AWE-GEN-2d Nadav Peleg, Simone Fatichi, Athanasios Paschalis, Peter Molnar, Paolo Burlando
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
AWE-GEN-2d Nadav Peleg, Simone Fatichi, Athanasios Paschalis, Peter Molnar, Paolo Burlando SWGEN-Hydro Berlin, 2017
Outline 1. Motivation and general overview 2. AWE-GEN-2d structure 3. Short demo 4. Advantages and weaknesses – from model validation 5. Applications 6. Next steps
Motivation The potential to narrow uncertainty in projections of regional precipitation change (Clim Dyn) Local scale Local scale Hawkins and Sutton (2011)
Motivation Uncertainty partition challenges the predictability of vital details of climate change (Earth’s Future) Local scale Natural climate variability is important also at small- local-scales! Fatichi et al. (2016)
Motivation Some impact studies require high-resolution climate data, e.g. for estimating streamflow in Alpine catchments (snow melt, glacier retreat, soil properties). Gap – to the best of our knowledge, no sub-daily sub-kilometer stochastic weather generator exists. Solution – Yes, another weather generator!
AWE-GEN-2d in a nutshell • AWE-GEN-2d (Advanced WEather GENerator for 2-Dimensional grid) follows the philosophy of combining physical and stochastic approaches to generate gridded climate variables in a high spatial and temporal resolution.
AWE-GEN-2d in a nutshell • AWE-GEN-2d (Advanced WEather GENerator for 2-Dimensional grid) follows the philosophy of combining physical and stochastic approaches to generate gridded climate variables in a high spatial and temporal resolution. • It is relatively fast and parsimonious in terms of computational demand. • It allows generating many stochastic realizations of present and future climates.
Ancestors AWE-GEN-2d model Peleg et al. (2017) HiReS-WG model Peleg and Morin (2014)
Climate variables • Sub-daily temporal resolution. • Similar output as of AWE-GEN model, but meteorological variables are spatially distributed: Precipitation Cloud cover Temperature Radiation (incoming longwave, incoming shortwave, diffusive) Relative humidity Vapor pressure Atmospheric pressure Wind speed
Model structure • Composed of eight modules. • There is dependency between most, but not all, meteorological variables. LWR
Storm mean areal statistics Paschalis et al. (2013)
Storm mean areal statistics • WAR-IMF-CAR process • Matérn cross-covariance function for tri-variate random time-series (using the Parsimonious Multivariate Matérn model, Gneiting et al., 2010). Paschalis et al. (2013)
Precipitation • The simulated WAR and IMF time series for each storm are transformed into the space- time evolution of the precipitation fields using latent Gaussian fields (simulated using 2-D FFT method). Nerini et al. (2017)
Precipitation • The fields takes into account the exponential decay of the spatial autocorrelation (spectral density analysis) and the ARMA process (temporal correlation). Peleg et al. (2017c)
Precipitation and advection • Due to the symmetries of the fast Fourier transform the generated fields can be folded. Paschalis et al. (2013)
Interitmtent precipitation fields • Lognormal distribution (Paschalis et al., 2013): , , = −1 ( , , ) , , It looks like rainfall! ( ) = But… precipitation are ( ) 2 + 1 = 2 + 1 homogenous at space • Weibull distribution (Skinner et al., 2017): , , = −1 ( , , ) , , 2 Γ 1+ ( )2 2 + 1 = 1 2 Γ 1+ ( ) = 1 Γ 1 + ( ) Paschalis et al. (2013)
Precipitation occurrence filter
U[G(x,y)] R(x,y) Precipitation UWAR=0.5[G(x,y)] UWAR=0.5[G*(x,y)] occurrence filter
Precipitation intensity filter Quantile [-] Direct link to GCM/RCM Extreme rainfall intensity are overestimated (at sub-hourly scales) Daily rain [mm]
Cloud cover • During the intra-storm period, the cloudiness and rainfall fields are cross-correlated and cloud cover has always an equal or larger extent than the precipitation field. • During an inter-storm period, the existence of a ‘‘fair weather’’ region is assumed, meaning that the cloudiness decreases as a function of the time passed since the end of the previous storm and increases toward the beginning of the next storm, using a two term exponential function. • The stochastic component of the cloud cover series during an inter-storm period is simulated through an ARMA model.
Near-surface air temperature • Air temperature is generated with an hourly time step first for a given reference level and then spatially extrapolated to all grid cells using a stochastic lapse rate. • The deterministic temperature component is assumed to be directly related to the incoming longwave radiation and to the hourly position of the Sun and site geographic location. • The stochastic temperature component is estimated through an autoregressive model AR(1). • The air temperature at a given reference level can be generated with or without considering the shading effect of the terrain. The shadow effect is calculated as a binary coefficient: the sloping surface is shadowed by neighboring terrain (==0) when the horizon angle is lower than the zenith angle for a given solar azimuth.
Near-surface wind speed • Near-surface wind speed is computed from the geostrophic wind speed. • The near-surface wind speed is calculated using the geostrophic drag law (mass conservation component is not implemented), which requires a simplified computation of the Monin-Obukhov length, the friction velocity per grid cell, and the PBL height. • The atmospheric stability is determined through the Pasquill stability classes (from extreme unstable to very stable atmospheric conditions), which is determined based on the shortwave incoming solar radiation and the cloud cover.
Case study: Engelberger catchment Domain size: 25 km by 25 km Elevation: 404 m to 3238 m Luzern Pilatus Altdorf Engelberg Titlis
Rainfall Fields Spatial resolution of 2 km by 2 km Temporal resolution of 5 min * Interpolated to a 100 m by 100 m grid for this example
High intensity cell evolving in time and space 00:20 pm Advection Rainfall intensity [mm h-1] 5 m s-1
High intensity cell evolving in time and space 00:25 pm Rainfall intensity [mm h-1]
High intensity cell evolving in time and space 00:30 pm Rainfall intensity [mm h-1]
High intensity cell evolving in time and space 00:35 pm Rainfall intensity [mm h-1]
High intensity cell evolving in time and space 00:40 pm Rainfall intensity [mm h-1]
Temperature Fields Spatial resolution of 100 m by 100 m Temporal resolution of 1 hour
06:00 am Temperature [°C] Jun, 17th Engelberg Titlis
07:00 am Temperature [°C] Jun, 17th Engelberg Titlis
08:00 am Temperature [°C] Jun, 17th
09:00 am Temperature [°C] Jun, 17th
10:00 am Temperature [°C] Jun, 17th
11:00 am Temperature [°C] Jun, 17th
12:00 pm Mean cloud cover for June, 17: 0.37 Mean cloud cover for June, 18: 0.13 Temperature [°C] Jun, 17th
12:00 pm Mean cloud cover for June, 17: 0.37 Mean cloud cover for June, 18: 0.13 Temperature [°C] Jun, 18th
01:00 pm Temperature [°C] Jun, 18th
02:00 pm Temperature [°C] Jun, 18th
03:00 pm Temperature [°C] Jun, 18th
04:00 pm Temperature [°C] Jun, 18th
05:00 pm Temperature [°C] Jun, 18th
06:00 pm Temperature [°C] Jun, 18th
Shortwave Incoming Radiation Fields Spatial resolution of 100 m by 100 m Temporal resolution of 1 hour
06:00 am Incoming radiation [W m-2] Jun, 17th Engelberg Titlis
Incoming radiation [W m-2] Jun, 17th 07:00 am
Incoming radiation [W m-2] Jun, 17th 08:00 am
Incoming radiation [W m-2] Jun, 17th 09:00 am
Incoming radiation [W m-2] Jun, 17th 10:00 am
Incoming radiation [W m-2] Jun, 17th 11:00 am
Incoming radiation [W m-2] Jun, 17th 12:00 pm
Incoming radiation [W m-2] Jun, 17th 01:00 pm
Incoming radiation [W m-2] Jun, 17th 02:00 pm
Incoming radiation [W m-2] Jun, 17th 03:00 pm
Incoming radiation [W m-2] Jun, 17th 04:00 pm
Incoming radiation [W m-2] Jun, 17th 05:00 pm
Incoming radiation [W m-2] Jun, 17th 06:00 pm
Near-Surface Wind Speed (2-m) Spatial resolution of 100 m by 100 m Temporal resolution of 1 hour
Wind speed [m s-1]
Inputs
Precipitation Advantages and weaknesses • Spatial patterns are well represented • Rainfall at daily scale is well reproduced • Model overestimate extreme rainfall intensity at sub-daily resolution • Obs.: 30 years of data • Sim.: 50 realizations, each of 30 years
Temperature Advantages and weaknesses • Hourly temperatures and dynamics are well represented • Thermal inversion is simulated but is not validated
Shortwave radiation Advantages and weaknesses
Vapor pressure Advantages and weaknesses • In general good representation • Hourly dynamics are not fully capture
Wind speed Advantages and weaknesses • Wind speed are not well represented at higher elevations
Applications High spatio-temporal resolution climate scenarios for snowmelt modelling in small alpine catchments Schirmer et al. (ongoing)
Applications Melt and surface sublimation across a glacier in a dry environment: distributed energy- balance modelling of Juncal Norte Glacier, Chile Ayala et al. (2017)
Applications • Spatial variability of extreme rainfall at radar subpixel scale (Peleg et al., 2016) • Partitioning the impacts of spatial and climatological rainfall variability in urban drainage modeling (Peleg et al., 2017) • Rainfall-discharge-sediment yield uncertainty cascade in a second generation landscape evolution models (Skinner et al.) • Simulating the effect of check dams on landscape evolution at centennial time scales (Ramirez et al.)
Applications • Spatial variability of extreme rainfall at radar subpixel scale (Peleg et al., 2016) • Partitioning the impacts of spatial and climatological rainfall variability in urban drainage modeling (Peleg et al., 2017) • Rainfall-discharge-sediment yield uncertainty cascade in a second generation landscape evolution models (Skinner et al.) • Simulating the effect of check dams on landscape evolution at centennial time scales (Ramirez et al.)
Future development • Re-parameterization AWE-GEN-2d for future climate • Diurnal cycle for precipitation (asymmetry?) • Coupling precipitation with temperature (extremes) Future • Replacing/improving wind estimates 2021-2030 2081-2090 Present
Thank you for your attention! Dr. Nadav Peleg Research Associate Hydrology and Water Resources Management Institute of Environmental Engineering (IfU) ETH Zurich HIL D21.1 Stefano-Franscini-Platz 5 8093 Zurich, Switzerland Office +41 44 633 27 17 nadav.peleg@sccer-soe.ethz.ch AWE-GEN-2d paper: Peleg, N., Fatichi, S., Paschalis, A., Molnar, P., and Burlando, P. (2017): An advanced stochastic weather generator for simulating 2-D high resolution climate variables. Journal of Advances in Modeling Earth Systems, 9, p. 1595-1627.
You can also read
