Fig. 1.Especially the period of 22 to 31 March was remarkably cloud free, e.g., seen by the total irradiance equalto the clear-sky radiation, and values for diffuse irradiance are the smallest in the period. These days recordedthe lowest AOD of the entire period and the lowest precipitable water in the atmosphere (Fig. 2), underliningthe cleanliness and dryness of the air. Radiosonde observations of De Bilt showed strikingly low amounts ofprecipitable water (not shown). Based on ERA5 Reanalysis at a similar location, March 22 to 26 had on average4.0 ± − precipitable water, far below the 1981–2019 mean of 11.5 ± − for the same period.Later, in May, multiple days had a very low AOD, including days with partial cloudiness, e.g., May 11 to 15. In order to assess the potential impact of the COVID-19 lockdown on the irradiance extremes, and to evaluateextremes in weather versus human activity, we discuss the anomalies in atmospheric circulation (Fig. 1a), in3 .0 0.1 0.2 0.3 0.4 0.5 0.6Aerosol optical depth (-)200420062008201020122014201620182020
Figure 3:
Box plot of 550 nm AOD for Cabauw (NL) . Plot is based on hourly CAMS McClear [15]data for spring (March, April, May). The box-and-whisker plots span the 5th to 95th percentile, red dots arethe means, blue squares are the minima. The AOD has a linear trend in the means of -0.0035 AOD per year(r =0.46, p-value=0.0026) and no significant trend in the minima.AOD (Fig. 3), and in contrail formation with respect to their climatology.The weather conditions favouring high irradiance discussed earlier are reflected by a positive anomaly inthe 500-hPa geopotential height in the lockdown period centered over the south of the United Kingdom anda negative anomaly in northern Scandinavia and Russia (Fig. 1a). This pattern is typical for atmosphericblocking conditions [16], which are often drivers of heatwaves in summer and cold spells in winter [17]. Inabsence of temperature extremes, springtime blocking conditions attract relatively less interest [18], despiteregular occurrence [19]. They are, as this study shows, a contributor to surface irradiance extremes due itsrelated cloud-free skies, caused by its dry and sinking air masses.Within cloud-free conditions, irradiance increases with decreasing humidity and AOD. We documented manydays in 2020 with exceptionally low humidity and a low AOD (Fig. 2). To appreciate the AOD observations, wehave to acknowledge the challenge in separating aerosols from homogeneous, optically thin cirrus in observationsusing sun photometers [20] or satellite products [21], which can result in a positively biased AOD. To avoid thisdifficulty and to assess a longer statistic of AOD, we use here the hourly AOD values from the CAMS aerosolproduct (Fig. 3), which compares well against ground-based observations over Europe [22]. Our analysishighlights that spring 2020 had the lowest median in hourly AOD since 2004. Spring 2020 was also among thesprings with the lowest mean AOD, with only 2015 and 2017 being lower, but its minimum and 5th percentilevalues did not stand out from the statistics of hourly AODs. Therefore, the reduction in anthropogenic aerosolpollution due to the lockdowns did not lead to new extremely low hourly values of AOD, but rather to frequenthours with low AODs.A reduction in contrail-cirrus due to the drop in flight activity (Fig. 2) is another pathway for the lockdownsto enhance surface irradiance. This is particularly true in Western Europe, which is a hot spot for contrail-cirrus, owing to a combination of high aviation activity, and suitable meteorological conditions. To obtain a4ough estimate of the effect, we compared spring 2020 to 2011 and 2015, which are both among the top-fiveyears in terms of surface irradiance (Fig. 1b), while having contrasting AODs. Year 2011 had the highestmedian AOD in recent years (Fig. 3), whereas 2015 had an AOD statistic comparable to 2020 (Fig. 3). Themeteorological conditions for persistent contrail-cirrus formation at 250 hPa, close to the typical flight levelof 230 hPa [23], are only slightly less favourable (see Supplementary Table 1 and Fig. 5), thus we expect toobserve less contrail-cirrus in 2020. Manual inspection of cirrus and contrail occurrence (see Methods) in NASAWorldview [24] imagery for the Netherlands gives results consistent with our expectation. The images showedthat 2011 and 2015 had about twice as much cirrus, but with 50 % more contrail contamination, compared to2020. Given that contrail-cirrus has a net shortwave radiative forcing in the order of -1 W m − over WesternEurope [25], but can enhance diffuse irradiance with tens of percents [26, 27, 28], we speculate that the lowpresence of contrails contributed to the extremely low diffuse fraction that was observed in 2020. This is likelyto play a small role in the total irradiance extreme compared to the generally low cloud cover in spring 2020(Figs. 1b and 2). W m − a ) Clear-sky, no aerosols, no water vapourClear-sky, no aerosols Clear-sky, with aerosolsClouds, no aerosols W m − b ) ERA5Observations Figure 4:
Modelled surface irradiance with a radiative transfer model . Irradiance is shown for a) casestudy of 29 April, 2020 compared against Cabauw observations, and b) averaged over the months March, April,and May compared against Cabauw ( (cid:63) ) and Veenkampen (+) observations and against the surface irradianceof ERA5. See Table 2 in Supplementary Table 2 for exact values corresponding to bars. We present here estimations of i) the relative importance of different contributing factors to the extreme insurface irradiance, and ii) the anomalies in cloud radiative forcing and the direct aerosol effects. To this end,we used a contemporary radiative transfer model [29] to first reproduce the observed surface irradiance, andsubsequently repeat the calculation without individual components to assess their quantitative contribution tothe surface irradiance. We used the McClear clear-sky radiation product to infer the aerosol contribution [15](see Supplementary Fig. 7 for a validation of the direct aerosol effect as a function of AOD). Further details are5iven in the Methods section. The combination of the experiments with the radiative transfer model and theclear-sky data provides surface irradiance under four conditions, compared against the observations in Fig. 4: • Experiment dark blue resembles the reality, but without aerosols in the atmosphere. • Experiment red additionally removes the clouds. • Experiment orange additionally removes the water vapour. • Experiment light blue is the clear-sky product, thus without clouds but with aerosols.We first show the modelled surface solar irradiance for a single day on 29 April in Fig. 4a for illustratingthe transient behaviour of the radiative transfer model results. The experiment dark blue with clouds, butwithout aerosols closely follows the observed slowly increasing irradiance due to vanishing clouds. It confirmsthe ability of the model to reproduce the time series of surface irradiance (see Supplementary Fig. 6 for detailedvalidation). At noon, removing clouds ( red ) increases the irradiance by 250.7 W m − , and removing watervapour ( orange ) increases irradiance by a further 130.8 W m − , whereas the presence of aerosols ( light blue )lowers the irradiance only by 12.5 W m − . Both the removal of water vapour and clouds have a larger effect onthe irradiance than removing aerosols. This is not surprising due to the typically larger optical depth of cloudsthan aerosols, but is noteworthy in the context of the following statistical assessment.We expand our analysis to the entire spring period for each of the past 10 years in Fig. 4b. Here, the topof each bar segment indicates the total surface irradiance for the situation indicated by its color. If clouds,aerosols, and water vapour are removed, all years have a bar of approximately equal depth, indicating that ourexperiment captures the essence and that leap years and year-to-year variability in atmospheric pressure, ozone,and temperature are of minor importance to the irradiance extremes in 2020. The variation over the years iscomparable between the two measurement stations and is closely following the radiative transfer computationswith clouds ( dark blue ), but without aerosols. The observations are consistently lower than the model simu-lation, consistent with the here removed aerosols. Furthermore, the Cabauw station has a consistently higherirradiance than Veenkampen, due to its closer proximity to the coast, where clouds are less common [30].We quantify the cloud radiative effect at the surface as the difference between the experiment with cloudsand the clear-sky experiment without aerosols ( dark blue minus red ). The irradiance increase due to thereduction in clouds is +30.7 W m − in 2020 with respect to the 2010–2019 mean cloud radiative effect of -88.6W m − . Similarly, we quantify the aerosol effect from the difference between the clear-sky experiment withoutaerosols and the McClear data ( light blue minus ( red ). This computation indicates an increase in irradianceof +2.3 W m − , with respect to the 2010–2019 mean aerosol effect of -13.0 W m − . The water vapour effect isquantified as the clear-sky experiment minus the dry experiment ( red minus orange ) and gives an enhancementof only +1.5 W m − with respect to the 2010–2019 mean water vapour effect of -45.4 W m − . The vapourenhancement is only the contribution to the optical properties of the clear-sky radiation, and that the mostimportant signal of the atmospheric moisture anomaly of 2020 is tightly linked with the low cloud cover. The lowhumidity also affects the aerosol optical depth in the sense that less water vapour is available for condensation6n the aerosol surface keeping the aerosol optical depth smaller than in moist conditions. The quantification ofthe three effects highlights the relative importance of variations in cloud cover over the years in explaining thesurface irradiance. It emphasizes that the sunny weather played the most important role in setting the 2020record in surface irradiance, while the reduced emission of anthropogenic aerosols is of smaller importance, tothe extent that even without the reduction the irradiance record had occurred. During the exceptionally sunny spring in western Europe amid the COVID-19 pandemic in 2020, The Nether-lands received the most solar radiation at the surface since the start of the measurements in 1928 and neverexperienced so little scattering of light (Fig. 1b). The particularly dry atmosphere (Fig. 2) and weather pat-terns favouring sunny weather (Fig. 1a) led to fewer clouds than in previous years. Based on radiative transfercalculations, we estimated the relative contributions of aerosols, water vapour, and clouds and argue that thelatter is the dominant contributor to the new irradiance extreme (Fig. 4), while the impact of COVID-19lockdowns is at least an order of magnitude less.Our analysis separates the contribution of aerosols from that of clouds, allowing us to estimate the irradianceincrease in 2020 due to a reduction in the direct aerosol effect at 2.3 W m − . Measured by the median, spring2020 was the cleanest on record in The Netherlands since 2004 (Fig. 3b). The trend of -3.5 · − y − in themeans continues the generally decreasing trend in anthropogenic aerosol emissions in Europe, an effect knownas brightening from earlier studies [31, 32, 33]. Today’s already low anthropogenic aerosol emissions in Europeare expected to stay small or even further decrease in the future [34], which could enable even larger irradianceextremes. Changes in future circulation and related cloudiness could enforce or counter new extremes, but theuncertainty in projections thereof [18, 35] must be reduced before more definitive statements on future irradianceextremes can be made.Further effects of the aerosol removal are possible, including rapid adjustments of clouds and circulationto associated temperature changes and aerosol effects on cloud microphysical processes, but these can not bequantitatively assessed with the model used here. Qualitatively, a reduction in anthropogenic aerosols could forinstance reduce the cloud albedo and enhance the surface irradiance in cloudy conditions, but the uncertainty inthe quantification of aerosol-cloud interactions is still large [36]. If we account for additional aerosol effects, theimpact of the reduced anthropogenic aerosols due to the lockdown is expected to be larger than our estimateof the instantaneous clear-sky radiative effects of the aerosol removal. Despite all uncertainty, larger effectsof aerosol reductions from COVID-19 lockdowns are expected for typically more polluted regions than westernEurope. With the current spread of the pandemic in strongly polluted regions, more data to test our expectationwill become available in the near future. 7 uthor contributions CvH, WM, and MV designed the study. WM performed the data analysis of the observations. MV performedthe radiative transfer model experiments. CvH, WM, MV, IB and SF interpreted the results. CvH, WM, MV,and SF wrote the manuscript. BH, WK, SK, and NK provided observational data and expertise thereof. Allauthors read the final manuscript and provided feedback.
Acknowledgments
CvH, WM, MV, and BH acknowledge funding from the Dutch Research Council (NWO) (grant: VI.Vidi.192.068).SF acknowledges the funding of the Hans-Ertel-Centre for Weather Research by the German Federal Ministryfor Transportation and Digital Infrastructure, (grant: BMVI/DWD 4818DWDP5A).
Effects of aviation are estimated by comparing two top-5 irradiance years with relatively high and low AOD with2020, hence the choice of 2011 and 2015. Environmental conditions favourable for persistent contrails, namelya low air temperature at flight level [23], and often occurring supersaturation with respect to ice [37, 38],are quantified using ERA5 reanalysis for a domain covering approximately The Netherlands at 250 and 300hPa. Cirrus occurrence and whether it is contaminated by contrails is manually counted by looking at highresolution satellite images from Terra and Aqua MODIS, available on the NASA Worldview website [24]. Theinspected area covers the Netherlands and closely neighbouring regions (Belgium, western Germany and partof the North Sea). This area is part of the irradiance extreme coverage (Fig. 3 and helps offset the factonly two images close to noon per day are available. Only cirrus and contrails optically thick enough to bedetectable by eye can be counted, anything that is too thin to detect is assumed to have only a very smallimpact on irradiance. Contrail-contamination is counted when there is cirrus with five or more linear (typicallyoverlapping) or unnatural looking (dispersed) condensation trails present. See Supplementary Fig. 5 for a clearexample.
We used the Radiative Transfer for Energetics and RRTM for General circulation model applications—Parallel(RTE+RRTMGP) [29] to reproduce the surface irradiance observations at Cabauw, The Netherlands for spring2020 (March, April, May) in order to construct Fig. 4. The computation requires hourly atmospheric profilesof pressure, temperature, water vapour, liquid water, ice, cloud cover, cloud liquid water, cloud ice, and ozoneat a 0 . × . ◦ grid resolution taken from the ERA5 reanalysis [1]. We used the data on 37 pressure levelsinstead of the 137 native model levels, but the vertical integrals are approximately conserved. We assume i)8louds to be horizontally homogeneous within one grid cell, ii) that adjacent cloud layers have overlap, iii) thatthe spatial correlation between two clouds layers decreases exponentially (using a decorrelation length of 2 km)with increasing vertical distance between the layers, and iv) that separated cloud layers have random overlap.To obtain a statistical distribution of the cloud fields, we sampled 100 vertical profiles, calculated radiativefluxes for each profile and subsequently averaged the surface irradiance. To infer the effect of aerosols, we usedthe surface irradiance product of the Copernicus Atmosphere Monitoring service (CAMS) McClear Clear-SkyIrradiating service [15]. All data used in this manuscript is either available from public sources or included with this manuscript. Publicsources are cited in the manuscript. Data that is non-public will be made publicly available after paper isaccepted and a DOI will be attached. Until then a zip file is included with the manuscript.
References [1] Hersbach, H. et al.
The ERA5 global reanalysis.
Quarterly Journal of the Royal Meteorological Society , 1999–2049 (2020).[2] Madge, G. May 2020 becomes the sunniest calendar month on record.
United Kingdom Met Office (2020). URL .[3] DWD. German weather in spring 2020.
Deutsche Wetter Dienst (2020). URL .[4] KNMI. Sunniest spring since start of measurements.
Royal Dutch Mete-orological Institute (2020). URL .[5] KMI. A dry and exceptionally sunny spring.
Royal Meteorological Institute (2020). URL .[6] Knap, W. & Meirink, J. F. Uitzonderlijk blauwe voorjaarsluchten.
Royal Nether-lands Meteorological Institute (2020). URL .[7] COVID-19 dashboard by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins Uni-versity (JHU). URL https://coronavirus.jhu.edu/map.html . Accessed: 2020-07-21.[8] European Centre for Disease Prevention and Control, COVID-19 pandemic. URL . Accessed: 2020-07-21.99] Forster, P. M. et al.
Current and future global climate impacts resulting from COVID-19.
Nature ClimateChange (2020).[10] Muhammad, S., Long, X. & Salman, M. COVID-19 pandemic and environmental pollution: A blessing indisguise?
Science of The Total Environment , 138820 (2020).[11] Kazadzis, S., Kouremeti, N., Nyeki, S., Gr¨obner, J. & Wehrli, C. The world optical depth research andcalibration center (WORCC) quality assurance and quality control of GAW-PFR AOD measurements.
Geoscientific Instrumentation, Methods and Data Systems , 39–53 (2018).[12] Schafer, M., Strohmeier, M., Lenders, V., Martinovic, I. & Wilhelm, M. Bringing up OpenSky: A large-scale ADS-b sensor network for research. In IPSN-14 Proceedings of the 13th International Symposium onInformation Processing in Sensor Networks (IEEE, 2014).[13] Knap, W. Basic and other measurements of radiation at station Cabauw (2020-02) (2020). URL https://doi.pangaea.de/10.1594/PANGAEA.913473 .[14] Driemel, A. et al.
Baseline surface radiation network (BSRN): structure and data description (1992-2017).
Earth System Science Data , 1491–1501 (2018).[15] Gschwind, B. et al. Improving the McClear model estimating the downwelling solar radiation at groundlevel in cloud-free conditions – McClear-v3.
Meteorologische Zeitschrift , 147–163 (2019).[16] Folland, C. K. et al. The summer North Atlantic Oscillation: Past, present, and future.
Journal of Climate , 1082–1103 (2009).[17] Pfahl, S. & Wernli, H. Quantifying the relevance of atmospheric blocking for co-located temperatureextremes in the Northern Hemisphere on (sub-) daily time scales. Geophysical Research Letters (2012).[18] Woollings, T. et al. Blocking and its response to climate change.
Current climate change reports , 287–300(2018).[19] Brunner, L., Hegerl, G. C. & Steiner, A. K. Connecting atmospheric blocking to European temperatureextremes in spring. Journal of Climate , 585–594 (2016).[20] Chew, B. N. et al. Tropical cirrus cloud contamination in sun photometer data.
Atmospheric En-vironment , 6724 – 6731 (2011). URL .[21] Kaufman, Y. J. et al. A critical examination of the residual cloud contamination and diurnal samplingeffects on modis estimates of aerosol over ocean.
IEEE Transactions on Geoscience and Remote Sensing , 2886–2897 (2005).[22] Gueymard, C. A. & Yang, D. Worldwide validation of cams and merra-2 reanalysis aerosol optical depthproducts using 15 years of aeronet observations. Atmospheric Environment , 117216 (2020).1023] Burkhardt, U. & K¨archer, B. Global radiative forcing from contrail cirrus.
Nature Climate Change ,54–58 (2011).[24] NASA Worldview. URL https://worldview.earthdata.nasa.gov . Accessed: 2020-07-25.[25] Stuber, N., Forster, P., R¨adel, G. & Shine, K. The importance of the diurnal and annual cycle of air trafficfor contrail radiative forcing. Nature , 864–867 (2006).[26] Feister, U. & Shields, J. Cloud and radiance measurements with the VIS/NIR Daylight Whole Sky Imagerat Lindenberg (Germany).
Meteorologische Zeitschrift , 627–639 (2005).[27] Gueymard, C. A. Temporal variability in direct and global irradiance at various time scales as affected byaerosols. Solar Energy , 3544 – 3553 (2012). URL . Solar Resources.[28] Weihs, P. et al. Potential impact of contrails on solar energy gain.
Atmospheric Measurement Techniques , 1089–1096 (2015).[29] Pincus, R., Mlawer, E. J. & Delamere, J. S. Balancing accuracy, efficiency, and flexibility in radiationcalculations for dynamical models. Journal of Advances in Modeling Earth Systems Geophysical Research Letters , L11201(2004).[32] Wild, M. et al. From dimming to brightening: Decadal changes in solar radiation at earth’s surface.
Science , 847–850 (2005).[33] Wild, M. Enlightening Global Dimming and Brightening.
Bulletin of the American Meteorological Society , 27–37 (2012).[34] Fiedler, S. et al. First forcing estimates from the future CMIP6 scenarios of anthropogenic aerosol opticalproperties and an associated Twomey effect.
Geoscientific Model Development , 989–1007 (2019).[35] Bony, S. et al. Clouds, circulation and climate sensitivity.
Nature Geoscience , 261–268 (2015).[36] Bellouin, N. et al. Bounding global aerosol radiative forcing of climate change.
Reviews of Geophysics ,e2019RG000660 (2020).[37] Jensen, E. J. et al. Environmental conditions required for contrail formation and persistence.
Journal ofGeophysical Research: Atmospheres , 3929–3936 (1998).[38] Lee, D. S. et al.
Aviation and global climate change in the 21st century.
Atmospheric Environment ,3520–3537 (2009). 11 upplementary material The supplementary material contains a series of figures and tables with additional material to complement andvalidate the analyses and radiative transfer model computations that are discussed in the main text. All itemsare referenced from the main text.Table 1: Percentage of days in March, April and May with visible cirrus and the percentage those days whichare visibly contaminated with contrails. Based on Terra and Aqua MODIS imagery on NASA Worldview [24].2011 2015 2020Cirrus 51 60 30Contrail contaminated 62 59 42Figure 5: Example from NASA Worldview of cirrus that contains (and is therefore enhanced in coverage andoptical thickness) contrails, from narrow and straight ones to older and more dispersed. The date of this imageis 11 May 2015.Table 2: Surface irradiance per experiment in units of W m − for the period March, April, May for each of theyears 2010–2020. 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020Dry ( orange ) 308.2 308.1 310.0 309.1 308.6 307.7 309.9 309.4 308.5 307.8 310.1Clear ( red ) 264.6 262.4 263.4 265.3 261.9 263.8 264.8 262.9 261.4 262.6 266.2McClear ( light blue ) 252.1 247.1 249.0 251.6 250.7 254.3 253.2 249.2 246.6 249.3 255.4Clouds ( dark blue ) 170.5 191.9 168.5 153.4 173.5 179.6 173.6 182.1 179.9 174.2 208.312 year: 2010 year: 2011 year: 2012 year: 2013 year: 2014 year: 2015 year: 2016 year: 2017 year: 2018 year: 2019 year: 2020 CabauwVeenkampen Observations [W m − ] RR T M G P ( C l o ud , n o a e r o s o l s ) [ W m − ] Figure 6: Daily mean surface irradiances based on RTE+RRTMGP simulation with clouds but without aerosolsagainst daily mean surface irradiances based on observations at Cabauw and at Veenkampen. Each year is shownin a single panel and the data points represent all individual days in March, April and May.13 year: 2010 year: 2011 year: 2012 year: 2013 year: 2014 year: 2015 year: 2016 year: 2017 year: 2018 year: 2019 year: 2020
Daily mean Aerosol Optical Depth [ − ] D i r ec t a e r o s o l e ff ec t [ W m − ]]