Precipitation:
Gridded netCDF:
hydrological model
For each RP, cell:
Statistical RP analysis
hydraulic model
For each RP, cell:
(multiple simulations)
Validation for
RegCM
We have access to several Italian observational datasets provided by the University of L'Aquila for:
~2000-2017
only!
Some examples...
Outliers
(by applying first-stage filtering procedures only)
Before | After | Diff | |
---|---|---|---|
Total valid values | 250M | 243M | -2.6% |
Total flags | 324468 | 58008 | -82.1% |
pr > Mean + 20SD | 3240 | 2538 | -21.7% |
pr > Median + 20IQR | 49753 | 22519 | -54.7% |
pr > 100 mm/h | 221822 | 711 | -99.6% |
Other softer flags | 49646 | 32240 | -35.1% |
FILTERED
ORIGINAL
Statistical analysis helps, but a longer time period is required
EURO-CORDEX Simulations (thanks James Ciarlo`!):
Available datasets:
E-OBS
REGCM
HR-OBS
CHyM Is a distributed (gridded) hydrological model. Peculiarities:
Creating the river network is sometimes easier said than done...
Real Po river
CHyM simulation 1
CHyM simulation 2
Sanity check:
can we reproduce river basins?
Sanity check: skill of the model driven with the observations
Sanity check: how does the model assimilate RegCM's precipitation?
How does the model perform if driven with RegCM, compared with if driven with observations? (Liguria example)
CHyM (RegCM-driven)
CHyM (OBS-driven)
How does the model perform if driven with RegCM, compared with if driven with observations? (Central-South Italy example)
How to estimate hundred-years floods with only a few (~20) years worth of data?
The methodology is taken from Maione et al., 2003
Annual maxima
Gumbel extreme value distribution
Fit parameters
SDH: "Typical" flood timing curve for each river cell
CA2D input data
Precipitation:
Gridded netCDF:
hydrological model
For each RP, cell:
Statistical RP analysis
For each RP, cell:
(multiple simulations)
Validation for
CA2D model
afantini@ictp.it