FEATURED TOOL
CoSMoS: Complete Stochastic Modelling System
CoSMoS is an award-winning open-source R package that generates realistic synthetic time series and spatial fields of any hydroclimatic variable (precipitation, streamflow, wind speed, humidity) in seconds. You define the statistical properties you need — CoSMoS produces the ensemble.
- Univariate, multivariate, and random field simulation of any hydroclimatic variable
- Preserves marginal distributions, correlation structure, and intermittency across scales
- Space-time storm simulation including advection from local rainfall to cyclone-scale fields
Maintained by TycheLab — Papalexiou, Serinaldi, Shook
35,000+ downloads on CRAN since 2019
Recognition ▼
- 2022 STAHY Best Paper Award — International Commission of Statistical Hydrology (ICSH) — for the unified stochastic modelling framework. [Paper]
- Water Resources Research 2020 Editors' Choice Award (AGU, awarded 2022) — for the random field simplification and simulation approach. [Paper]
- AGU Eos Research Spotlight (2021) — Improving Weather Simulations Through Increased Generality. Eos ↗
Core papers — what each one solved ▼
- A unified framework for stochastic modelling of hydroclimatic processes — preserving marginal distributions, correlation structures, and intermittency simultaneously.
- Precise temporal disaggregation of time series while preserving marginals and correlations, for both stationary and nonstationary processes.
- A simplified approach to random field simulation that preserves marginal distributions, spatial correlations, and intermittency — applied from rainfall to humidity fields.
- Space-time simulation of random fields at scale — extending the framework to capture storm dynamics, cyclone structures, and advection.
- Revisiting rainfall generation: CoSMoS-2s introduces copula-based intermittent time series modelling with improved realism at sub-daily scales.
- Operational blueprint for large-domain multisite precipitation generation — demonstrated at 1,000 simultaneous sites.
