{
"cells": [
{
"cell_type": "markdown",
"id": "8d1f3060-1bba-4336-bbd0-e074a80ed0e7",
"metadata": {},
"source": [
"# TerraFirma UKESM long-term stabilisation and overshoot scenarios"
]
},
{
"cell_type": "markdown",
"id": "ae3d7093-ab76-4465-bd05-c5dfd14239e8",
"metadata": {},
"source": [
"#### Steps to do\n",
"- decide how the bias correction should be applied\n",
" - according to Tom Mitcham: simulation years 10 to 50 of the ramp-up phase are most equal to 1975-2025 in terms of warming trend\n",
" - global mean temperature is similar (+0.26°C), but there is another period (simulation year 0 to 40 which would fit better\n",
" - [ ] double-check the warming trend?\n",
"- [ ] create a `process_terrafirma_ukesm_data()`,\n",
"- [ ] test with HEF using W5E5 or ERA5, RGI7?, regional spinup ...\n",
" - do they need that we start from pre-industrial state, or would it be ok, if we just start at simulation year 30 (which would correspond best to a \"climate\" as we had around year 2000, which is around RGI/glavier volume inventory year ... I guess the problem is that their ice sheet / sea-level rise output will be relative to \"pre-industrial\", and therefore, we also need the glaciers to have the \"pre-industrial\" baseline...? \n",
"- [ ] do I need to do sth. about the growing out of border issue ...https://github.com/OGGM/oggm/issues/1822 \n",
" "
]
},
{
"cell_type": "code",
"execution_count": 65,
"id": "6d0764b7-cede-4d23-b4e2-82768d9b2434",
"metadata": {},
"outputs": [],
"source": [
"import xarray as xr\n",
"xr.set_options(use_new_combine_kwarg_defaults=True)\n",
"import numpy as np\n",
"import matplotlib.pyplot as plt"
]
},
{
"cell_type": "markdown",
"id": "7fac9731-8e55-4db6-86dc-342e81ab93bd",
"metadata": {},
"source": [
"- \"esm-up2p0\" is the \"ramp-up\" phase.\n",
"- \"esm-up2p0-gwl2p0\" is stabilisation at 2C \n",
"- \"esm-up2p0-gwl2p0-50y-dn2p0\" is a \"ramp-down\" from the 2C stabilisation after 50 years'\n",
"- \"esm-up2p0-gwl4p0\" is stabilisation at 4C\n",
"- \"esm-up2p0-gwl4p0-50y-dn2p0\" is a \"ramp-down\" from the 4C stabilisation after 50 years'\n",
"- \"esm-hist\" is the same model (UKESM1.2-LL) being forced through the CMIP historical forcing'\n",
"- \"esm-piControl\" is the same model with pre-industrial forcing'"
]
},
{
"cell_type": "code",
"execution_count": 66,
"id": "c79ac93c-beb3-4686-b36b-947e0b53f077",
"metadata": {},
"outputs": [],
"source": [
"variables = ['pr','tas']\n",
"period = '205001-213912'\n",
"scenarios = ['hist', # same model (UKESM1.2-LL) being forced through the CMIP historical forcing\n",
" 'up2p0', # ramp-up phase\n",
" 'up2p0-gwl2p0', # stabilisation at 2°C\n",
" 'up2p0-gwl2p0-50y-dn2p0', # \"ramp-down\" from 2C stabilisation after 50 years'\n",
" 'up2p0-gwl4p0', # stabilisation at 4C\n",
" \"up2p0-gwl4p0-50y-dn2p0\", # \"ramp-down\" from 4C stabilisation after 50 years'\n",
" \"piControl\" # same model with pre-industrial forcing'\n",
" ]"
]
},
{
"cell_type": "code",
"execution_count": 67,
"id": "0449132e-6d07-452f-806d-cdb688dc8ba8",
"metadata": {},
"outputs": [],
"source": [
"time_coder = xr.coders.CFDatetimeCoder(use_cftime=True)\n",
"run = False\n",
"if run:\n",
" for var in ['pr','tas']:\n",
" _dt = xr.open_mfdataset(f'terrafirma_data/raw_data/{var}_Amon_UKESM1-2-LL_esm-{scen}_r1i1p1f1_gn_*.nc', \n",
" decode_times=time_coder).load()\n",
" if var == 'tas':\n",
" _dt['tas'] = _dt.tas\n",
" _dt.tas.attrs['units'] = '°C'\n",
" else:\n",
" # already in correct format... \n",
" assert 'kg m-2 s-1' in _dt.pr.units, 'Precip units not understood'\n",
" _dt = _dt.drop_vars(['time_bnds','lat_bnds', 'lon_bnds'])\n",
" y0=_dt.time.dt.year.values[0]\n",
" y1=_dt.time.dt.year.values[-1]\n",
" _dt.to_netcdf(f'terrafirma_data/{var}_Amon_UKESM1-2-LL_esm-{scen}_r1i1p1f1_gn.nc')"
]
},
{
"cell_type": "code",
"execution_count": 91,
"id": "6f34a523-934c-41df-9114-a6e6d19dc386",
"metadata": {},
"outputs": [],
"source": [
"run = False\n",
"if run:\n",
" for var in ['pr','tas']:\n",
" time_coder = xr.coders.CFDatetimeCoder(use_cftime=True)\n",
" global_var_mean_dict = {}\n",
" for scen in scenarios:\n",
" _dt = xr.open_dataset(f'terrafirma_data/{var}_Amon_UKESM1-2-LL_esm-{scen}_r1i1p1f1_gn.nc',\n",
" decode_times=time_coder).groupby('time.year').mean().load()\n",
" ### check the global mean temperature and precipitation time series\n",
" # weighted average over latitudes\n",
" if var == 'tas':\n",
" _dt[var] = _dt[var]-273.15\n",
" weight = np.cos(np.deg2rad(_dt.lat))\n",
" weight = weight / weight.sum()\n",
" global_var_mean_dict[scen] = (_dt.mean(dim='lon') * weight).sum(dim='lat')\n",
" keys = list(global_var_mean_dict.keys())\n",
" vals = list(global_var_mean_dict.values())\n",
"\n",
" ds_global_var_mean = xr.concat(vals, dim=\"scenario\", join='outer')\n",
" ds_global_var_mean = ds_global_var_mean.assign_coords(scenario=keys)\n",
" ds_global_var_mean.to_netcdf(f'terrafirma_global_annual_mean_{var}_scenarios.nc')\n",
"ds_global_tas_mean = xr.open_dataset('terrafirma_global_annual_mean_tas_scenarios.nc')\n",
"ds_global_pr_mean = xr.open_dataset('terrafirma_global_annual_mean_pr_scenarios.nc')\n",
"\n",
"ds_w5e5 = xr.open_dataset('/home/www/oggm/climate/gswp3-w5e5/unflattened/monthly/gswp3-w5e5_obsclim_tas_global_monthly_1901_2019.nc')\n",
"ds_w5e5 = ds_w5e5.groupby('time.year').mean().load()\n",
"ds_w5e5['tas'] = ds_w5e5['tas'] -273.15\n",
"weight = np.cos(np.deg2rad(ds_w5e5.lat))\n",
"weight = weight / weight.sum()\n",
"ds_global_tas_mean_w5e5 = (ds_w5e5.mean(dim='lon') * weight).sum(dim='lat')"
]
},
{
"cell_type": "code",
"execution_count": 93,
"id": "efdea528-12f3-44d4-9ccd-6ebe1506de08",
"metadata": {},
"outputs": [
{
"data": {
"text/html": [
"
GSWP3-W5E5 observational climate input data for ISIMIP3a
institution :
Potsdam Institute for Climate Impact Research (PIK)
project :
Inter-Sectoral Impact Model Intercomparison Project phase 3a (ISIMIP3a)
contact :
ISIMIP cross-sectoral science team <info@isimip.org> <https://www.isimip.org>
summary :
GSWP3 v1.09 bias-adjusted to W5E5 v2.0 with ISIMIP3BASD v2.5.0 (for 1901-1978) combined with W5E5 v2.0 (for 1979-2019)
references :
Kim (2017) <https://doi.org/10.20783/DIAS.501> for GSWP3; Cucchi et al. (2020) <https://doi.org/10.5194/essd-2020-28> and Lange et al. (2021) <https://doi.org/10.48364/ISIMIP.342217> for W5E5; Lange (2019) <https://doi.org/10.5194/gmd-12-3055-2019> and Lange (2021) <https://doi.org/10.5281/zenodo.4686991> for ISIMIP3BASD
"
],
"text/plain": [
" Size: 8B\n",
"array(39.)\n",
"Coordinates:\n",
" height float64 8B 1.5"
]
},
"execution_count": 72,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"rolling_40yr_ramp = ds_global_tas_mean.sel(scenario='up2p0').tas.rolling(year=40, center=False).mean().dropna(dim='year') - ref_period_1975_2014\n",
"year_match = np.abs(rolling_40yr_ramp).idxmin()\n",
"year_match-1850\n",
"# when just checking the global average temperature using 1850 to 1889 works best,"
]
},
{
"cell_type": "markdown",
"id": "ce2b0828-0854-4239-8e0c-813b8f9b2193",
"metadata": {},
"source": [
"- In my opinion, year 0 to 40 of the ramp up scenario (up2p0) fits best to the historical 1975-2014 period in terms of both, the warming trend and the global mean temperature. However, according to the paper it is year 10 to 50. "
]
},
{
"cell_type": "markdown",
"id": "a2a628ac-292b-4399-895a-b94124b5aba3",
"metadata": {},
"source": [
"----\n",
"- Tom Mitcham: the years 10-50 of the \"ramp-up\" is nearest to the years 1975-2015 of the \"esm-hist\" simulation, at least when comparing global average warming across a set time-period compared to the pre-industrial simulation.\n",
"- referred paper: \"We therefore compare the evolution of the UKESM1.2 CMIP6 historical runs and the TIPMIP ramp-up runs over a common GMSAT space, namely for the 40-year period of observed warming from 0.2°C to 1.0°C and the same 40-year warming period in the ramp-up run. This equates to 1975 to 2015 in the historical runs and years 10 to 50 of the TIPMIP ramp-up.\"\n",
"---"
]
},
{
"cell_type": "code",
"execution_count": 99,
"id": "b747662d-27d5-4131-aa17-913e3d982b97",
"metadata": {},
"outputs": [
{
"data": {
"text/html": [
"