Searching for new physics: Using explainable AI to understand deep learned parameterizations of turbulent heat fluxes

2021 ◽  
Author(s):  
Andrew Bennett ◽  
Bart Nijssen
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Deep Learning ◽  
Coupled System ◽  
Hydrologic Model ◽  
Heat Fluxes ◽  
Turbulent Heat ◽  
Phase Lag ◽  
Geophysical Research ◽  
Turbulent Heat Fluxes

<p>Machine learning (ML), and particularly deep learning (DL), for geophysical research has shown dramatic successes in recent years. However, these models are primarily geared towards better predictive capabilities, and are generally treated as black box models, limiting researchers’ ability to interpret and understand how these predictions are made. As these models are incorporated into larger models and pushed to be used in more areas it will be important to build methods that allow us to reason about how these models operate. This will have implications for scientific discovery that will ensure that these models are robust and reliable for their respective applications. Recent work in explainable artificial intelligence (XAI) has been used to interpret and explain the behavior of machine learned models.</p><p>Here, we apply new tools from the field of XAI to provide physical interpretations of a system that couples a deep-learning based parameterization for turbulent heat fluxes to a process based hydrologic model. To develop this coupling we have trained a neural network to predict turbulent heat fluxes using FluxNet data from a large number of hydroclimatically diverse sites. This neural network is coupled to the SUMMA hydrologic model, taking imodel derived states as additional inputs to improve predictions. We have shown that this coupled system provides highly accurate simulations of turbulent heat fluxes at 30 minute timesteps, accurately predicts the long-term observed water balance, and reproduces other signatures such as the phase lag with shortwave radiation. Because of these features, it seems this coupled system is learning physically accurate relationships between inputs and outputs. </p><p>We probe the relative importance of which input features are used to make predictions during wet and dry conditions to better understand what the neural network has learned. Further, we conduct controlled experiments to understand how the neural networks are able to learn to regionalize between different hydroclimates. By understanding how these neural networks make their predictions as well as how they learn to make predictions we can gain scientific insights and use them to further improve our models of the Earth system.</p>

Deep Learning based cloud parametrization for the Community Atmosphere Model

2020 ◽  
Author(s):  
Gunnar Behrens ◽  
Veronika Eyring ◽  
Pierre Gentine ◽  
Mike S. Pritchard ◽  
Tom Beucler ◽  
...  
Keyword(s):  
Neural Network ◽  
Machine Learning ◽  
Neural Networks ◽  
Deep Learning ◽  
Climate Models ◽  
Large Contribution ◽  
Geophysical Research ◽  
Validation Data ◽  
Community Atmosphere Model ◽  
Atmosphere Model

<p>Despite significant progress in climate modeling over the last decades, the spread in simulated equilibrium climate sensitivity, i.e. the change in global mean surface temperature after a doubling of atmospheric CO2 concentrations, has not decreased since the 1970s. Estimates from the Coupled Model Intercomparison Project Phase 5 (CMIP5) range between 2.1 and 4.7°C, and this range is even larger in CMIP6. A large contribution to this uncertainty stems from differences in the representation of clouds and convection occurring at scales smaller than the resolved model grid resolution. Machine learning approaches have recently been shown to be potentially powerful alternatives to physical parametrizations of convection (Gentine et al. 2018; Rasp et al. 2018). Despite improving understanding of the minimization of generalization errors in out-of-sample climate states (Beucler et al. 2019) these approaches struggle to capture the full variability observed in the training and validation data. First experiments with autoencoder-decoder structures illustrate a compatible skill in comparison to feedforward neural networks.</p><p>In this study we modify the existing neural network parametrization using more advanced neural network structures. The new parametrization is tested with the Community Atmosphere Model Version 3 for different climate states. This work is a contribution to the development of complex subgrid-scale convective and turbulence parametrizations performed within the recently awarded European Research Council (ERC) Synergy Grant “Understanding and modelling the Earth system with machine learning (USMILE)”.</p><p> </p><p> </p><p><strong>References:</strong></p><p><span>Gentine, P., Pritchard, M.S., Rasp, S., Reinaudi, G., & Yacalis, G. (2018): Could Machine Learning Break the Convection Parameterization Deadlock? </span><em><span>Geophysical Research Letters, 45</span></em><span>, 5742-5751</span></p><p><span>Rasp, S., Pritchard, M.S., & Gentine, P. (2018): Deep learning to represent subgrid processes in climate models. </span><em><span>Proceedings of the National Academy of Sciences</span></em></p><p><span>Beucler, T., Rasp, S., Pritchard, M.S., Gentine, P.,Ott, J. & Baldi, P. (2019)</span><span>: </span><span>Enforcing Analytic Constraints in Neural-Networks Emulating Physical Systems.</span><em><span> Physical Review Letters, https://arxiv.org/abs/1909.00912</span></em></p>

scholarly journals Convolutional Neural Network for the Semantic Segmentation of Remote Sensing Images

2021 ◽  
Vol 26 (1) ◽  
pp. 200-215
Author(s):  
Muhammad Alam ◽  
Jian-Feng Wang ◽  
Cong Guangpei ◽  
LV Yunrong ◽  
Yuanfang Chen
Keyword(s):  
Neural Network ◽  
Remote Sensing ◽  
Neural Networks ◽  
Image Processing ◽  
Deep Learning ◽  
Semantic Segmentation ◽  
Natural Scene ◽  
Remote Sensing Images ◽  
Advantages And Disadvantages ◽  
Target Segmentation

AbstractIn recent years, the success of deep learning in natural scene image processing boosted its application in the analysis of remote sensing images. In this paper, we applied Convolutional Neural Networks (CNN) on the semantic segmentation of remote sensing images. We improve the Encoder- Decoder CNN structure SegNet with index pooling and U-net to make them suitable for multi-targets semantic segmentation of remote sensing images. The results show that these two models have their own advantages and disadvantages on the segmentation of different objects. In addition, we propose an integrated algorithm that integrates these two models. Experimental results show that the presented integrated algorithm can exploite the advantages of both the models for multi-target segmentation and achieve a better segmentation compared to these two models.

scholarly journals Classification of Skin Disease Using Deep Learning Neural Networks with MobileNet V2 and LSTM

Sensors ◽  
2021 ◽  
Vol 21 (8) ◽  
pp. 2852
Author(s):  
Parvathaneni Naga Srinivasu ◽  
Jalluri Gnana SivaSai ◽  
Muhammad Fazal Ijaz ◽  
Akash Kumar Bhoi ◽  
Wonjoon Kim ◽  
...  
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Deep Learning ◽  
Convolutional Neural Network ◽  
Skin Disease ◽  
Network Architecture ◽  
Large Scale ◽  
Short Term Memory ◽  
Convolutional Networks ◽  
Occurrence Matrix

Deep learning models are efficient in learning the features that assist in understanding complex patterns precisely. This study proposed a computerized process of classifying skin disease through deep learning based MobileNet V2 and Long Short Term Memory (LSTM). The MobileNet V2 model proved to be efficient with a better accuracy that can work on lightweight computational devices. The proposed model is efficient in maintaining stateful information for precise predictions. A grey-level co-occurrence matrix is used for assessing the progress of diseased growth. The performance has been compared against other state-of-the-art models such as Fine-Tuned Neural Networks (FTNN), Convolutional Neural Network (CNN), Very Deep Convolutional Networks for Large-Scale Image Recognition developed by Visual Geometry Group (VGG), and convolutional neural network architecture that expanded with few changes. The HAM10000 dataset is used and the proposed method has outperformed other methods with more than 85% accuracy. Its robustness in recognizing the affected region much faster with almost 2× lesser computations than the conventional MobileNet model results in minimal computational efforts. Furthermore, a mobile application is designed for instant and proper action. It helps the patient and dermatologists identify the type of disease from the affected region’s image at the initial stage of the skin disease. These findings suggest that the proposed system can help general practitioners efficiently and effectively diagnose skin conditions, thereby reducing further complications and morbidity.

Estimation of turbulent heat fluxes and exchange coefficients for heat at Dumont d'Urville, East Antarctica

1999 ◽  
Vol 11 (1) ◽  
pp. 93-99 ◽  
Author(s):  
S. Argentini ◽  
G. Mastrantonio ◽  
A. Viola
Keyword(s):  
Boundary Layer ◽  
Heat Fluxes ◽  
East Antarctica ◽  
Turbulent Heat ◽  
Convective Boundary ◽  
Doppler Sodar ◽  
Turbulent Heat Fluxes ◽  
Unstable Boundary Layer ◽  
Sodar Measurements ◽  
The Antarctic

Simultaneous acoustic Doppler sodar and tethersonde measurements were used to study some of the characteristics of the unstable boundary layer at Dumont d'Urville, Adélie Land, East Antarctica during the summer 1993–94. A description of the convective boundary layer and its behaviour in connection with the wind regime is given along with the frequency distribution of free convection episodes. The surface heat flux has been evaluated using the vertical velocity variance derived from sodar measurements. The turbulent exchange coefficients, estimated by coupling sodar and tethered balloon measurements, are in strong agreement with those present in literature for the Antarctic regions.

The Effect of Soil on the Summertime Surface Energy Budget of a Humid Subarctic Tundra in Northern Quebec, Canada

2021 ◽  
Vol 22 (10) ◽  
pp. 2547-2564
Author(s):  
Georg Lackner ◽  
Daniel F. Nadeau ◽  
Florent Domine ◽  
Annie-Claude Parent ◽  
Gonzalo Leonardini ◽  
...  
Keyword(s):  
Heat Flux ◽  
Latent Heat ◽  
Latent Heat Flux ◽  
Land Surface ◽  
Sensible Heat Flux ◽  
Arctic Region ◽  
Heat Fluxes ◽  
Surface Energy Budget ◽  
Turbulent Heat ◽  
Turbulent Heat Fluxes

AbstractRising temperatures in the southern Arctic region are leading to shrub expansion and permafrost degradation. The objective of this study is to analyze the surface energy budget (SEB) of a subarctic shrub tundra site that is subject to these changes, on the east coast of Hudson Bay in eastern Canada. We focus on the turbulent heat fluxes, as they have been poorly quantified in this region. This study is based on data collected by a flux tower using the eddy covariance approach and focused on snow-free periods. Furthermore, we compare our results with those from six Fluxnet sites in the Arctic region and analyze the performance of two land surface models, SVS and ISBA, in simulating soil moisture and turbulent heat fluxes. We found that 23% of the net radiation was converted into latent heat flux at our site, 35% was used for sensible heat flux, and about 15% for ground heat flux. These results were surprising considering our site was by far the wettest site among those studied, and most of the net radiation at the other Arctic sites was consumed by the latent heat flux. We attribute this behavior to the high hydraulic conductivity of the soil (littoral and intertidal sediments), typical of what is found in the coastal regions of the eastern Canadian Arctic. Land surface models overestimated the surface water content of those soils but were able to accurately simulate the turbulent heat flux, particularly the sensible heat flux and, to a lesser extent, the latent heat flux.

The signature of mesoscale eddies on the air‐sea turbulent heat fluxes in the South Atlantic Ocean

2015 ◽  
Vol 42 (6) ◽  
pp. 1856-1862 ◽  
Author(s):  
A. B. Villas Bôas ◽  
O. T. Sato ◽  
A. Chaigneau ◽  
G. P. Castelão
Keyword(s):  
Atlantic Ocean ◽  
Mesoscale Eddies ◽  
South Atlantic ◽  
Heat Fluxes ◽  
South Atlantic Ocean ◽  
Turbulent Heat ◽  
The South ◽  
Turbulent Heat Fluxes

scholarly journals Evaluation of the heat balance components over the Baltic Sea using four gridded meteorological databases and direct observations

2005 ◽  
Vol 36 (4-5) ◽  
pp. 381-396 ◽  
Author(s):  
A. Rutgersson ◽  
A. Omstedt ◽  
Y. Chen
Keyword(s):  
Baltic Sea ◽  
Short Wave ◽  
Heat Fluxes ◽  
Wave Radiation ◽  
Turbulent Heat ◽  
The Baltic Sea ◽  
Short Wave Radiation ◽  
Turbulent Heat Fluxes ◽  
The Baltic ◽  
The Heat Balance

In this paper, which reports on part of the BALTEX project, various components of the heat balance over the Baltic Sea are calculated using a number of gridded meteorological databases. It is the heat exchange between the Baltic Sea surface and the atmosphere that is of interest. The databases have different origins, comprising synoptic data, data re-analysed with a 3D assimilation system, an ocean model forced with gridded synoptic data, ship data and satellite data. We compared the databases and found that the greatest variation between them is in the long- and short-wave radiation values. However, considerable upward long-wave radiation is followed by considerable downward short-wave radiation, so the total radiation component is partly compensated for in the total budget. The variation in the total heat transport in the databases therefore appears smaller (1.5±3 W m−2) as the average and one standard deviation. The turbulent heat fluxes estimated from satellite data have very low values; this can largely be explained by the method of calculating air temperature, which also produces an unrealistic stratification over the Baltic Sea. The ERA40 data was compared with measured values: there, we found a certain land influence even in the centre of the Baltic proper. The indicated turbulent heat fluxes were too large, mainly in the fall and winter, and the sensible heat flux was too large in a downward direction in spring and summer.

Sign in / Sign up

Export Citation Format

Share Document