scholarly journals Feasibility study of multi-pixel retrieval of optical thickness and droplet effective radius of inhomogeneous clouds using deep learning

2017 ◽  
Vol 10 (12) ◽  
pp. 4747-4759 ◽  
Author(s):  
Rintaro Okamura ◽  
Hironobu Iwabuchi ◽  
K. Sebastian Schmidt
Keyword(s):  
Deep Learning ◽  
Radiative Transfer ◽  
Optical Thickness ◽  
Cloud Droplet ◽  
Effective Radius ◽  
Radiative Transfer Model ◽  
Transfer Model ◽  
Optical Remote Sensing ◽  
Transfer Effects ◽  
Cloud Properties

Abstract. Three-dimensional (3-D) radiative-transfer effects are a major source of retrieval errors in satellite-based optical remote sensing of clouds. The challenge is that 3-D effects manifest themselves across multiple satellite pixels, which traditional single-pixel approaches cannot capture. In this study, we present two multi-pixel retrieval approaches based on deep learning, a technique that is becoming increasingly successful for complex problems in engineering and other areas. Specifically, we use deep neural networks (DNNs) to obtain multi-pixel estimates of cloud optical thickness and column-mean cloud droplet effective radius from multispectral, multi-pixel radiances. The first DNN method corrects traditional bispectral retrievals based on the plane-parallel homogeneous cloud assumption using the reflectances at the same two wavelengths. The other DNN method uses so-called convolutional layers and retrieves cloud properties directly from the reflectances at four wavelengths. The DNN methods are trained and tested on cloud fields from large-eddy simulations used as input to a 3-D radiative-transfer model to simulate upward radiances. The second DNN-based retrieval, sidestepping the bispectral retrieval step through convolutional layers, is shown to be more accurate. It reduces 3-D radiative-transfer effects that would otherwise affect the radiance values and estimates cloud properties robustly even for optically thick clouds.

scholarly journals Retrieval of optical thickness and droplet effective radius of inhomogeneous clouds using deep learning

2017 ◽  
Author(s):  
Rintaro Okamura ◽  
Hironobu Iwabuchi ◽  
K. Sebastian Schmidt
Keyword(s):  
Deep Learning ◽  
Radiative Transfer ◽  
Optical Thickness ◽  
Three Dimensional ◽  
Cloud Droplet ◽  
Effective Radius ◽  
Radiative Transfer Model ◽  
Transfer Model ◽  
Transfer Effects ◽  
Cloud Properties

Abstract. Three-dimensional (3D) radiative transfer effects are a major source of retrieval errors in satellite-based optical re- mote sensing of clouds. In this study, we present two retrieval methods based on deep learning. We use deep neural networks (DNNs) to retrieve multipixel estimates of cloud optical thickness and column-mean cloud droplet effective radius simultane- ously from multispectral, multipixel radiances. Cloud field data are obtained from large-eddy simulations, and a 3D radiative transfer model is employed to simulate upward radiances from clouds. The cloud and radiance data are used to train and test the DNNs. The proposed DNN-based retrieval is shown to be more accurate than the existing look-up table approach that assumes plane-parallel, homogeneous clouds. By using convolutional layers, the DNN method estimates cloud properties robustly, even for optically thick clouds, and can correct the 3D radiative transfer effects that would otherwise affect the radiance values.

scholarly journals The Plane-Parallel Albedo Bias of Liquid Clouds from MODIS Observations

2007 ◽  
Vol 20 (20) ◽  
pp. 5114-5125 ◽  
Author(s):  
Lazaros Oreopoulos ◽  
Robert F. Cahalan ◽  
Steven Platnick
Keyword(s):  
Optical Thickness ◽  
Radiative Forcing ◽  
Mean Value ◽  
Time Of Day ◽  
Effective Radius ◽  
Radiative Transfer Model ◽  
Transfer Model ◽  
Plane Parallel ◽  
Cloud Properties ◽  
Horizontal Inhomogeneity

Abstract The authors present the global plane-parallel shortwave albedo bias of liquid clouds for two months, July 2003 and January 2004. The cloud optical properties necessary to perform the bias calculations come from the operational Moderate Resolution Imaging Spectroradiometer (MODIS) Terra and MODIS Aqua level-3 datasets. These data, along with ancillary surface albedo and atmospheric information consistent with the MODIS retrievals, are inserted into a broadband shortwave radiative transfer model to calculate the fluxes at the atmospheric column boundaries. The plane-parallel homogeneous (PPH) calculations are based on the mean cloud properties, while independent column approximation (ICA) calculations are based either on 1D histograms of optical thickness or joint 2D histograms of optical thickness and effective radius. The (positive) PPH albedo bias is simply the difference between PPH and ICA albedo calculations. Two types of biases are therefore examined: 1) the bias due to the horizontal inhomogeneity of optical thickness alone (the effective radius is set to the grid mean value) and 2) the bias due to simultaneous variations of optical thickness and effective radius as derived from their joint histograms. The authors find that the global bias of albedo (liquid cloud portion of the grid boxes only) is ∼+0.03, which corresponds to roughly 8% of the global liquid cloud albedo and is only modestly sensitive to the inclusion of horizontal effective radius variability and time of day, but depends strongly on season and latitude. This albedo bias translates to ∼3–3.5 W m−2 of bias (stronger negative values) in the diurnally averaged global shortwave cloud radiative forcing, assuming homogeneous conditions for the fraction of the grid box not covered by liquid clouds; zonal values can be as high as 8 W m−2. Finally, the (positive) broadband atmospheric absorptance bias is about an order of magnitude smaller than the albedo bias. The substantial magnitude of the PPH bias underlines the importance of predicting subgrid variability in GCMs and accounting for its effects on cloud–radiation interactions.

Consistent retrieval of cloud/aerosol single scattering properties and surface reflectance

2021 ◽  
Author(s):  
Marta Luffarelli ◽  
Yves Govaerts
Keyword(s):  
Radiative Transfer ◽  
Optical Thickness ◽  
Solution Space ◽  
Single Scattering ◽  
Aerosol Optical Thickness ◽  
Radiative Transfer Model ◽  
Transfer Model ◽  
Surface Reflectance ◽  
Earth System ◽  
System Components

<p>The CISAR (Combined Inversion of Surface and AeRosols) algorithm is exploited in the framework of the ESA Aerosol Climate Change Initiatiave (CCI) project, aiming at providing a set of atmospheric (cloud and aerosol) and surface reflectance products derived from S3A/SLSTR observations using the same radiative transfer physics and assumptions. CISAR is an advance algorithm developed by Rayference originally designed for the retrieval of aerosol single scattering properties and surface reflectance from both geostationary and polar orbiting satellite observations.  It is based on the inversion of a fast radiative transfer model (FASTRE). The retrieval mechanism allows a continuous variation of the aerosol and cloud single scattering properties in the solution space.</p><p> </p><p>Traditionally, different approaches are exploited to retrieve the different Earth system components, which could lead to inconsistent data sets. The simultaneous retrieval of different atmospheric and surface variables over any type of surface (including bright surfaces and water bodies) with the same forward model and inversion scheme ensures the consistency among the retrieved Earth system components. Additionally, pixels located in the transition zone between pure clouds and pure aerosols are often discarded from both cloud and aerosol algorithms. This “twilight zone” can cover up to 30% of the globe. A consistent retrieval of both cloud and aerosol single scattering properties with the same algorithm could help filling this gap.</p><p> </p><p>The CISAR algorithm aims at overcoming the need of an external cloud mask, discriminating internally between aerosol and cloud properties. This approach helps reducing the overestimation of aerosol optical thickness in cloud contaminated pixels. The surface reflectance product is delivered both for cloud-free and cloudy observations.  </p><p> </p><p>Global maps obtained from the processing of S3A/SLSTR observations will be shown. The SLSTR/CISAR products over events such as, for instance, the Australian fire in the last months of 2019, will be discussed in terms of aerosol optical thickness, aerosol-cloud discrimination and fine/coarse mode fraction.</p>

scholarly journals Retrieval of Ice Cloud Properties from AIRS and MODIS Observations Based on a Fast High-Spectral-Resolution Radiative Transfer Model

2013 ◽  
Vol 52 (3) ◽  
pp. 710-726 ◽  
Author(s):  
Chenxi Wang ◽  
Ping Yang ◽  
Steven Platnick ◽  
Andrew K. Heidinger ◽  
Bryan A. Baum ◽  
...  
Keyword(s):  
Radiative Transfer ◽  
Spectral Resolution ◽  
Spectral Response ◽  
Thermal Infrared ◽  
High Spectral Resolution ◽  
Radiative Transfer Model ◽  
Transfer Model ◽  
Retrieval Method ◽  
Ice Cloud ◽  
Cloud Properties

AbstractA computationally efficient high-spectral-resolution cloudy-sky radiative transfer model (HRTM) in the thermal infrared region (700–1300 cm−1, 0.1 cm−1 spectral resolution) is advanced for simulating the upwelling radiance at the top of atmosphere and for retrieving cloud properties. A precomputed transmittance database is generated for simulating the absorption contributed by up to seven major atmospheric absorptive gases (H2O, CO2, O3, O2, CH4, CO, and N2O) by using a rigorous line-by-line radiative transfer model (LBLRTM). Both the line absorption of individual gases and continuum absorption are included in the database. A high-spectral-resolution ice particle bulk scattering properties database is employed to simulate the radiation transfer within a vertically nonisothermal ice cloud layer. Inherent to HRTM are sensor spectral response functions that couple with high-spectral-resolution measurements in the thermal infrared regions from instruments such as the Atmospheric Infrared Sounder (AIRS) and Infrared Atmospheric Sounding Interferometer. When compared with the LBLRTM and the discrete ordinates radiative transfer model (DISORT), the root-mean-square error of HRTM-simulated single-layer cloud brightness temperatures in the thermal infrared window region is generally smaller than 0.2 K. An ice cloud optical property retrieval scheme is developed using collocated AIRS and Moderate Resolution Imaging Spectroradiometer (MODIS) data. A retrieval method is proposed to take advantage of the high-spectral-resolution instrument. On the basis of the forward model and retrieval method, a case study is presented for the simultaneous retrieval of ice cloud optical thickness τ and effective particle size Deff that includes a cloud-top-altitude self-adjustment approach to improve consistency with simulations.

scholarly journals Assessment of cloud properties from the reanalysis with satellite observations over East Asia

2019 ◽  
Author(s):  
Bin Yao ◽  
Chao Liu ◽  
Yan Yin ◽  
Zhiquan Liu ◽  
Chunxiang Shi ◽  
...  
Keyword(s):  
Radiative Transfer ◽  
East Asia ◽  
Deep Convection ◽  
Satellite Observation ◽  
Reanalysis Data ◽  
Satellite Observations ◽  
Radiative Transfer Model ◽  
Transfer Model ◽  
China Meteorological Administration ◽  
Cloud Properties

Abstract. Extensive observational and numerical investigations have been performed to better characterize cloud properties. However, due to the large variations of cloud spatiotemporal distributions and physical properties, quantitative depictions of clouds in different atmospheric reanalysis datasets are still highly uncertain, and cloud parameters in the models to produce those datasets remain largely unconstrained. A radiance-based evaluation approach is introduced and performed to assess the quality of cloud properties by directly comparing reanalysis-driven forward radiative transfer results with radiances from satellite observation. The newly developed China Meteorological Administration Reanalysis data (CRA), the ECMWF’s Fifth-generation Reanalysis (ERA5), and the Modern-Era Retrospective Analysis for Applications, Version 2 (MERRA-2) are considered in the present study. To avoid the unrealistic assumptions and uncertainties on satellite retrieval algorithms and products, the radiative transfer model (RTM) is used as a bridge to “translate” the reanalysis to corresponding satellite observations. The simulated reflectance and brightness temperatures (BTs) are directly compared with observations from the Advanced Himawari Imager (AHI) onboard the Himawari-8 satellite in the region from 80° E to 160° W between 60° N and 60° S, especially for results over East Asia. Comparisons of the reflectance in the solar and BTs in the infrared (IR) window channels reveal that CRA reanalysis better represents the total cloud cover than the other two reanalysis datasets. The simulated BTs for CRA and ERA5 are close to each other in many pixels, whereas the vertical distributions of cloud properties are significantly different, and ERA5 depicts a better deep convection structure than CRA reanalysis. Comparisons of the BT differences (BTDs) between the simulations and observations suggest that the water clouds are generally overestimated in ERA5 and MERRA-2, whereas the ice cloud is responsible for the overestimation over the center of cyclones in ERA5. Overall, the cloud from CRA, ERA5, and MERRA-2 show their own advantages in different aspects. The ERA5 reanalysis is found the most capability in representing the cloudy atmosphere over East Asia, and the results in CRA are close to those in ERA5.

scholarly journals Simulated reflectance above snow constrained by airborne measurements of solar radiation: implications for the snow grain morphology in the Arctic

2021 ◽  
Vol 14 (1) ◽  
pp. 369-389
Author(s):  
Soheila Jafariserajehlou ◽  
Vladimir V. Rozanov ◽  
Marco Vountas ◽  
Charles K. Gatebe ◽  
John P. Burrows
Keyword(s):  
Radiative Transfer ◽  
Grain Morphology ◽  
Effective Radius ◽  
Ice Crystals ◽  
Radiative Transfer Model ◽  
Transfer Model ◽  
Snow Layer ◽  
Airborne Measurements ◽  
Size And Shape ◽  
Atmosphere System

Abstract. Accurate knowledge of the reflectance from snow/ice-covered surfaces is of fundamental importance for the retrieval of snow parameters and atmospheric constituents from space-based and airborne observations. In this paper, we simulate the reflectance in a snow–atmosphere system, using the phenomenological radiative transfer model SCIATRAN, and compare the results with that of airborne measurements. To minimize the differences between measurements and simulation, we determine and employ the key atmospheric and surface parameters, such as snow grain morphologies (or habits). First, we report on a sensitivity study. This addresses the requirement for adequate a priori knowledge about snow models and ancillary information about the atmosphere. For this aim, we use the well-validated phenomenological radiative transfer model, SCIATRAN. Second, we present and apply a two-stage snow grain morphology (i.e., size and shape of ice crystals in the snow) retrieval algorithm. We then describe the use of this new retrieval for estimating the most representative snow model, using different types of snow morphologies, for the airborne observation conditions performed by NASA's Cloud Absorption Radiometer (CAR). Third, we present a comprehensive comparison of the simulated reflectance (using retrieved snow grain size and shape and independent atmospheric data) with that from airborne CAR measurements in the visible (0.670 µm) and near infrared (NIR; 0.870 and 1.6 µm) wavelength range. The results of this comparison are used to assess the quality and accuracy of the radiative transfer model in the simulation of the reflectance in a coupled snow–atmosphere system. Assuming that the snow layer consists of ice crystals with aggregates of eight column ice habit and having an effective radius of ∼99 µm, we find that, for a surface covered by old snow, the Pearson correlation coefficient, R, between measurements and simulations is 0.98 (R2∼0.96). For freshly fallen snow, assuming that the snow layer consists of the aggregate of five plates ice habit with an effective radius of ∼83 µm and having surface inhomogeneity, the correlation is ∼0.97 (R2∼0.94) in the infrared and 0.88 (R2∼0.77) in the visible wavelengths. The largest differences between simulated and measured values are observed in the glint area (i.e., in the angular regions of specular and near-specular reflection), with relative azimuth angles <±40∘ in the forward-scattering direction. The absolute difference between the modeled results and measurements in off-glint regions, with a viewing zenith angle of less than 50∘, is generally small ∼±0.025 and does not exceed ±0.05. These results will help to improve the calculation of snow surface reflectance and relevant assumptions in the snow–atmosphere system algorithms (e.g., aerosol optical thickness retrieval algorithms in the polar regions).

AMSU-B Observations of Mixed-Phase Clouds over Land

2005 ◽  
Vol 44 (1) ◽  
pp. 72-85 ◽  
Author(s):  
M. N. Deeter ◽  
J. Vivekanandan
Keyword(s):  
Water Vapor ◽  
Radiative Transfer ◽  
Mixed Phase ◽  
Radiative Transfer Model ◽  
Transfer Model ◽  
Radiative Effects ◽  
Liquid Water Path ◽  
Water Path ◽  
Cloud Properties ◽  
Mixed Phase Clouds

Abstract Measurements from passive microwave satellite instruments such as the Advanced Microwave Sounding Unit B (AMSU-B) are sensitive to both liquid and ice cloud particles. Radiative transfer modeling is exploited to simulate the response of the AMSU-B instrument to mixed-phase clouds over land. The plane-parallel radiative transfer model employed for the study accounts for scattering and absorption from cloud ice as well as absorption and emission from trace gases and cloud liquid. The radiative effects of mixed-phase clouds on AMSU-B window channels (i.e., 89 and 150 GHz) and water vapor line channels (i.e., 183 ± 1, 3, and 7 GHz) are studied. Sensitivities to noncloud parameters, including surface temperature, surface emissivity, and atmospheric temperature and water vapor profiles, are also quantified. Modeling results indicate that both cloud phases generally have significant radiative effects and that the 150- and 183 ± 7-GHz channels are typically the most sensitive channels to integrated cloud properties (i.e., liquid water path and ice water path). However, results also indicate that AMSU-B measurements alone are probably insufficient for retrieving all mixed-phase cloud properties of interest. These results are supported by comparisons of AMSU-B observations of a mixed-phase cloud over the Atmospheric Radiation Measurement (ARM) Program’s Southern Great Plains (SGP) site with corresponding calculated clear-sky values.

Using deep learning to emulate and accelerate a radiative-transfer model

2021 ◽  
Author(s):  
Ryan Lagerquist ◽  
David Turner ◽  
Imme Ebert-Uphoff ◽  
Jebb Stewart ◽  
Venita Hagerty
Keyword(s):  
Deep Learning ◽  
Radiative Transfer ◽  
Heating Rate ◽  
Weather Prediction ◽  
Computing Time ◽  
Single Layer ◽  
Radiative Heating ◽  
Radiative Transfer Model ◽  
Transfer Model ◽  
Heating Rates

AbstractThis paper describes the development of U-net++ models, a type of neural network that performs deep learning, to emulate the shortwave Rapid Radiative-transfer Model (RRTM). The goal is to emulate the RRTM accurately in a small fraction of the computing time, creating a U-net++ that could be used as a parameterization in numerical weather prediction (NWP). Target variables are surface downwelling flux, top-of-atmosphere upwelling flux (), net flux, and a profile of radiative-heating rates. We have devised several ways to make the U-net++ models knowledge-guided, recently identified as a key priority in machine learning (ML) applications to the geosciences. We conduct two experiments to find the best U-net++ configurations. In Experiment 1, we train on non-tropical sites and test on tropical sites, to assess extreme spatial generalization. In Experiment 2, we train on sites from all regions and test on different sites from all regions, with the goal of creating the best possible model for use in NWP. The selected model from Experiment 1 shows impressive skill on the tropical testing sites, except four notable deficiencies: large bias and error for heating rate in the upper stratosphere, unreliable for profiles with single-layer liquid cloud, large heating-rate bias in the mid-troposphere for profiles with multi-layer liquid cloud, and negative bias at lowzenith angles for all flux components and tropospheric heating rates. The selected model from Experiment 2 corrects all but the first deficiency, and both models run ~104 times faster than the RRTM. Our code is available publicly.

scholarly journals Numerical Experiments of an Advanced Radiative Transfer Model in the U.S. Navy Operational Global Atmospheric Prediction System

2012 ◽  
Vol 51 (3) ◽  
pp. 554-570 ◽  
Author(s):  
Ming Liu ◽  
Young-Joon Kim ◽  
Qingyun Zhao
Keyword(s):  
Radiative Transfer ◽  
Effective Radius ◽  
Radiative Transfer Model ◽  
Transfer Model ◽  
Forecast Errors ◽  
Prediction System ◽  
Surface Data ◽  
Ice Water ◽  
The Impact ◽  
Mean Square Errors

AbstractA high-order accurate radiative transfer (RT) model developed by Fu and Liou has been implemented into the Navy Operational Global Atmospheric Prediction System (NOGAPS) to improve the energy budget and forecast skill. The Fu–Liou RT model is a four-stream algorithm (with a two-stream option) integrating over 6 shortwave bands and 12 longwave bands. The experimental 10-day forecasts and analyses from data assimilation cycles are compared with the operational output, which uses a two-stream RT model of three shortwave and five longwave bands, for both winter and summer periods. The verifications against observations of radiosonde and surface data show that the new RT model increases temperature accuracy in both forecasts and analyses by reducing mean bias and root-mean-square errors globally. In addition, the forecast errors also grow more slowly in time than those of the operational NOGAPS because of accumulated effects of more accurate cloud–radiation interactions. The impact of parameterized cloud effective radius in estimating liquid and ice water optical properties is also investigated through a sensitivity test by comparing with the cases using constant cloud effective radius to examine the temperature changes in response to cloud scattering and absorption. The parameterization approach is demonstrated to outperform that of constant radius by showing smaller errors and better matches to observations. This suggests the superiority of the new RT model relative to its operational counterpart, which does not use cloud effective radius. An effort has also been made to improve the computational efficiency of the new RT model for operational applications.

Estimation of Surface and Top-of-Atmosphere Shortwave Irradiance in Biomass-Burning Regions during SCAR-B

2000 ◽  
Vol 39 (10) ◽  
pp. 1742-1753 ◽  
Author(s):  
Sundar A. Christopher ◽  
Xiang Li ◽  
Ronald M. Welch ◽  
Jeffrey S. Reid ◽  
Peter V. Hobbs ◽  
...  
Keyword(s):  
Radiative Transfer ◽  
Biomass Burning ◽  
Optical Thickness ◽  
Radiative Forcing ◽  
Wavelength Dependence ◽  
Single Scattering ◽  
Radiative Transfer Model ◽  
Transfer Model ◽  
Aerosol Radiative Forcing ◽  
Mean Square Errors

Abstract Using in situ measurements of aerosol optical properties and ground-based measurements of aerosol optical thickness (τs) during the Smoke, Clouds and Radiation—Brazil (SCAR-B) experiment, a four-stream broadband radiative transfer model is used to estimate the downward shortwave irradiance (DSWI) and top-of-atmosphere (TOA) shortwave aerosol radiative forcing (SWARF) in cloud-free regions dominated by smoke from biomass burning in Brazil. The calculated DSWI values are compared with broadband pyranometer measurements made at the surface. The results show that, for two days when near-coincident measurements of single-scattering albedo ω0 and τs are available, the root-mean-square errors between the measured and calculated DSWI for daytime data are within 30 W m−2. For five days during SCAR-B, however, when assumptions about ω0 have to be made and also when τs was significantly higher, the differences can be as large as 100 W m−2. At TOA, the SWARF per unit optical thickness ranges from −20 to −60 W m−2 over four major ecosystems in South America. The results show that τs and ω0 are the two most important parameters that affect DSWI calculations. For SWARF values, surface albedos also play an important role. It is shown that ω0 must be known within 0.05 and τs at 0.55 μm must be known to within 0.1 to estimate DSWI to within 20 W m−2. The methodology described in this paper could serve as a potential strategy for determining DSWI values in the presence of aerosols. The wavelength dependence of τs and ω0 over the entire shortwave spectrum is needed to improve radiative transfer calculations. If global retrievals of DSWI and SWARF from satellite measurements are to be performed in the presence of biomass-burning aerosols on a routine basis, a concerted effort should be made to develop methodologies for estimating ω0 and τs from satellite and ground-based measurements.

Sign in / Sign up

Export Citation Format

Share Document