Software and methods

I develop numerical methods that turn polarised-light physics into simulations of planetary atmosphere observations. The centrepiece is MONKI: a three-dimensional Monte Carlo radiative-transfer code written in Fortran.

How the methods are built

First-principles derivation

Deriving the equations for polarised radiative transfer from the underlying physics, rather than treating the model as a black box.

Numerical implementation

Translating the physics into robust Fortran code, including photon-path sampling, scattering events, Stokes-vector bookkeeping, absorption, and statistical convergence.

Benchmarking and validation

Comparing model output with established radiative-transfer benchmarks and using discrepancies to improve the method and implementation.

Physical interpretation

Using simulations to explain why specific radiance and polarisation features appear in Earth-observation, Venus, and exoplanet signals.

MONKI · 3D Monte Carlo radiative transfer

Three-dimensional radiative transfer for total and polarised light

MONKI is the central code development in my work: a three-dimensional Monte Carlo radiative-transfer model for total and polarised radiation reflected and transmitted by planetary atmospheres. I wrote the Fortran code from scratch, including the photon sampling, local-estimation method, polarisation bookkeeping, absorption treatment, and parallel execution needed for large atmospheric simulations.

The code can be used for horizontally homogeneous atmospheres and for fully three-dimensional scenes, such as cloud fields from 3D cloudy atmospheric models, including large-eddy simulations. MONKI is currently used mainly for Earth and Venus applications, where it computes total and polarised light reflected and transmitted by complex planetary atmospheres.

MONKI paper Zenodo code archive

MONKI simulation of intensity and degree of polarisation for a 3D cloudy atmosphere — MONKI simulation of intensity and degree of polarisation in a 3D cloudy atmosphere. Figure by Victor J. H. Trees.

Sketch of the 3D MONKI computational domain and photon trajectories — Photon trajectories and local estimation in the MONKI 3D computational domain. Figure adapted from Trees et al. (2025).

DARCLOS · TROPOMI cloud-shadow detection

Detecting displaced cloud shadows in satellite observations

DARCLOS is a cloud-shadow detection method for TROPOMI. It uses the geometry of the Sun, cloud height, surface location, and satellite viewing direction to estimate where a cloud shadow should appear on the Earth’s surface, even though the cloud itself is observed at a different apparent location by the satellite.

The resulting cloud-shadow flag helps identify scenes where three-dimensional cloud effects influence measured reflectances and derived atmospheric products. This connects directly to my Earth-observation work on aerosols, surface reflectance, and the physical interpretation of satellite measurements.

DARCLOS paper

VIIRS true-colour scene and DARCLOS cloud-shadow flag on the TROPOMI grid — VIIRS true colour next to the DARCLOS cloud-shadow flag on the TROPOMI grid. Figure adapted from Trees et al. (2022).

Sketch of the cloud-shadow geometry used in DARCLOS — Geometry linking the apparent cloud pixel, true cloud position, and displaced shadow. Figure adapted from Trees et al. (2022).

Solar-eclipse correction · Reflectance restoration

Restoring satellite measurements during solar eclipses

The solar-eclipse correction method computes how much of the solar disk is blocked by the Moon for each satellite pixel and uses that geometry to recover satellite measurements. This turns eclipse scenes from problematic observations into usable measurements of the atmosphere inside the Moon’s partial shadow.

I developed this correction to investigate how reduced sunlight affects measured radiation fields. It later made it possible to study the rapid response of shallow cumulus clouds to solar eclipses.

Eclipse-correction paper Cloud-response paper

Satellite images during the 2005 solar eclipse before and after correction for reduced sunlight — Satellite measurements during the 3 October 2005 solar eclipse before and after correcting for reduced sunlight. Figure by Victor J. H. Trees.

Sketch of the solar and lunar disk geometry used in the eclipse-correction method — Solar and lunar disk geometry used to compute eclipse obscuration. Figure adapted from Trees et al. (2021).