Retrieval of aerosol optical depth from surface solar radiation measurements using machine learning algorithms, non-linear regression and a radiative transfer-based look-up table

Huttunen, Jari T.; Kokkola, Harri; Mielonen, Tero; Mononen, Mika E.; Lipponen, Antti; Reunanen, Juha; Lindfors, Anders V.; Mikkonen, Santtu; Lehtinen, K. E. J.; Kouremeti, Natalia; Bais, A. F.; Niska, Harri; Arola, Antti

doi:10.5194/acp-16-8181-2016

Cited by 26 publications

(14 citation statements)

References 34 publications

(38 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This equation is widely used to calculate g from b (Andrews et al, 2006;Horvath et al, 2016;Kassianov et al, 2007). We use the field measurement results to test its reliability.…”

Section: Influence Of Rh On Gmentioning

confidence: 99%

Calculating the aerosol asymmetry factor based on measurements from the humidified nephelometer system

Zhao

Kuang

et al. 2018

Atmos. Chem. Phys.

View full text Add to dashboard Cite

Abstract. The aerosol asymmetry factor (g) is one of the most important factors for assessing direct aerosol radiative forcing. However, little attention has been paid to the measurement and parameterization of g. In this study, the characteristics of g are studied based on field measurements over the North China Plain (NCP) using the Mie scattering theory. The results show that calculated g values for dry aerosol can vary over a wide range (between 0.54 and 0.67). Furthermore, when ambient relative humidity (RH) reaches 90 %, g is significantly enhanced by a factor of 1.2 due to aerosol hygroscopic growth. For the first time, a novel method of calculating g based on measurements from the humidified nephelometer system is proposed. This method can constrain the uncertainty of g to within 2.56 % for dry aerosol populations and 4.02 % for ambient aerosols, providing that aerosol hygroscopic growth is taken into account. Sensitivity studies show that aerosol hygroscopicity plays a vital role in the accuracy of predicting g.

show abstract

“…This equation is widely used to calculate g from b (Andrews et al, 2006;Horvath et al, 2016;Kassianov et al, 2007). We use the field measurement results to test its reliability.…”

Section: Influence Of Rh On Gmentioning

confidence: 99%

Calculating the aerosol asymmetry factor based on measurements from the humidified nephelometer system

Zhao

Kuang

et al. 2018

Atmos. Chem. Phys.

View full text Add to dashboard Cite

show abstract

“…However, the number of sites where direct irradiance is routinely measured with a pyrheliometer is still low, and was so low before the 1970s that this avenue might not be of any practical importance in climate studies. Recently, a sophisticated method was proposed to estimate AOD from global horizontal irradiance (GHI) using machine-learning algorithms (Huttunen et al, 2016). GHI observations might be relatively more common than those of direct irradiance, but are less directly linked to AOD than the latter, and are affected by larger experimental uncertainties.…”

Section: Introductionmentioning

confidence: 99%

Reconstruction of historical aerosol optical depth time series over Romania during summertime

Dumitrescu

Gueymard²,

Badescu

2017

Intl Journal of Climatology

View full text Add to dashboard Cite

Ground‐based meteorological observations are used here as proxy data, as well as aerosol reanalysis estimates for training, to reconstruct aerosol optical depth (AOD) time series over Romania during a multidecadal period (1961–2015). The approach uses sunshine duration fraction (SDF) data computed for cloudless summer days (identified from 6‐hourly cloud observations) and AOD data from the Modern‐Era Retrospective analysis for Research and Applications, version 2 (MERRA‐2) reanalysis. MERRA‐2 AOD data were validated first by comparing them to high‐quality AErosol RObotic NETwork (AERONET) ground observations at eight stations located in Central and Eastern Europe. The comparison of MERRA‐2 AOD data with AERONET ground observations showed that the reanalysis estimates are reasonably accurate, but tend to underestimate AOD in the region of interest. Spaceborne AOD observations from Moderate Resolution Imaging Spectroradiometer (MODIS) were also considered, but were found too biased and not complete enough for this task. A significant linear relationship between MERRA‐2's AOD and SDF data at the 57 of the 86 available stations was found, allowing the use of both a linear and bivariate regression models to back‐calculate the historical AOD time series from SDF observations. This reconstruction is restricted to clear summer days here because of the limited temporal resolution of the available cloud observations. The long‐term trends of the computed AOD time series correctly delineate the expected dimming and brightening periods that have been widely reported over Europe. The proposed method is suitable over areas where very long and concomitant time series of daily sunshine and hourly cloud observations are available, but also has limitations that are discussed.

show abstract

“…To our knowledge, this method has not been used routinely in NPF event analysis. Hyvönen et al (2005) used data mining techniques to analyze aerosol formation in Hyytiälä, Finland, and Mikkonen et al (2006Mikkonen et al ( , 2011 applied similar methods to datasets recorded in SPC (2006) and in Melpitz and Hohenpeißenberg, Germany (2011). They studied different variables and parameters that may be behind NPF but they did not make automatic classification for NPF events.…”

Section: Resultsmentioning

confidence: 99%

“…Developing the NPF classification to also take into account other data measured at the same site, such as meteorological data and concentrations of trace gases, has allowed utilization of statistical methods to search for variables which could best explain and predict the occurrence of NPF. Hyvönen et al (2005) and Mikkonen et al (2006) applied discriminant analysis for multiyear datasets of aerosol size distributions and several gas and meteorological parameters measured at Hyytiälä, Finland and San Pietro Capofiume, Italy, respectively. Both of these stud-ies were able to find the characteristic conditions for NPF event days in each site and it was seen that the conditions differ significantly.…”

Section: Introductionmentioning

confidence: 99%

Identification of new particle formation events with deep learning

et al. 2018

Self Cite

View full text Add to dashboard Cite

Abstract. New particle formation (NPF) in the atmosphere is globally an important source of climate relevant aerosol particles. Occurrence of NPF events is typically analyzed by researchers manually from particle size distribution data day by day, which is time consuming and the classification of event types may be inconsistent. To get more reliable and consistent results, the NPF event analysis should be automatized. We have developed an automatic analysis method based on deep learning, a subarea of machine learning, for NPF event identification. To our knowledge, this is the first time that a deep learning method, i.e., transfer learning of a convolutional neural network (CNN), has successfully been used to automatically classify NPF events into different classes directly from particle size distribution images, similarly to how the researchers carry out the manual classification. The developed method is based on image analysis of particle size distributions using a pretrained deep CNN, named AlexNet, which was transfer learned to recognize NPF event classes (six different types). In transfer learning, a partial set of particle size distribution images was used in the training stage of the CNN and the rest of the images for testing the success of the training. The method was utilized for a 15-yearlong dataset measured at San Pietro Capofiume (SPC) in Italy. We studied the performance of the training with different training and testing of image number ratios as well as with different regions of interest in the images. The results show that clear event (i.e., classes 1 and 2) and nonevent days can be identified with an accuracy of ca. 80 %, when the CNN classification is compared with that of an expert, which is a good first result for automatic NPF event analysis. In the event classification, the choice between different event classes is not an easy task even for trained researchers, and thus overlapping or confusion between different classes occurs. Hence, we cross-validated the learning results of CNN with the expert-made classification. The results show that the overlapping occurs, typically between the adjacent or similar type of classes, e.g., a manually classified Class 1 is categorized mainly into classes 1 and 2 by CNN, indicating that the manual and CNN classifications are very consistent for most of the days. The classification would be more consistent, by both human and CNN, if only two different classes are used for event days instead of three classes. Thus, we recommend that in the future analysis, event days should be categorized into classes of "quantifiable" (i.e., clear events, classes 1 and 2) and "nonquantifiable" (i.e., weak events, Class 3). This would better describe the difference of those classes: both formation and growth rates can be determined for quantifiable days but not both for nonquantifiable days. Furthermore, we investigated more deeply the days that are classified as clear events by experts and recognized as nonevents by the CNN and vice versa. Clear misclassifications seem to ...

show abstract

Retrieval of aerosol optical depth from surface solar radiation measurements using machine learning algorithms, non-linear regression and a radiative transfer-based look-up table

Cited by 26 publications

References 34 publications

Calculating the aerosol asymmetry factor based on measurements from the humidified nephelometer system

Calculating the aerosol asymmetry factor based on measurements from the humidified nephelometer system

Reconstruction of historical aerosol optical depth time series over Romania during summertime

Identification of new particle formation events with deep learning

Contact Info

Product

Resources

About