English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

We test the reliability of two neural network interpretation techniques,backward optimization and layerwise relevance propagation, within geoscientificapplications by applying them to a commonly studied geophysical phenomenon, theMadden-Julian Oscillation. The Madden-Julian Oscillation is a multi-scalepattern within the tropical atmosphere that has been extensively studied overthe past decades, which makes it an ideal test case to ensure theinterpretability methods can recover the current state of knowledge regardingits spatial structure. The neural networks can, indeed, reproduce the currentstate of knowledge and can also provide new insights into the seasonality ofthe Madden-Julian Oscillation and its relationships with atmospheric statevariables.  The neural network identifies the phase of the Madden-Julian Oscillationtwice as accurately as linear regression, which means that nonlinearities usedby the neural network are important to the structure of the Madden-JulianOscillation. Interpretations of the neural network show that it accuratelycaptures the spatial structures of the Madden-Julian Oscillation, suggest thatthe nonlinearities of the Madden-Julian Oscillation are manifested through theuniqueness of each event, and offer physically meaningful insights into itsrelationship with atmospheric state variables. We also use the interpretationsto identify the seasonality of the Madden-Julian Oscillation, and find that theconventionally defined extended seasons should be shifted later by one month.More generally, this study suggests that neural networks can be reliablyinterpreted for geoscientific applications and may there

Testing the Reliability of Interpretable Neural Networks in Geoscience Using the Madden-Julian Oscillation