SAGE
A stochastic weather generator producing unlimited, physically consistent synthetic hourly climate records for a single location.
SAGE is a stochastic weather generator that creates long, physically consistent synthetic records of hourly weather for a single location. Calibrated on a relatively short observed record, it can generate an unlimited number of equally plausible realizations, jointly simulating five correlated variables.
The synthetic series preserve the statistical character of the local climate at every scale: the distribution of each variable, the diurnal cycle, hour-to-hour and day-to-day persistence, the clustering of wet and dry spells, seasonal and longer-term variation, the behaviour of extremes, and the physical correlations between variables (such as the drop in sunshine and rise in humidity during rainfall).
- 01Precipitation
- 02Air temperature
- 03Global (solar) radiation
- 04Relative humidity
- 05Mean wind speed
Built on a latent-Gaussian framework of transformed and censored Gaussian processes, in the tradition of Papalexiou (2018) and Papalexiou & Serinaldi (2020) (CoSMoS). Each variable is driven by an underlying Gaussian process and mapped onto its observed distribution, so the statistical properties of the real climate are reproduced by construction.
- 01Papalexiou, S. M. (2018). Unified theory for stochastic modelling of hydroclimatic processes: Preserving marginal distributions, correlation structures, and intermittency. Advances in Water Resources, 115, 234–252. https://doi.org/10.1016/j.advwatres.2018.02.013
- 02Papalexiou, S. M., & Serinaldi, F. (2020). Random fields simplified: Preserving marginal distributions, correlations, and intermittency, with applications from rainfall to humidity. Water Resources Research, 56(2), e2019WR026331. https://doi.org/10.1029/2019WR026331
Sample stochastic realization · Hasselt · 8,760 hours
One ensemble member, played back at hourly resolution. Five co-evolving variables trace the seasons; scrub or pause to inspect any moment of the year.