Michael Taylor

Applied Mathematician, UK

CURRENT WORK: I work at UEA CRU on the GloSAT project.




Independent Research

Sep 2019 – July 2020 UK
Re-training in climate science. I built a quadrifilar helix antenna to receive APT weather satellite data with an RTL-SDR dongle and learned how to program a Raspberry Pi3 to automate the downlink and serve the imagery via a web app to set up an amateur remote sensing station. I also installed NCAR’s Weather Research and Forcasting (WRF) model. The app uses API to draw in hourly meteorological and hydrological data from UKMO and UKEA. To help during the pandemic I wrote 2 Plotly python reactive apps to visualise and provide a global daily update of Coronavirus lockdown status and to run a MCMC estimated country-level hindcast/forecast. I have been working with Amy Catanzano to develop an algorithm for quantum poetry.



Department of Meteorology, University of Reading

May 2017 – Aug 2019 UK
H2020 FIDUCEO: Climate data from Earth observation. ESA SST_CCI: SST observation density analysis. C3S EQCO: Quality assurance of ECV products.


Laboratory of Atmospheric Physics, Aristotle University of Thessaloniki

Mar 2016 – Mar 2017 Greece
Contributing author to SPARC/IO3C/GAW Report 9 ‘Long-term Ozone Trends and Uncertainties in the Stratosphere’. ESA Ozone_CCI: construction of ozone datasets for the study of climate. EUMETSAT AC-SAF validation reports for MetOp-B OUV products.

Research Fellow

National Observatory of Athens

Mar 2012 – Mar 2017 Greece
H2020 GEO-CRADLE SENSE pilot (WP4): Use of neural networks to develop solar energy potential. MARIE-CURIE IEF AEROMAP: Global mapping of aerosol properties using neural network inversions of ground and satellite based data.


IAASARS, National Observatory of Athens

Oct 2008 – Oct 2009 Greece
Greek National Scholarships Foundation (IKY) research award: ‘Space weather prediction using nonlinear techniques’.


Spanish National Research Council (CSIC)

Mar 2007 – Dec 2007 Spain
ESA/Herschel Space Telescope (HST) grant. Department of Astrophysical Molecules & the InfraRed (DAMIR): ‘3D inverse models and reconstruction of depth information’.


Departamento de Fisica Teorica, Universidad Autonoma de Madrid (UAM)

Jan 2004 – Mar 2007 Spain
ESTALLIDOS: ‘3D radiative transfer modelling of astrophysical photoionization regions’.


Department of Physics, University of Bath

Sep 2001 – Sep 2002 UK
Undergraduate Courses: Vibrations & Waves, Electrical Circuits I, Magnetism III, Matrices, Mathematics I & II. MSc supervision (4 postgrads).

Books Commissioning Editor

Institute of Physics Publishing

Mar 1998 – Jun 2000 UK
Commissioning editor for popular science, graduate textbooks and research monographs.

Research Associate

Space & Atmospheric Physics Group, Blackett Laboratory, Imperial College

Jan 1996 – Dec 1996 UK
NASA/JPL Galileo mission grant: ‘Magnetometer data analysis of ion cyclotron waves in the Jupiter aurorae’

Research Associate

Government Fusion Division, Culham Laboratory, Oxford

Jan 1995 – Dec 1995 UK
EURATOM-JET/MST CASE award: ‘Wave propagation through cyclotron resonance in the presence of large Larmor radius particles’.



Multi-lognormal fitting of AERONET aerosol volume size distributions.

Archived version of the website for the FP7 Marie-Curie IEF project 'AEROMAP - Global mapping of aerosol properties using neural network inversions of ground and satellite based data' (2012-2014):

Cartopy-based routine to allow plotting of L1C orbital parameters in various projection modes.

Code to generate the L1C AVHRR FCDR with metrologically-traceable uncertainty estimates. .

Code to generate the L1C AVHRR FCDR ensemble used as input to surface temperature CDR. .

FIDUCEO L1C harmonisation comparison with CCI OE-based harmonisation using SST matchups.

Noise characteristics of the thermal channels of the AVHRR.

ESA SST CDR observation density and retrieval uncertainty calculation from the AVHRR and ATSR series of sensors. .

ESA SST CCI work forward modeled comparison between L3U and L4 SST variation with latitude.

Development code for optimal estimation (OE)-driven re-harmonisation of the measurement equations used for retrieval of SST from AVHRR IR channels.

Calculation of high resolution latitudinal variation of ocean + sea ice area using the L4 landsea mask produced with the OSTIA reanalysis sytem v3.0.

MATLAB code to download, read, parse and write operational data from NDACC and WOUDC.

Plotting tools for post-processing the output of the chemical transport model CHIMERE. .

Development version of code to convert .mov clips of birds in flight to ornitographic trajectories inspired by the work of http://www.xavibou.com/

Optimisation of SSA in R using genetic programming

Neural network radiative transfer solver for generation of surface solar spectral radiance maps and UV products

Remote sensing of weather satellites with a quadrifilar helix (QFH) antenna

Coronavirus global lockdown status daily dashboard

Coronavirus MCMC hindcast/forecast dashboard

Python implementation of World Lines quantum poetry algorithm in collaboration with


H2020 FIDUCEO progress update

ESA SST_CCI progress update

C3S EQCO reportback

H2020 FIDUCEO progress report

H2020 GEO-CRADLE SENSE pilot progress update

H2020 GEO-CRADLE SENSE pilot progress update

Marie-Curie IEF AEROMAP reportback

ESTALLIDOS-IV progress update


Quickly discover relevant content by filtering publications.

We report on the development of a spatial back-propagation neural network (S-BPNN) model designed to retrieve PM2.5 concentrations over China from space by specifically making spatial correlations implicit by incorporating an ArcGIS-derived spatial lag variable (SLV) as a virtual input variable.

This paper presents the framework for Evaluation and Quality Control (EQC) of climate data products derived from satellite and in situ observations to be catalogued within the C3S Climate Data Store (CDS). The EQC framework will be implemented by C3S as part of their operational quality assurance programme.

This Report summarises the main results obtained during the first year of the LOTUS activity, which was targeted at providing timely inputs to the 2018 WMO/UNEP Ozone Assessment. The scope of this Report is on changes in ozone levels in the middle and the upper stratosphere outside the polar regions as observed by merged satellite records.

This study estimates the impact of dust aerosols on surface solar radiation and solar energy in Egypt based on Earth Observation (EO) related techniques. For this purpose, we exploited the synergy of monthly mean and daily post processed satellite remote sensing observations from the MODerate resolution Imaging Spectroradiometer (MODIS), radiative transfer model (RTM) simulations utilising machine learning, in conjunction with 1-day forecasts from the Copernicus Atmosphere Monitoring Service (CAMS).

This study focuses on the assessment of surface solar radiation (SSR) based on operational neural network (NN) and multi-regression function (MRF) modelling techniques that produce instantaneous (in less than 1min) outputs using real-time cloud and aerosol optical properties inputs from SEVIRI onboard MSG and the Copernicus Atmosphere Monitoring Service (CAMS).

This study assesses the impact of dust on surface solar radiation focussing on an extreme dust event. For this purpose, we exploited the synergy of AERONET measurements and passive and active satellite remote sensing (MODIS and CALIPSO) observations, in conjunction with radiative transfer model (RTM) and chemical transport model (CTM) simulations and the 1-day forecasts from the Copernicus Atmosphere Monitoring Service (CAMS).

This study aims to cross-validate ground-based and satellite-based models of three photobiological UV effective dose products - the Commission Internationale de l’Eclairage (CIE) erythemal UV, the production of vitamin D in the skin, and DNA damage, using high-temporal-resolution surface-based measurements of solar UV spectral irradiances from a synergy of instruments and models.

We report on the construction of generic models to calculate photosynthetically active radiation (PAR) from global horizontal irradiance (GHI), and vice versa. Our study took place at stations of the Greek UV network (UVNET) and the Hellenic solar energy network (HNSE) with measurements from NILU-UV multi-filter radiometers and CM pyranometers, chosen due to their long (~1M record/site) high temporal resolution (1 min) record that captures a broad range of atmospheric environments and cloudiness conditions.

We report on the development and validation of a neural network (NN) model of PM10 concentrations in terms of photochemical measurements of NO, NO₂ and O₃ and temporal parameters that include the day of the week and the day of the year with its sinusoidal variation.

This paper reports on the development of a neural network (NN) model for instantaneous and accurate estimation of solar radiation spectra and budgets geared toward satellite cloud data using a ~ 2.4 M record, high-spectral resolution look up table (LUT) generated with the radiative transfer model libRadtran.

Cluster analysis is applied to seven years of three-hourly gridded aerosol optical depth data from the Goddard Chemistry Aerosol Radiation and Transport model, to construct a spatial partition of the globe into a finite number of aerosol mixtures. The optimal number of aerosol mixtures is obtained with a k-means algorithm with smart seeding in conjunction with a stopping condition based on applying the ‘Law of diminishing returns’ to the norm of the Euclidean distance. The partition is used to extract AERONET Level-2 V2 inversion products in each cluster for estimating the values of key optical and microphysical parameters to help characterise aerosol mixtures.

Here, a methodology based on neural networks is developed to retrieve such parameters from satellite inputs and to validate them with ground-based remote sensing data. For key combinations of input variables available from the MODerate resolution Imaging Spectroradiometer (MODIS) and the Ozone Measuring Instrument (OMI) Level 3 data sets, a grid of 100 feed-forward neural network architectures is produced to derive aerosol optical and microphysical properties.

We present a method for determining the mode separation point for equivalent-volume bi-lognormal distributions based on optimization of the root mean squared error and the coefficient of determination. The extracted secondary parameters are compared with those provided by AERONET’s Level 2.0 Version 2 inversion algorithm for a set of benchmark dominant aerosol types, including desert dust, biomass burning aerosol, urban sulphate and sea salt.

We demonstrate improvements in CALIPSO (Cloud–Aerosol Lidar and Infrared Pathfinder Satellite Observations) dust extinction retrievals over northern Africa and Europe when corrections are applied regarding the Saharan dust lidar ratio assumption, the separation of the dust portion in detected dust mixtures, and the averaging scheme introduced in the Level 3 CALIPSO product.

In this work, we have developed a generalisation of the second order NESM model (Tsallis et al. (2003)) to higher orders and we have fit the complete spectrum including the ankle with third order NESM. We find that, towards the GDZ limit, a new mechanism comes into play. Surprisingly it also presents as a modulation akin to that in our own local neighbourhood of cosmic rays emitted by the sun. We propose that this is due to modulation at the source and is possibly due to processes in the shell of the originating supernova.

This paper presents a review of the analytical approach to obtain exact solutions for the populations of n-level ions, and summarizes the ideas behind detailed balance and the statistical physics of collisionally-excited ions. We show that the analytical method of solution to the problem using matrices and symbolic mathematics is straightforward, and we illustrate through theoretical, numerical, and empirical checks the validity of its results.

Here, we report on a new artificial intelligence-based approach to determine metallicity indicators that shows promise for the provision of improved empirical fits. The method hinges on the application of an evolutionary neural network to observational emission line data.