scholarly journals Identifying Aerosol Subtypes from CALIPSO Lidar Profiles Using Deep Machine Learning

Atmosphere ◽  
2020 ◽  
Vol 12 (1) ◽  
pp. 10
Author(s):  
Shan Zeng ◽  
Ali Omar ◽  
Mark Vaughan ◽  
Macarena Ortiz ◽  
Charles Trepte ◽  
...  
Keyword(s):  
Machine Learning ◽  
Semantic Segmentation ◽  
Retrieval Algorithm ◽  
Orthogonal Polarization ◽  
Ice Clouds ◽  
Microphysical Properties ◽  
Texture Information ◽  
Vertical Distributions ◽  
Aerosol Classification ◽  
Aerosol Types

The Cloud–Aerosol Lidar with Orthogonal Polarization (CALIOP), on-board the Cloud–Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) platform, is an elastic backscatter lidar that has been providing vertical profiles of the spatial, optical, and microphysical properties of clouds and aerosols since June 2006. Distinguishing between feature types (i.e., clouds vs. aerosol) and subtypes (e.g., ice clouds vs. water clouds and dust aerosols from smoke) in the CALIOP measurements is currently accomplished using layer-integrated measurements acquired by co-polarized (parallel) and cross-polarized (perpendicular) 532 nm channels and a single 1064 nm channel. Newly developed deep machine learning (DML) semantic segmentation methods now have the ability to combine observations from multiple channels with texture information to recognize patterns in data. Instead of focusing on a limited set of layer integrated values, our new DML feature classification technique uses the full scope of range-resolved information available in the CALIOP attenuated backscatter profiles. In this paper, one of the convolutional neural networks (CNN), SegNet, a fast and efficient DML model, is used to distinguish aerosol subtypes directly from the CALIOP profiles. The DML method is a 2D range bin-to-range bin aerosol subtype classification algorithm. We compare our new DML results to the classifications generated by CALIOP’s 1D layer-to-layer operational retrieval algorithm. These two methods, which take distinctly different approaches to aerosol classification, agree in over 60% of the comparisons. Higher levels of agreement are found in homogeneous scenes containing only a single aerosol type (i.e., marine, stratospheric aerosols). Disagreement between the two techniques increases in regions containing mixture of different aerosol types. The multi-dimensional texture information leveraged by the DML method shows advantages in differentiating between aerosol types based on their classification scores, as well as in distinguishing vertical distributions of aerosol types within individual layers. However, untangling mixtures of aerosol subtypes is still challenging for both the DML and operational algorithms.

scholarly journals Thermodynamic phase retrieval of convective clouds: impact of sensor viewing geometry and vertical distribution of cloud properties

2013 ◽  
Vol 6 (3) ◽  
pp. 539-547 ◽  
Author(s):  
E. Jäkel ◽  
J. Walter ◽  
M. Wendisch
Keyword(s):  
Phase Retrieval ◽  
Near Infrared ◽  
Three Dimensional ◽  
Ice Clouds ◽  
Convective Clouds ◽  
Microphysical Properties ◽  
Cloud Properties ◽  
Reflectivity Spectra ◽  
Vertical Distributions ◽  
Thermodynamic Phase

Abstract. The sensitivity of passive remote sensing measurements to retrieve microphysical parameters of convective clouds, in particular their thermodynamic phase, is investigated by three-dimensional (3-D) radiative transfer simulations. The effects of different viewing geometries and vertical distributions of the cloud microphysical properties are investigated. Measurement examples of spectral solar radiance reflected by cloud sides (passive) in the near-infrared (NIR) spectral range are performed together with collocated lidar observations (active). The retrieval method to distinguish the cloud thermodynamic phase (liquid water or ice) exploits different slopes of cloud side reflectivity spectra of water and ice clouds in the NIR. The concurrent depolarization backscattering lidar provides geometry information about the cloud distance and height as well as the depolarization.

Aerosol classification using sun photometer and lidar data within a machine learning algorithm-a possible nowcasting application

2021 ◽  
Author(s):  
Joelle Buxmann ◽  
Martin Osborne ◽  
Mike Protts ◽  
Debbie O'Sullivan
Keyword(s):  
Machine Learning ◽  
Biomass Burning ◽  
Machine Learning Algorithms ◽  
Supervised Machine Learning ◽  
Training Set ◽  
Angstrom Exponent ◽  
Ångström Exponent ◽  
Aerosol Classification ◽  
Aerosol Types ◽  
Ångstrom Exponent

<p>The Met Office operates a ground based operational network of nine polarisation Raman lidars (aerosol profiling instruments) and sun photometers (column integrated information). An aerosol classification scheme using supervised machine learning has been developed. The concept of Mahalanobis (~normalized) distance to identify the aerosol type  from individual Aerosol Robotic Network (AERONET) measurements including Extinction Angstrom Exponent, Absorption Angstrom Exponent, Single Scattering Albedo and Index of refraction is used for a subset of AERONET stations around the globe of known main aerosol types (training set). The aerosol types  include maritime, urban industrial, biomass burning and dust. We build a predictive model from this training set using K nearest neighbour machine learning algorithms. The relation of particle polarisation ratio and lidar ratio from the Raman lidar is used as a sanity check.  We apply the model to 3- 4 years of AERONET and profiling data across the UK, with instruments evenly distributed across the country, from Camborne in Cornwall to Lerwick in the Shetland Islands. We are showing more detailed data of a dust event in May 2016, dust/biomass burning aerosol mix from October 2017 (hurricane Ophelia) and more recent aerosol transported from the Canadian wild fires in September 2020. AERONET Level 2.0  data is compared to level 1.5 in order to determine the implications for the aerosol classification. Level 1.5 data are cloud-screened, but not quality assured and may not have the final calibration applied. Level 2.0  data have pre- and post-field calibration applied, are cloud-screened, and quality-assured data. As level 2.0 data is usually only available after 1-2 years (after a new calibration has been performed), it is important to understand the  usefulness of more readily available level 1.5 (cloud screened) data.</p><p>The aim is to build a real time aerosol classification application that can be used in Nowcasting.</p>

scholarly journals Aerosol classification from airborne HSRL and comparisons with the CALIPSO vertical feature mask

2013 ◽  
Vol 6 (5) ◽  
pp. 1397-1412 ◽  
Author(s):  
S. P. Burton ◽  
R. A. Ferrare ◽  
M. A. Vaughan ◽  
A. H. Omar ◽  
R. R. Rogers ◽  
...  
Keyword(s):  
High Spectral Resolution ◽  
Orthogonal Polarization ◽  
Dependent Observations ◽  
Direct Measurements ◽  
Aerosol Classification ◽  
Lidar Ratio ◽  
Langley Research Center ◽  
High Spectral Resolution Lidar ◽  
Aerosol Types ◽  
Aerosol Type

Abstract. Aerosol classification products from the NASA Langley Research Center (LaRC) airborne High Spectral Resolution Lidar (HSRL-1) on the NASA B200 aircraft are compared with coincident V3.01 aerosol classification products from the Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) instrument on the CALIPSO satellite. For CALIOP, aerosol classification is a key input to the aerosol retrieval, and must be inferred using aerosol loading-dependent observations and location information. In contrast, HSRL-1 makes direct measurements of aerosol intensive properties, including the lidar ratio, that provide information on aerosol type. In this study, comparisons are made for 109 underflights of the CALIOP orbit track. We find that 62% of the CALIOP marine layers and 54% of the polluted continental layers agree with HSRL-1 classification results. In addition, 80% of the CALIOP desert dust layers are classified as either dust or dusty mix byHSRL-1. However, agreement is less for CALIOP smoke (13%) and polluted dust (35%) layers. Specific case studies are examined, giving insight into the performance of the CALIOP aerosol type algorithm. In particular, we find that the CALIOP polluted dust type is overused due to an attenuation-related depolarization bias. Furthermore, the polluted dust type frequently includes mixtures of dust plus marine aerosol. Finally, we find that CALIOP's identification of internal boundaries between different aerosol types in contact with each other frequently do not reflect the actual transitions between aerosol types accurately. Based on these findings, we give recommendations which may help to improve the CALIOP aerosol type algorithms.

scholarly journals Satellite-Based Aerosol Classification for Capital Cities in Asia Using a Random Forest Model

2021 ◽  
Vol 13 (13) ◽  
pp. 2464
Author(s):  
Wonei Choi ◽  
Hyeongwoo Kang ◽  
Dongho Shin ◽  
Hanlim Lee
Keyword(s):  
Random Forest ◽  
Classification Model ◽  
Seasonal Effects ◽  
Classification Models ◽  
Capital Cities ◽  
Microphysical Properties ◽  
Aerosol Classification ◽  
Absorbing Aerosols ◽  
Aerosol Types ◽  
Aerosol Type

Aerosol types in Asian capital cities were classified using a random forest (RF) satellite-based aerosol classification model during 2018–2020 in an investigation of the contributions of aerosol types, with or without Aerosol Robotic Network (AERONET) observations. In this study, we used the recently developed RF aerosol classification model to detect and classify aerosols into four types: pure dust, dust-dominated aerosols, strongly absorbing aerosols, and non-absorbing aerosols. Aerosol optical and microphysical properties for each aerosol type detected by the RF model were found to be reasonably consistent with those for typical aerosol types. In Asian capital cities, pollution-sourced aerosols, especially non-absorbing aerosols, were found to predominate, although Asian cities also tend to be seasonally affected by natural dust aerosols, particularly in East Asia (March–May) and South Asia (March–August). No specific seasonal effects on aerosol type were detected in Southeast Asia, where there was a predominance of non-absorbing aerosols. The aerosol types detected by the RF model were compared with those identified by other aerosol classification models. This study indicates that the satellite-based RF model may be used as an alternative in the absence of AERONET sites or observations.

scholarly journals A First Approach to Aerosol Classification Using Space-Borne Measurement Data: Machine Learning-Based Algorithm and Evaluation

2021 ◽  
Vol 13 (4) ◽  
pp. 609
Author(s):  
Wonei Choi ◽  
Hanlim Lee ◽  
Jeonghyeon Park
Keyword(s):  
Machine Learning ◽  
Measurement Data ◽  
Wavelength Dependence ◽  
Single Scattering ◽  
New Method ◽  
Spherical Particles ◽  
Training Dataset ◽  
Aerosol Classification ◽  
Input Variables ◽  
Aerosol Types

A new method was developed for classifying aerosol types involving a machine-learning approach to the use of satellite data. An Aerosol Robotic NETwork (AERONET)-based aerosol-type dataset was used as a target variable in a random forest (RF) model. The contributions of satellite input variables to the RF-based model were quantified to determine an optimal set of input variables. The new method, based on inputs of satellite variables, allows the classification of seven aerosol types: pure dust, dust-dominant mixed, pollution-dominant mixed aerosols, and pollution aerosols (strongly, moderately, weakly, and non-absorbing). The performance of the model was statistically evaluated using AERONET data excluded from the model training dataset. Model accuracy for classifying the seven aerosol types was 59%, improving to 72% for four types (pure dust, dust-dominant mixed, strongly absorbing, and non-absorbing). The performance of the model was evaluated against an earlier aerosol classification method based on the wavelength dependence of single-scattering albedo (SSA) and fine-mode-fraction values from AERONET. Typical wavelength dependences of SSA for individual aerosol types are consistent with those obtained for aerosol types by the new method. This study demonstrates that an RF-based model is capable of satellite aerosol classification with sensitivity to the contribution of non-spherical particles.

scholarly journals The impact of the microphysical properties of aerosol on the atmospheric correction of hyperspectral data in coastal waters

2015 ◽  
Vol 8 (3) ◽  
pp. 1593-1604 ◽  
Author(s):  
C. Bassani ◽  
C. Manzo ◽  
F. Braga ◽  
M. Bresciani ◽  
C. Giardino ◽  
...  
Keyword(s):  
Accuracy Assessment ◽  
Atmospheric Correction ◽  
Solar Spectrum ◽  
Hyperspectral Data ◽  
High Spectral Resolution ◽  
Quality Analysis ◽  
Microphysical Properties ◽  
Quantitative Remote Sensing ◽  
The Impact ◽  
Aerosol Types

Abstract. Hyperspectral imaging provides quantitative remote sensing of ocean colour by the high spectral resolution of the water features. The HICO™ (Hyperspectral Imager for the Coastal Ocean) is suitable for coastal studies and monitoring. The accurate retrieval of hyperspectral water-leaving reflectance from HICO™ data is still a challenge. The aim of this work is to retrieve the water-leaving reflectance from HICO™ data with a physically based algorithm, using the local microphysical properties of the aerosol in order to overcome the limitations of the standard aerosol types commonly used in atmospheric correction processing. The water-leaving reflectance was obtained using the HICO@CRI (HICO ATmospherically Corrected Reflectance Imagery) atmospheric correction algorithm by adapting the vector version of the Second Simulation of a Satellite Signal in the Solar Spectrum (6SV) radiative transfer code. The HICO@CRI algorithm was applied on to six HICO™ images acquired in the northern Mediterranean basin, using the microphysical properties measured by the Acqua Alta Oceanographic Tower (AAOT) AERONET site. The HICO@CRI results obtained with AERONET products were validated with in situ measurements showing an accuracy expressed by r2 = 0.98. Additional runs of HICO@CRI on the six images were performed using maritime, continental and urban standard aerosol types to perform the accuracy assessment when standard aerosol types implemented in 6SV are used. The results highlight that the microphysical properties of the aerosol improve the accuracy of the atmospheric correction compared to standard aerosol types. The normalized root mean square (NRMSE) and the similar spectral value (SSV) of the water-leaving reflectance show reduced accuracy in atmospheric correction results when there is an increase in aerosol loading. This is mainly when the standard aerosol type used is characterized with different optical properties compared to the local aerosol. The results suggest that if a water quality analysis is needed the microphysical properties of the aerosol need to be taken into consideration in the atmospheric correction of hyperspectral data over coastal environments, because aerosols influence the accuracy of the retrieved water-leaving reflectance.

scholarly journals Semantic segmentation of PolSAR image data using advanced deep learning model

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Rajat Garg ◽  
Anil Kumar ◽  
Nikunj Bansal ◽  
Manish Prateek ◽  
Shashi Kumar
Keyword(s):  
Machine Learning ◽  
Remote Sensing ◽  
Deep Learning ◽  
Urban Area ◽  
Urban Areas ◽  
Learning Algorithms ◽  
Semantic Segmentation ◽  
Learning Model ◽  
Machine Learning Algorithms ◽  
Deep Learning Model

AbstractUrban area mapping is an important application of remote sensing which aims at both estimation and change in land cover under the urban area. A major challenge being faced while analyzing Synthetic Aperture Radar (SAR) based remote sensing data is that there is a lot of similarity between highly vegetated urban areas and oriented urban targets with that of actual vegetation. This similarity between some urban areas and vegetation leads to misclassification of the urban area into forest cover. The present work is a precursor study for the dual-frequency L and S-band NASA-ISRO Synthetic Aperture Radar (NISAR) mission and aims at minimizing the misclassification of such highly vegetated and oriented urban targets into vegetation class with the help of deep learning. In this study, three machine learning algorithms Random Forest (RF), K-Nearest Neighbour (KNN), and Support Vector Machine (SVM) have been implemented along with a deep learning model DeepLabv3+ for semantic segmentation of Polarimetric SAR (PolSAR) data. It is a general perception that a large dataset is required for the successful implementation of any deep learning model but in the field of SAR based remote sensing, a major issue is the unavailability of a large benchmark labeled dataset for the implementation of deep learning algorithms from scratch. In current work, it has been shown that a pre-trained deep learning model DeepLabv3+ outperforms the machine learning algorithms for land use and land cover (LULC) classification task even with a small dataset using transfer learning. The highest pixel accuracy of 87.78% and overall pixel accuracy of 85.65% have been achieved with DeepLabv3+ and Random Forest performs best among the machine learning algorithms with overall pixel accuracy of 77.91% while SVM and KNN trail with an overall accuracy of 77.01% and 76.47% respectively. The highest precision of 0.9228 is recorded for the urban class for semantic segmentation task with DeepLabv3+ while machine learning algorithms SVM and RF gave comparable results with a precision of 0.8977 and 0.8958 respectively.

Sign in / Sign up

Export Citation Format

Share Document