scholarly journals Neural Network Reflectance Prediction Model for Both Open Ocean and Coastal Waters

2020 ◽  
Vol 12 (9) ◽  
pp. 1421
Author(s):  
Lipi Mukherjee ◽  
Peng-Wang Zhai ◽  
Meng Gao ◽  
Yongxiang Hu ◽  
Bryan A. Franz ◽  
...  
Keyword(s):  
Neural Network ◽  
Remote Sensing ◽  
Radiative Transfer ◽  
Prediction Model ◽  
Open Ocean ◽  
Machine Learning Techniques ◽  
Global Ocean ◽  
Retrieval Algorithm ◽  
Test Cases ◽  
Polarized Reflectance

Remote sensing of global ocean color is a valuable tool for understanding the ecology and biogeochemistry of the worlds oceans, and provides critical input to our knowledge of the global carbon cycle and the impacts of climate change. Ocean polarized reflectance contains information about the constituents of the upper ocean euphotic zone, such as colored dissolved organic matter (CDOM), sediments, phytoplankton, and pollutants. In order to retrieve the information on these constituents, remote sensing algorithms typically rely on radiative transfer models to interpret water color or remote-sensing reflectance; however, this can be resource-prohibitive for operational use due to the extensive CPU time involved in radiative transfer solutions. In this work, we report a fast model based on machine learning techniques, called Neural Network Reflectance Prediction Model (NNRPM), which can be used to predict ocean bidirectional polarized reflectance given inherent optical properties of ocean waters. This supervised model is trained using a large volume of data derived from radiative transfer simulations for coupled atmosphere and ocean systems using the successive order of scattering technique (SOS-CAOS). The performance of the model is validated against another large independent test dataset generated from SOS-CAOS. The model is able to predict both polarized and unpolarized reflectances with an absolute error (AE) less than 0.004 for 99% of test cases. We have also shown that the degree of linear polarization (DoLP) for unpolarized incident light can be predicted with an AE less than 0.002 for 99% of test cases. In general, the simulation time of SOS-CAOS depends on optical depth, and required accuracy. When comparing the average speeds of the NNRPM against the SOS-CAOS model for the same parameters, we see that the NNRPM is able to predict the Ocean BRDF 6000 times faster than SOS-CAOS. Both ultraviolet and visible wavelengths are included in the model to help differentiate between dissolved organic material and chlorophyll in the study of the open ocean and the coastal zone. The incorporation of this model into the retrieval algorithm will make the retrieval process more efficient, and thus applicable for operational use with global satellite observations.

scholarly journals OPEN: A New Estimation of Global Ocean Heat Content for Upper 2000 Meters from Remote Sensing Data

2020 ◽  
Vol 12 (14) ◽  
pp. 2294
Author(s):  
Hua Su ◽  
Haojie Zhang ◽  
Xupu Geng ◽  
Tian Qin ◽  
Wenfang Lu ◽  
...  
Keyword(s):  
Neural Network ◽  
Remote Sensing ◽  
Network Architecture ◽  
Climate Changes ◽  
Heat Content ◽  
Remote Sensing Data ◽  
Ocean Heat Content ◽  
Global Ocean ◽  
Coefficient Of Determination ◽  
Sensing Data

Retrieving information concerning the interior of the ocean using satellite remote sensing data has a major impact on studies of ocean dynamic and climate changes; however, the lack of information within the ocean limits such studies about the global ocean. In this paper, an artificial neural network, combined with satellite data and gridded Argo product, is used to estimate the ocean heat content (OHC) anomalies over four different depths down to 2000 m covering the near-global ocean, excluding the polar regions. Our method allows for the temporal hindcast of the OHC to other periods beyond the 2005–2018 training period. By applying an ensemble technique, the hindcasting uncertainty could also be estimated by using different 9-year periods for training and then calculating the standard deviation across six ensemble members. This new OHC product is called the Ocean Projection and Extension neural Network (OPEN) product. The accuracy of the product is accessed using the coefficient of determination (R2) and the relative root-mean-square error (RRMSE). The feature combinations and network architecture are optimized via a series of experiments. Overall, intercomparison with several routinely analyzed OHC products shows that the OPEN OHC has an R2 larger than 0.95 and an RRMSE of <0.20 and presents notably accurate trends and variabilities. The OPEN product can therefore provide a valuable complement for studies of global climate changes.

scholarly journals Testing remote sensing on artificial observations: impact of drizzle and 3-D cloud structure on effective radius retrievals

2010 ◽  
Vol 10 (19) ◽  
pp. 9535-9549 ◽  
Author(s):  
T. Zinner ◽  
G. Wind ◽  
S. Platnick ◽  
A. S. Ackerman
Keyword(s):  
Remote Sensing ◽  
Particle Size ◽  
Radiative Transfer ◽  
Droplet Size ◽  
Optical Thickness ◽  
Marine Boundary Layer ◽  
Effective Radius ◽  
Retrieval Algorithm ◽  
Moderate Resolution Imaging Spectroradiometer ◽  
The Impact

Abstract. Remote sensing of cloud effective particle size with passive sensors like the Moderate Resolution Imaging Spectroradiometer (MODIS) is an important tool for cloud microphysical studies. As a measure of the radiatively relevant droplet size, effective radius can be retrieved with different combinations of visible through shortwave and midwave infrared channels. In practice, retrieved effective radii from these combinations can be quite different. This difference is perhaps indicative of different penetration depths and path lengths for the spectral reflectances used. In addition, operational liquid water cloud retrievals are based on the assumption of a relatively narrow distribution of droplet sizes; the role of larger precipitation particles in these distributions is neglected. Therefore, possible explanations for the discrepancy in some MODIS spectral size retrievals could include 3-D radiative transport effects, including sub-pixel cloud inhomogeneity, and/or the impact of drizzle formation. For three cloud cases the possible factors of influence are isolated and investigated in detail by the use of simulated cloud scenes and synthetic satellite data: marine boundary layer cloud scenes from large eddy simulations (LES) with detailed microphysics are combined with Monte Carlo radiative transfer calculations that explicitly account for the detailed droplet size distributions as well as 3-D radiative transfer to simulate MODIS observations. The operational MODIS optical thickness and effective radius retrieval algorithm is applied to these and the results are compared to the given LES microphysics. We investigate two types of marine cloud situations each with and without drizzle from LES simulations: (1) a typical daytime stratocumulus deck at two times in the diurnal cycle and (2) one scene with scattered cumulus. Only small impact of drizzle formation on the retrieved domain average and on the differences between the three effective radius retrievals is noticed for both cloud scene types for different reasons. For our, presumably typical, overcast stratocumulus scenes with an optical thickness of 8 to 9 and rain rates at cloud bottom up to 0.05 mm/h clear drizzle impact on the retrievals can be excluded. The cumulus scene does not show much drizzle sensitivity either despite extended drizzle areas being directly visible from above (locally >1 mm/h), which is mainly due to technical characteristics of the standard retrieval approach. 3-D effects, on the other hand, produce large discrepancies between the 1.6 and 2.1 μm channel observations compared to 3.7 μm retrievals in the latter case. A general sensitivity of MODIS particle size data to drizzle formation is not corroborated by our case studies.

A Fast Polarized Microwave Radiative Transfer Model for a Raining Atmosphere

2010 ◽  
Author(s):  
K. Srinivasa Ramanujam ◽  
C. Balaji
Keyword(s):  
Neural Network ◽  
Remote Sensing ◽  
Microwave Radiation ◽  
Tropical Rainfall Measuring Mission ◽  
Forward Model ◽  
Retrieval Algorithm ◽  
Earth’S Atmosphere ◽  
Brightness Temperatures ◽  
Infra Red ◽  
Earth's Atmosphere

Retrieval of vertical rain structure and hence the estimation of surface rain rate is of central importance to various missions involving remote sensing of the earth’s atmosphere. Typically, remote sensing involves scanning the earth’s atmosphere at visible, infra red and microwave frequencies. While the visible and infra red frequencies can scan the atmosphere with higher spatial resolution, they are not suited for scanning under cloudy conditions as clouds are opaque under these frequencies. However, the longer wavelength microwave radiation can partially penetrate through the clouds without much attenuation thereby making it more suitable for meteorological purposes. The retrieval algorithms used for passive microwave remote sensing involve modeling of the radiation in the earth’s atmosphere where in the clouds and precipitating rain (also known as hydrometeors) emit / absorb / scatter. Additionally, it has been observed that the rain droplets tend to polarize the microwave signal emitted by the earth’s surface. In view of this, the first step in the development of a rainfall retrieval algorithm for any satellite mission is to simulate the radiances (also known as brightness temperatures) that would have been measured by a typical radiometer for different sensor frequencies and resolutions. Towards this, a polarized microwave radiation transfer code has been developed in house for a plane parallel raining atmosphere (henceforth called as forward model) that depicts the physics as seen by a satellite. Physics based retrieval algorithm often involves repeated execution of the forward model for various raining scenario. However, due to the complexity involved in the radiation modeling of the raining atmosphere which is participating in nature, the forward model suffers from the drawback that it requires enormous computational effort. In the present work, a much quicker alternative is proposed wherein the forward model can be replaced with an Artificial Neural Network (ANN) based Fast Forward Model (AFFM). This AFFM can be used in conjunction with an appropriate inverse technique to retrieve the rain structure. Spectral microwave brightness temperatures at frequencies corresponding to the Tropical Rainfall Measuring Mission (TRMM) of National Aeronautics and Space Administration (NASA) and Japan Aerospace Exploration Agency (JAXA) are first simulated using an in-house polarized radiate on transfer code for sixteen past cyclones in the North Indian Ocean region in the period (2000–2005), using the hydrometeor profiles retrieved from the Goddard Profiling Algorithm (GPROF) of the Tropical Rainfall Measuring Mission (TRMM)’s Microwave Imager (TMI). This data is split into two sets: while the first set of data is used for training the network, the remainder of the data is used for testing the ANN. The results obtained are very encouraging and shows that neural network is capable of predicting the brightness temperature accurately with the correlation coefficient of over 99%. Furthermore, the execution of the forward model on an Intel Core 2 Quad 3.0 GHz processor based, 8 GB DDR3 RAM workstation took 3 days, while the AFFM delivers the results in 10 seconds.

Deep Learning for the derivation of GNSS Reflectometry global ocean wind speed

2021 ◽  
Author(s):  
Milad Asgarimehr ◽  
Caroline Arnold ◽  
Felix Stiehler ◽  
Tobias Weigel ◽  
Chris Ruf ◽  
...  
Keyword(s):  
Machine Learning ◽  
Remote Sensing ◽  
Deep Learning ◽  
Wind Speed ◽  
Sampling Rate ◽  
Machine Learning Techniques ◽  
Global Ocean ◽  
Deep Model ◽  
Learning Techniques ◽  
Unseen Data

&lt;p&gt;The Global Navigation Satellite System Reflectometry (GNSS-R) is a novel remote sensing technique exploiting GNSS signals after reflection off the Earth's surface. The capability of spaceborne GNSS-R to monitor ocean state and the surface wind is recently well demonstrated, which offers an unprecedented sampling rate and much robustness during rainfall. The Cyclone GNSS (CyGNSS) is the first spaceborne mission fully dedicated to GNSS-R, launched in December 2016.&lt;/p&gt;&lt;p&gt;Thanks to the low development costs of the GNSS-R satellite missions as well as the capability of tracking multiple reflected signals from numerous GNSS transmitters, the GNSS-R datasets are much bigger compared to those from conventional remote sensing techniques. The CyGNSS provides a high number of unique samples in the order of a few millions monthly.&amp;#160; Deep learning can therefore be implemented in GNSS-R even more efficiently than other remote sensing domains. With the upcoming GNSS-R CubeSats, the data volume is expected to increase in the near future and GNSS-R &amp;#8220;Big data&amp;#8221; can be a future challenge. Deep learning methods are additionally able to correct the potential effects, both technical and geophysical, dictated by data empirically when the mechanisms are not well described by the theoretical knowledge. This poses the question if GNSS-R should embrace deep learning and can benefit from this modern data scientific method like other Earth Observation domains.&lt;/p&gt;&lt;p&gt;The receivers onboard CyGNSS cross-correlate the reflected signals received at a nadir antenna to a locally generated replica. The cross-correlation power at a range of the signal delay and Doppler frequency shift is the observational output of the receivers being called delay-Doppler Maps (DDMs). The mapped power is inversely proportional to the ocean roughness and consequently surface winds.&lt;/p&gt;&lt;p&gt;Few recent studies innovatively show some merits of machine learning techniques for the derivations of ocean winds from the DDMs. However, the capability of machine learning techniques, especially deep learning for an operational data derivation needs to be better characterized. Normally, the operational retrieval algorithms are developed based on an existing dataset and are supposed to operate on the upcoming measurements. Therefore, machine learning-based models are supposed to generalize well on the unseen data in future periods. Herein, we aim at the characterization of deep learning capabilities for these GNSS-R operational purposes.&lt;/p&gt;&lt;p&gt;In this interdisciplinary study, we present a deep learning algorithm processing the CyGNSS measurements to derive wind speed data. The model is supposed to meet an acceptable level of generalization on the upcoming unseen data, and alternatively can be used as an operational processing algorithm. We propose a deep model based on convolutional and fully connected layers processing the DDMs besides ancillary input features. The model leads to the so-far best quality of global wind speed estimates using GNSS-R measurements with a general root mean square error of 1.3 m/s over unseen data in a time span different from that of the training data.&lt;/p&gt;

scholarly journals A neural network radiative transfer model approach applied to the Tropospheric Monitoring Instrument aerosol height algorithm

2019 ◽  
Vol 12 (12) ◽  
pp. 6619-6634 ◽  
Author(s):  
Swadhin Nanda ◽  
Martin de Graaf ◽  
J. Pepijn Veefkind ◽  
Mark ter Linden ◽  
Maarten Sneep ◽  
...  
Keyword(s):  
Neural Network ◽  
Radiative Transfer ◽  
European Space Agency ◽  
Forward Model ◽  
Radiative Transfer Model ◽  
Retrieval Algorithm ◽  
Transfer Model ◽  
Aerosol Layer ◽  
Layer Height ◽  
Monitoring Instrument

Abstract. To retrieve aerosol properties from satellite measurements of the oxygen A-band in the near-infrared, a line-by-line radiative transfer model implementation requires a large number of calculations. These calculations severely restrict a retrieval algorithm's operational capability as it can take several minutes to retrieve the aerosol layer height for a single ground pixel. This paper proposes a forward modelling approach using artificial neural networks to speed up the retrieval algorithm. The forward model outputs are trained into a set of neural network models to completely replace line-by-line calculations in the operational processor. Results comparing the forward model to the neural network alternative show an encouraging outcome with good agreement between the two when they are applied to retrieval scenarios using both synthetic and real measured spectra from TROPOMI (TROPOspheric Monitoring Instrument) on board the European Space Agency (ESA) Sentinel-5 Precursor mission. With an enhancement of the computational speed by 3 orders of magnitude, TROPOMI's operational aerosol layer height processor is now able to retrieve aerosol layer heights well within operational capacity.

scholarly journals Physically Based Approach for Combined Atmospheric and Topographic Corrections

2019 ◽  
Vol 11 (10) ◽  
pp. 1218 ◽  
Author(s):  
Federico Santini ◽  
Angelo Palombo
Keyword(s):  
Remote Sensing ◽  
Radiative Transfer ◽  
Radiative Transfer Model ◽  
Transfer Model ◽  
Test Cases ◽  
Remote Sensors ◽  
Sensing Applications ◽  
Physically Based ◽  
Remote Sensing Applications ◽  
The Impact

The enhanced spectral and spatial resolutions of the remote sensors have increased the need for highly performing preprocessing procedures. In this paper, a comprehensive approach, which simultaneously performs atmospheric and topographic corrections and includes second order corrections such as adjacency effects, was presented. The method, developed under the assumption of Lambertian surfaces, is physically based and uses MODTRAN 4 radiative transfer model. The use of MODTRAN 4 for the estimates of the radiative quantities was widely discussed in the paper and the impact on remote sensing applications was shown through a series of test cases.

scholarly journals A neural network radiative transfer model approach applied to TROPOMI’s aerosol height algorithm

2019 ◽  
Author(s):  
Swadhin Nanda ◽  
Martin de Graaf ◽  
J. Pepijn Veefkind ◽  
Mark ter Linden ◽  
Maarten Sneep ◽  
...  
Keyword(s):  
Neural Network ◽  
Radiative Transfer ◽  
Near Infrared ◽  
Forward Model ◽  
Radiative Transfer Model ◽  
Retrieval Algorithm ◽  
Transfer Model ◽  
Aerosol Layer ◽  
Neural Network Models ◽  
Layer Height

Abstract. To retrieve aerosol properties from satellite measurements of the oxygen A-band in the near infrared, a line-by-line radiative transfer model implementation requires a large number of calculations. These calculations severely restrict a retrieval algorithm's operational capability as it can take several minutes to retrieve aerosol layer height for a single ground pixel. This paper proposes a forward modeling approach using artificial neural networks to speed up the retrieval algorithm. The forward model outputs are trained into a set of neural network models to completely replace line-by-line calculations in the operational processor. Results of comparing the forward model to the neural network alternative show encouraging results with good agreements between the two when applied to retrieval scenarios using both synthetic and real measured spectra from TROPOMI (TROPOspheric Monitoring Instrument) on board the ESA Sentinel-5 Precursor mission. With an enhancement of the computational speed by three orders of magnitude, TROPOMI's operational aerosol layer height processor is now able to retrieve aerosol layer heights well within operational capacity.

scholarly journals Combining radiative transfer calculations and a neural network for the remote sensing of volcanic ash using MSG/SEVIRI

2021 ◽  
Author(s):  
Luca Bugliaro ◽  
Dennis Piontek ◽  
Stephan Kox ◽  
Marius Schmidl ◽  
Bernhard Mayer ◽  
...  
Keyword(s):  
Neural Network ◽  
Remote Sensing ◽  
Radiative Transfer ◽  
Volcanic Ash ◽  
Correlation Coefficients ◽  
Probability Of Detection ◽  
High Mass ◽  
Data Set ◽  
Novel Method ◽  
Lidar Observations

Abstract. After the eruption of volcanoes all over the world the monitoring of the dispersion of ash in the atmosphere is an important task for satellite remote sensing since ash represents a threat to air traffic. In this work we present a novel method that uses thermal observations of the SEVIRI imager aboard the geostationary Meteosat Second Generation satellite to detect ash clouds and determine their mass column concentration and top height during day and night. This approach requires the compilation of an extensive data set of synthetic SEVIRI observations to train an artificial neural network. This is done by means of the RTSIM tool that combines atmospheric, surface and ash properties and runs automatically a large number of radiative transfer calculations for the entire SEVIRI disk. The resulting algorithm is called VADUGS (Volcanic Ash Detection Using Geostationary Satellites) and has been evaluated against independent radiative transfer simulations. VADUGS detects ash contaminated pixels with a probability of detection of 0.84 and a false alarm rate of 0.05. Ash column concentrations are provided by VADUGS with correlations up to 0.5, a scatter up to 0.6 g m−2 for concentrations smaller than 2.0 g m−2 and small overestimations in the range 5–50 % for moderate viewing angles 35–65°, but up to 300 % for satellite viewing zenith angles close to 90° or 0°. Ash top heights are mainly underestimated, with the smallest underestimation of −9 % for viewing zenith angles between 40° and 50°. Absolute errors are smaller than 70 % and with high correlation coefficients up to 0.7 for ash clouds with high mass column concentrations. A comparison against spaceborne lidar observations by CALIPSO/CALIOPconfirms these results. VADUGS is run operationally at the German Weather Service and this application is presented as well.

Sign in / Sign up

Export Citation Format

Share Document