Search NASA⌕ Search
Home/Archive/DOE OSTI

DOE OSTI · 1826668

Physics-Guided Machine Learning for Prediction of Cloud Properties in Satellite-Derived Solar Data

Buster, Grant·Bannister, Mike·Habte, Aron·Hettinger, Dylan·Maclaurin, Galen·Rossol, Michael·Sengupta, Manajit (ORCID:0000000156224503)·Xie, Yu
View original source ↗Export metadata ↓

Abstract

With over 20 years of high-resolution surface irradiance data covering most of the western hemisphere, the National Solar Radiation Database (NSRDB) is a vital public data asset. The NSRDB uses a two-step Physical Solar Model (PSM) that explicitly considers the effects of clouds and other atmospheric variables on radiative transfer. High-quality physical and optical cloud properties derived from satellite imagery are perhaps the most important data inputs to the PSM, representing the greatest source of radiation attenuation and scattering. However, traditional methods for cloud property retrieval have their own limitations and are unable to accurately predict cloud properties outside of nominal conditions. We introduce a physics-guided neural network that can accurately predict cloud properties when traditional methods fail or are inaccurate. Using this framework, we show reductions in relative Root Mean Square Error (RMSE) for Global Horizontal Irradiance (GHI) up to 13 percentage points for timesteps that previously had missing or low-quality cloud property data. We expect that this methodology will be effective in improving the quality of cloud property and solar irradiance data in the NSRDB.

Explore related subjects

cloud propertiesmachine learningOTHER INSTRUMENTATION,SOLAR ENERGYphysics-guided neural networksremote sensingsatellite-derived irradiancesolar resource data

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Buster, Grant, Bannister, Mike, Habte, Aron, Hettinger, Dylan, Maclaurin, Galen, Rossol, Michael, Sengupta, Manajit (ORCID:0000000156224503), Xie, Yu. 2021-08-26. Physics-Guided Machine Learning for Prediction of Cloud Properties in Satellite-Derived Solar Data. https://doi.org/10.1109/pvsc43889.2021.9519065

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related reports

NASA NTRS2019

Observation of absorbing aerosols above clouds over the south-east Atlantic Ocean from the geostationary satellite SEVIRI – Part 1: Method description and sensitivity

High-temporal-resolution observations from satellites have a great potential for studying the impact of biomass burning aerosols and clouds over the south-east Atlantic Ocean (SEAO). This paper presents a method developed to simultaneously retrieve aerosol and cloud properties in aerosol above-cloud conditions from the geostationary instrument Meteosat Second Generation/Spinning Enhanced Visible and Infrared Imager (MSG/SEVIRI). The above-cloud aerosol optical thickness (AOT), the cloud optical thickness (COT) and the cloud droplet effective radius (CER) are derived from the spectral contrast and the magnitude of the signal measured in three channels in the visible to shortwave infrared region. The impact of the absorption from atmospheric gases on the satellite signal is corrected by applying transmittances calculated using the water vapour profiles from a Met Office forecast model. The sensitivity analysis shows that a 10 % error on the humidity profile leads to an 18.5 % bias on the above-cloud AOT, which highlights the importance of an accurate atmospheric correction scheme. In situ measurements from the CLARIFY-2017 airborne field campaign are used to constrain the aerosol size distribution and refractive index that is assumed for the aforementioned retrieval algorithm. The sensitivities in the retrieved AOT, COT and CER to the aerosol model assumptions are assessed. Between 09:00 and 15:00 UTC, an uncertainty of 40 % is estimated on the above-cloud AOT, which is dominated by the sensitivity of the retrieval to the single-scattering albedo. The absorption AOT is less sensitive to the aerosol assumptions with an uncertainty generally lower than 17 % between 09:00 and 15:00 UTC. Outside of that time range, as the scattering angle decreases, the sensitivity of the AOT and the absorption AOT to the aerosol model increases. The retrieved cloud properties are only weakly sensitive to the aerosol model assumptions throughout the day, with biases lower than 6 % on the COT and 3 % on the CER. The stability of the retrieval over time is analysed. For observations outside of the backscattering glory region, the time series of the aerosol and cloud properties are physically consistent, which confirms the ability of the retrieval to monitor the temporal evolution of aerosol above-cloud events over the SEAO.

cloud properties↗
NASA NTRS2014

Systematic Satellite Observations of the Impact of Aerosols from Passive Volcanic Degassing on Local Cloud Properties

The impact of volcanic emissions is a significant source of uncertainty in estimations of aerosol indirect radiative forcing, especially with respect to emissions from passive de-gassing and minor explosions. Understanding the impact of volcanic emissions on indirect radiative forcing is important assessing present day atmospheric properties and also to define the pre-industrial baseline to assess anthropogenic perturbations. We present observations of the time-averaged indirect aerosol effect within 200 km downwind of isolated island volcanoes in regions of low present-day aerosol burden using MODIS and AATSR data. Retrievals of aerosol and cloud properties at Kilauea (Hawaii), Yasur (Vanuatu) and Piton de la Fournaise (Reunion) are rotated about the volcanic vent according to wind direction, so that retrievals downwind of the volcano can be averaged to improve signal to noise ratio. The emissions from all three volcanoes, including those from passive degassing, strombolian activity and minor explosions lead to measurably increased aerosol optical depth downwind of the active vent. Average cloud droplet effective radius is lower downwind of the volcano in all cases, with the peak difference in effective radius ranging from 48 microns at the different volcanoes. A comparison of these observations with cloud properties at isolated islands with no significant source of aerosol suggests that these patterns are not purely orographic in origin. This approach sets out a first step for the systematic measurement of the effects of present day low altitude volcanic emissions on cloud properties, and our observations of unpolluted, isolated marine settings may capture processes similar to those in the preindustrial marine atmosphere.

cloud properties↗
NASA NTRS2013

Effects of Tunable Data Compression on Geophysical Products Retrieved from Surface Radar Observations with Applications to Spaceborne Meteorological Radars

This paper presents results and analyses of applying an international space data compression standard to weather radar measurements that can easily span 8 orders of magnitude and typically require a large storage capacity as well as significant bandwidth for transmission. By varying the degree of the data compression, we analyzed the non-linear response of models that relate measured radar reflectivity and/or Doppler spectra to the moments and properties of the particle size distribution characterizing clouds and precipitation. Preliminary results for the meteorologically important phenomena of clouds and light rain indicate that for a 0.5 dB calibration uncertainty, typical for the ground-based pulsed-Doppler 94 GHz (or 3.2 mm, W-band) weather radar used as a proxy for spaceborne radar in this study, a lossless compression ratio of only 1.2 is achievable. However, further analyses of the non-linear response of various models of rainfall rate, liquid water content and median volume diameter show that a lossy data compression ratio exceeding 15 is realizable. The exploratory analyses presented are relevant to future satellite missions, where the transmission bandwidth is premium and storage requirements of vast volumes of data, potentially problematic.

cloud properties↗