Checking the spatio-temporal build up of the largest simulated events¶

In this notebook we will compare the UNSEEN data to the observations.

##Load pacakages
import xarray as xr
import matplotlib.pyplot as plt
import numpy as np
import cartopy
import cartopy.crs as ccrs
import pandas as pd
import calendar

##This is so variables get printed within jupyter
from IPython.core.interactiveshell import InteractiveShell 
InteractiveShell.ast_node_interactivity = "all"

### Load Data
dirname = r'/home/tike/'

Observed¶

Observations_pooled_monthly = xr.open_dataarray('../Data/Observations_pooled_monthly.nc')
Observations_pooled_monthly

xarray.DataArray

'Streamflow'

index: 360

9.682e+04 1.107e+05 1.258e+05 ... 2.177e+05 1.826e+05 1.214e+05

array([ 96820.165677, 110682.521733, 125771.958613, ..., 217669.384226,
       182610.150774, 121367.412367])

Coordinates: (1)

index

(index)

datetime64[ns]

1980-10-31 ... 2010-09-30

array(['1980-10-31T00:00:00.000000000', '1980-11-30T00:00:00.000000000',
       '1980-12-31T00:00:00.000000000', ..., '2010-07-31T00:00:00.000000000',
       '2010-08-31T00:00:00.000000000', '2010-09-30T00:00:00.000000000'],
      dtype='datetime64[ns]')

Attributes: (0)

The temporal characteristics of the floods¶

Now we will compare the flood seasonality between UNSEEN simulations and the observations. We plot the 2, 20 and 2000 year monthly flood values for UNSEEN and the 2, 20 and 52 year values for the observations.

The plot shows that there are large UNSEEN floods in July-September, that are not seen in the observations.

##Create a time array for indexing the time dimension
#The time axis is currently from 0:2000, I will change them to their original years 2035-2040 (copied from the precipitation dataset). 
# The runs are five years long, repeated 400 times (16 starts x 25 ensembles). 
# Time_year represents one run, that should be repeated 400 times to match the time dimension. 
time_year=np.arange('2035-01', '2040-01', dtype='datetime64[M]')
time_year

array(['2035-01', '2035-02', '2035-03', '2035-04', '2035-05', '2035-06',
       '2035-07', '2035-08', '2035-09', '2035-10', '2035-11', '2035-12',
       '2036-01', '2036-02', '2036-03', '2036-04', '2036-05', '2036-06',
       '2036-07', '2036-08', '2036-09', '2036-10', '2036-11', '2036-12',
       '2037-01', '2037-02', '2037-03', '2037-04', '2037-05', '2037-06',
       '2037-07', '2037-08', '2037-09', '2037-10', '2037-11', '2037-12',
       '2038-01', '2038-02', '2038-03', '2038-04', '2038-05', '2038-06',
       '2038-07', '2038-08', '2038-09', '2038-10', '2038-11', '2038-12',
       '2039-01', '2039-02', '2039-03', '2039-04', '2039-05', '2039-06',
       '2039-07', '2039-08', '2039-09', '2039-10', '2039-11', '2039-12'],
      dtype='datetime64[M]')

### Set extent and projection
plt.rcParams["font.family"] = "sans-serif" ##change font
plt.rcParams['font.size'] = 7  ## change font size (6 was used below)
plt.rcParams['svg.fonttype'] = 'none' ## so inkscape recognized texts in svg file
plt.rcParams['pdf.fonttype'] = 42 ## so illustrator can recognize text

# plt.figure(figsize=(90/25.4, 40/25.4))
ax = plt.axes()
Quantiles_obs = Observations_pooled_monthly.groupby("index.month").quantile([0,1-1/2, 1-1/20, 1], dim="index")
ax.fill_between(list(calendar.month_abbr[1:]), Quantiles_obs.sel(quantile=1-1/2), Quantiles_obs.sel(quantile=1-1/20), color='blue', alpha=0.2,label = '20-year')
ax.fill_between(list(calendar.month_abbr[1:]), Quantiles_obs.sel(quantile=0), Quantiles_obs.sel(quantile=1-1/2), color='blue', alpha=0.1,label = '2-year')

plt.plot(list(calendar.month_abbr[1:]),
         Amazon.assign_coords(
             time=np.tile(time_year, 400)).isel(
             time=slice(1567 * 12, 1568 * 12)
         ).values,
         marker='o',
         markersize = 4,
         label='largest simulation',
         color='orange'
        )

plt.plot(list(calendar.month_abbr[1:]),
         Observations_pooled_monthly.sel(
             index=slice('2009', '2009-12')
         ).values,
         marker='o',
         markersize = 4,
         label='2009',
         color='blue'
        )

plt.ylabel('Discharge [m3/s]')
plt.legend()
# plt.savefig('../Graphs/paper/Discharge_temporal_obs_flood1.pdf',bbox_inches='tight', dpi =300)       

<matplotlib.collections.PolyCollection at 0x2ad54d9e1e20>

<matplotlib.collections.PolyCollection at 0x2ad54d9e8580>

[<matplotlib.lines.Line2D at 0x2ad54d9e1e80>]

[<matplotlib.lines.Line2D at 0x2ad54d9fb070>]

Text(0, 0.5, 'Discharge [m3/s]')

<matplotlib.legend.Legend at 0x2ad54d9e8e80>

../_images/Spatial_Temporal_characteristics_11_6.png

Unseen or unrealistic?

Checking the spatio-temporal build up of the largest simulated events¶

Observed¶

UNSEEN¶

The temporal characteristics of the floods¶

The spatial characteristics¶