scholarly journals A map of global peatland distribution created using machine learning for use in terrestrial ecosystem and earth system models

2017 ◽  
Author(s):  
Yuanqiao Wu ◽  
Ed Chan ◽  
Joe R. Melton ◽  
Diana L. Verseghy
Keyword(s):  
Machine Learning ◽  
Statistical Model ◽  
Machine Learning Method ◽  
Learning Method ◽  
Earth System ◽  
Geophysical Field ◽  
Fractional Coverage ◽  
System Models ◽  
Earth System Models ◽  
Global Carbon

Abstract. Peatlands store large amounts of soil carbon and constitute an important component of the global carbon cycle. Accurate information on the global extent and distribution of peatlands is presently lacking but it important for earth system models (ESMs) to be able to simulate the effects of climate change on the global carbon balance. The most comprehensive peatland map produced to date is a qualitative presence/absence product. Here, we present a spatially continuous global map of peatland fractional coverage using the extremely randomized tree machine learning method suitable for use as a prescribed geophysical field in an ESM. Inputs to our statistical model include spatially distributed climate data, soil data and topographical slopes. Available maps of peatland fractional coverage for Canada and West Siberia were used along with a proxy for non-peatland areas to train and test the statistical model. Regions where the peatland fraction is expected to be zero were estimated from a map of topsoil organic carbon content below a threshold value of 13 kg/m2. The modelled coverage of peatlands yields a root mean square error of 4 % and a coefficient of determination of 0.91 for the 10,978 tested 0.5 degree grid cells. We then generated a complete global peatland fractional coverage map. In comparison with earlier qualitative estimates, our global modelled peatland map is able to reproduce peatland distributions in places remote from the training areas and capture peatland hot spots in both boreal and tropical regions, as well as in the southern hemisphere. Additionally we demonstrate that our machine-learning method has greater skill than solely setting peatland areas based on histosols from a soil database.

scholarly journals Physically regularized machine learning emulators of aerosol activation

2021 ◽  
Vol 14 (5) ◽  
pp. 3067-3077
Author(s):  
Sam J. Silva ◽  
Po-Lun Ma ◽  
Joseph C. Hardin ◽  
Daniel Rothenberg
Keyword(s):  
Machine Learning ◽  
Model Development ◽  
Computational Cost ◽  
Earth System ◽  
Cloud Droplets ◽  
Physical Constraints ◽  
System Models ◽  
Machine Learning Model ◽  
Aerosol Activation ◽  
Earth System Models

Abstract. The activation of aerosol into cloud droplets is an important step in the formation of clouds and strongly influences the radiative budget of the Earth. Explicitly simulating aerosol activation in Earth system models is challenging due to the computational complexity required to resolve the necessary chemical and physical processes and their interactions. As such, various parameterizations have been developed to approximate these details at reduced computational cost and accuracy. Here, we explore how machine learning emulators can be used to bridge this gap in computational cost and parameterization accuracy. We evaluate a set of emulators of a detailed cloud parcel model using physically regularized machine learning regression techniques. We find that the emulators can reproduce the parcel model at higher accuracy than many existing parameterizations. Furthermore, physical regularization tends to improve emulator accuracy, most significantly when emulating very low activation fractions. This work demonstrates the value of physical constraints in machine learning model development and enables the implementation of improved hybrid physical and machine learning models of aerosol activation into next-generation Earth system models.

scholarly journals A machine learning-guided adaptive algorithm to reduce the computational cost of atmospheric chemistry in Earth System models: application to GEOS-Chem versions 12.0.0 and 12.9.1

2021 ◽  
Author(s):  
Lu Shen ◽  
Daniel J. Jacob ◽  
Mauricio Santillana ◽  
Kelvin Bates ◽  
Jiawei Zhuang ◽  
...  
Keyword(s):  
Machine Learning ◽  
Atmospheric Chemistry ◽  
Computational Cost ◽  
Coupled System ◽  
Atmospheric Composition ◽  
Earth System ◽  
Major Barrier ◽  
System Models ◽  
Computational Bottleneck ◽  
Earth System Models

Abstract. Atmospheric composition plays a crucial role in determining the evolution of the atmosphere, but the high computational cost has been the major barrier to include atmospheric chemistry into Earth system models. Here we present an adaptive and efficient algorithm that can remove this barrier. Our approach is inspired by unsupervised machine learning clustering techniques and traditional asymptotic analysis ideas. We first partition species into 13 blocks, using a novel machine learning approach that analyzes the species network structures and their production and loss rates. Building on these blocks, we pre-select 20 submechanisms, as defined by unique assemblages of the species blocks, and then pick locally on the fly which submechanism to use based on local chemical conditions. In each submechanism, we isolate slow species and unimportant reactions from the coupled system. Application to a global 3-D model shows that we can cut the computational costs of the chemical integration by 50 % with accuracy losses smaller than 1 % that do not propagate in time. Tests show that this algorithm is highly chemically coherent making it easily portable to new models without compromising its performance. Our algorithm will significantly ease the computational bottleneck and will facilitate the development of next generation of earth system models.

scholarly journals Learning earth system models from observations: machine learning or data assimilation?

2021 ◽  
Vol 379 (2194) ◽  
pp. 20200089
Author(s):  
A. J. Geer
Keyword(s):  
Machine Learning ◽  
Data Assimilation ◽  
Bayesian Networks ◽  
Physical Models ◽  
Grand Challenge ◽  
Climate Modelling ◽  
Earth System ◽  
System Models ◽  
Physical Knowledge ◽  
Earth System Models

Recent progress in machine learning (ML) inspires the idea of improving (or learning) earth system models directly from the observations. Earth sciences already use data assimilation (DA), which underpins decades of progress in weather forecasting. DA and ML have many similarities: they are both inverse methods that can be united under a Bayesian (probabilistic) framework. ML could benefit from approaches used in DA, which has evolved to deal with real observations—these are uncertain, sparsely sampled, and only indirectly sensitive to the processes of interest. DA could also become more like ML and start learning improved models of the earth system, using parameter estimation, or by directly incorporating machine-learnable models. DA follows the Bayesian approach more exactly in terms of representing uncertainty, and in retaining existing physical knowledge, which helps to better constrain the learnt aspects of models. This article makes equivalences between DA and ML in the unifying framework of Bayesian networks. These help illustrate the equivalences between four-dimensional variational (4D-Var) DA and a recurrent neural network (RNN), for example. More broadly, Bayesian networks are graphical representations of the knowledge and processes embodied in earth system models, giving a framework for organising modelling components and knowledge, whether coming from physical equations or learnt from observations. Their full Bayesian solution is not computationally feasible but these networks can be solved with approximate methods already used in DA and ML, so they could provide a practical framework for the unification of the two. Development of all these approaches could address the grand challenge of making better use of observations to improve physical models of earth system processes. This article is part of the theme issue ‘Machine learning for weather and climate modelling’.

scholarly journals Physically Regularized Machine Learning Emulators of Aerosol Activation

2020 ◽  
Author(s):  
Sam J. Silva ◽  
Po-Lun Ma ◽  
Joseph C. Hardin ◽  
Daniel Rothenberg
Keyword(s):  
Machine Learning ◽  
Model Development ◽  
Computational Cost ◽  
Earth System ◽  
Cloud Droplets ◽  
Physical Constraints ◽  
System Models ◽  
Machine Learning Model ◽  
Aerosol Activation ◽  
Earth System Models

Abstract. The activation of aerosol into cloud droplets is an important step in the formation of clouds, and strongly influences the radiative budget of the Earth. Explicitly simulating aerosol activation in Earth system models is challenging due to the computational complexity required to resolve the necessary chemical and physical processes and their interactions. As such, various parameterizations have been developed to approximate these details at reduced computational cost and accuracy. Here, we explore how machine learning emulators can be used to bridge this gap in computational cost and parameterization accuracy. We evaluate a set of emulators of a detailed cloud parcel model using physically regularized machine learning regression techniques. We find that the emulators can reproduce the parcel model at higher accuracy than many existing parameterizations. Furthermore, physical regularization tends to improve emulator accuracy, most significantly when emulating very low activation fractions. This work demonstrates the value of physical constraints in machine learning model development and enables the implementation of improved hybrid physical-machine learning models of aerosol activation into next generation Earth system models.

Sign in / Sign up

Export Citation Format

Share Document