Assessment of meteorological extremes using a synoptic weather generator and a downscaling model based on analogues

Abstract

Abstract. Natural risk studies such as flood risk assessments require long series of weather variables. As an alternative to observed series, which have a limited length, these data can be provided by weather generators. Among the large variety of existing ones, resampling methods based on analogues have the advantage of guaranteeing the physical consistency between local weather variables at each time step. However, they cannot generate values of predictands exceeding the range of observed values. Moreover, the length of the simulated series is typically limited to the length of the synoptic meteorological records used to characterize the large-scale atmospheric configuration of the generation day. To overcome these limitations, the stochastic weather generator proposed in this study combines two sampling approaches based on atmospheric analogues: (1) a synoptic weather generator in a first step, which recombines days of the 20th century to generate a 1000-year sequence of new atmospheric trajectories, and (2) a stochastic downscaling model in a second step applied to these atmospheric trajectories, in order to simulate long time series of daily regional precipitation and temperature. The method is applied to daily time series of mean areal precipitation and temperature in Switzerland. It is shown that the climatological characteristics of observed precipitation and temperature are adequately reproduced. It also improves the reproduction of extreme precipitation values, overcoming previous limitations of standard analogue-based weather generators.

Comprehensive evaluation of an improved large‐scale multi‐site weather generator for Germany

“…To cover larger spatial scales than those characterized by a single location, multi-site weather generators have been proposed in the recent decades. Multi-site stochastic models are based on (a) resampling techniques, such as analogue (Zorita and von Storch, 1999;Chardon et al, 2018;Raynaud et al, 2020) and nearest-neighbour (Beersma and Buishand, 2003;Caraway et al, 2014), (b) point-process simulation of rainfall in combination with Markov chain simulation of precipitation occurrence (Fowler et al, 2005;Cowpertwait, 2006) and (c) time series generation using the Markov-chain approach in combination with frequency distributions (Wilks, 1998;Breinl et al, 2013;Keller et al, 2015) to simulate occurrence and magnitude, respectively. Furthermore, latent Gaussian variable models in combination with auto-regressive approach considering the spatial correlation structure are employed in Bárdossy and Plate (1992), Hundecha et al (2009), Kleiber et al (2012), Rasmussen (2013) and Bennett et al (2018).…”

“…Sharif and Burn (2007) propose a perturbation technique in combination with the k-nearest resampling while still retaining the spatial correlation structure and major statistical properties of at-site precipitation. Finally, Raynaud et al (2020) propose a weather generator based on constructing plausible atmospheric trajectories, that is, series of atmospheric states, based on analogues in combination with distribution adjustments to generate unobserved, but plausible series of precipitation and temperature.…”

“…In a multi‐site approach, observations at all sites (climate stations or grid points) are reshuffled for different time steps (e.g., days or hours). Thus, resampling techniques fully preserve spatial correlations between various locations within a single time‐step and thus can be easily applied to both small scales as well as to large river basins (Beersma and Buishand, 2003; Raynaud et al ., 2020). Standard resampling techniques are generally not able to produce single‐day extreme precipitation beyond those observed in the past.…”

“…Finally, Raynaud et al . (2020) propose a weather generator based on constructing plausible atmospheric trajectories, that is, series of atmospheric states, based on analogues in combination with distribution adjustments to generate unobserved, but plausible series of precipitation and temperature.…”

Deterministic Chaos Detection and Simplicial Local Predictions Applied to Strawberry Production Time Series

“…Since the description of the method of analogs by Lorenz [26], this method has gained popularity for forecasting and has been applied in many studies [27][28][29][30], offering even more accurate results than other approaches that apply machine learning techniques [31,32].…”