pr-global_vs_HMA.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Projections GLOB vs HMA of Total Precipitation\n",
    "conda env: `phd_v3` (in `envs/phd`)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3.8.5 | packaged by conda-forge | (default, Jul 24 2020, 01:25:15) \n",
      "[GCC 7.5.0]\n"
     ]
    }
   ],
   "source": [
    "# To reload external files automatically (ex: utils)\n",
    "%load_ext autoreload\n",
    "%autoreload 2\n",
    "\n",
    "import xarray as xr\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import proplot as plot # New plot library (https://proplot.readthedocs.io/en/latest/)\n",
    "plot.rc['savefig.dpi'] = 300 # 1200 is too big! #https://proplot.readthedocs.io/en/latest/basics.html#Creating-figures\n",
    "from scipy import stats\n",
    "import xesmf as xe # For regridding (https://xesmf.readthedocs.io/en/latest/)\n",
    "\n",
    "import sys\n",
    "sys.path.insert(1, '/home/mlalande/notebooks/utils') # to include my util file in previous directory\n",
    "import utils as u # my personal functions\n",
    "u.check_python_version()\n",
    "# u.check_virtual_memory()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Set variables"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'Total Precipitation'"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "period_future = slice('2015','2100')\n",
    "period_past = slice('1979','2014')\n",
    "latlim, lonlim = u.get_domain_HMA()\n",
    "lonlim_HK, latlim_HK, lonlim_HM, latlim_HM, lonlim_TP, latlim_TP = u.get_zones()\n",
    "seasons = ['Annual', 'DJFMA', 'JJAS']\n",
    "# seasons = ['Annual']\n",
    "\n",
    "# HMA for full domain and the following for the above zones\n",
    "zones = ['GLOB', 'GLOB-land', 'NH-land', 'NH-land-60', 'HMA', 'HK', 'HM', 'TP']\n",
    "zones_df = pd.DataFrame(\n",
    "    [\n",
    "        [slice(None), slice(None)], \n",
    "        [slice(None), slice(None)], \n",
    "        [slice(None), slice(0,90)], \n",
    "        [slice(None), slice(0,60)], \n",
    "        [lonlim, latlim],\n",
    "        [lonlim_HK, latlim_HK],\n",
    "        [lonlim_HM, latlim_HM],\n",
    "        [lonlim_TP, latlim_TP]\n",
    "    ], \n",
    "    columns=pd.Index(['lonlim', 'latlim'], name='Limits'),\n",
    "    index=pd.Index(zones, name='Zones')\n",
    ")\n",
    "\n",
    "var = 'pr'\n",
    "table = u.get_table(var)\n",
    "label, units, levels, cmap, extend, \\\n",
    "    levels_diff, cmap_diff, extend_diff, \\\n",
    "    levels_bias, cmap_bias, extend_bias = u.get_var_infos(var)\n",
    "label"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load topography"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "ds = xr.open_dataset('GMTED2010_15n240_1000deg.nc').drop_dims('nbounds').swap_dims(\n",
    "    {'nlat': 'latitude', 'nlon': 'longitude'}).drop({'nlat', 'nlon'}).rename(\n",
    "    {'latitude': 'lat', 'longitude': 'lon'})\n",
    "elevation = ds.elevation\n",
    "elevation_std = ds.elevation_stddev"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load land/sea mask\n",
    "https://ldas.gsfc.nasa.gov/gldas/vegetation-class-mask"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "mask = xr.open_dataset('/data/mlalande/GLDASp5_landmask_10d.nc4').GLDAS_mask[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "BCC-CSM2-MR (1/10)\n",
      "Load past\n",
      "1.125°x1.112°\n",
      "Annual\n",
      "Reuse existing file: bilinear_160x320_180x360.nc\n",
      "DJFMA\n",
      "Reuse existing file: bilinear_160x320_180x360.nc\n",
      "JJAS\n",
      "Reuse existing file: bilinear_160x320_180x360.nc\n",
      "Load future\n",
      "ssp126\n",
      "Annual\n",
      "DJFMA\n",
      "JJAS\n",
      "ssp245\n",
      "Annual\n",
      "DJFMA\n",
      "JJAS\n",
      "ssp370\n",
      "Annual\n",
      "DJFMA\n",
      "JJAS\n",
      "ssp585\n",
      "Annual\n",
      "DJFMA\n",
      "JJAS\n",
      "\n",
      "\n",
      "CNRM-CM6-1 (2/10)\n",
      "Load past\n",
      "1.406°x1.389°\n",
      "Annual\n",
      "Reuse existing file: bilinear_128x256_180x360.nc\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Reuse existing file: bilinear_128x256_180x360.nc\n",
      "JJAS\n",
      "Reuse existing file: bilinear_128x256_180x360.nc\n",
      "Load future\n",
      "ssp126\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "ssp245\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "ssp370\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "ssp585\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "\n",
      "\n",
      "CNRM-CM6-1-HR (3/10)\n",
      "Load past\n",
      "0.5°x0.495°\n",
      "Annual\n",
      "Reuse existing file: bilinear_360x720_180x360.nc\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Reuse existing file: bilinear_360x720_180x360.nc\n",
      "JJAS\n",
      "Reuse existing file: bilinear_360x720_180x360.nc\n",
      "Load future\n",
      "ssp126\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "ssp245\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "ssp370\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "ssp585\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "\n",
      "\n",
      "CNRM-ESM2-1 (4/10)\n",
      "Load past\n",
      "1.406°x1.389°\n",
      "Annual\n",
      "Reuse existing file: bilinear_128x256_180x360.nc\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Reuse existing file: bilinear_128x256_180x360.nc\n",
      "JJAS\n",
      "Reuse existing file: bilinear_128x256_180x360.nc\n",
      "Load future\n",
      "ssp126\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "ssp245\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "ssp370\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "ssp585\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "\n",
      "\n",
      "CanESM5 (5/10)\n",
      "Load past\n",
      "2.812°x2.767°\n",
      "Annual\n",
      "Reuse existing file: bilinear_64x128_180x360.nc\n",
      "DJFMA\n",
      "Reuse existing file: bilinear_64x128_180x360.nc\n",
      "JJAS\n",
      "Reuse existing file: bilinear_64x128_180x360.nc\n",
      "Load future\n",
      "ssp126\n",
      "Annual\n",
      "DJFMA\n",
      "JJAS\n",
      "ssp245\n",
      "Annual\n",
      "DJFMA\n",
      "JJAS\n",
      "ssp370\n",
      "Annual\n",
      "DJFMA\n",
      "JJAS\n",
      "ssp585\n",
      "Annual\n",
      "DJFMA\n",
      "JJAS\n",
      "\n",
      "\n",
      "IPSL-CM6A-LR (6/10)\n",
      "Load past\n",
      "2.5°x1.268°\n",
      "Annual\n",
      "Reuse existing file: bilinear_143x144_180x360.nc\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Reuse existing file: bilinear_143x144_180x360.nc\n",
      "JJAS\n",
      "Reuse existing file: bilinear_143x144_180x360.nc\n",
      "Load future\n",
      "ssp126\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "ssp245\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "ssp370\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "ssp585\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "\n",
      "\n",
      "MIROC-ES2L (7/10)\n",
      "Load past\n",
      "2.812°x2.767°\n",
      "Annual\n",
      "Reuse existing file: bilinear_64x128_180x360.nc\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Reuse existing file: bilinear_64x128_180x360.nc\n",
      "JJAS\n",
      "Reuse existing file: bilinear_64x128_180x360.nc\n",
      "Load future\n",
      "ssp126\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "ssp245\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "ssp370\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "ssp585\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "\n",
      "\n",
      "MIROC6 (8/10)\n",
      "Load past\n",
      "1.406°x1.389°\n",
      "Annual\n",
      "Reuse existing file: bilinear_128x256_180x360.nc\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Reuse existing file: bilinear_128x256_180x360.nc\n",
      "JJAS\n",
      "Reuse existing file: bilinear_128x256_180x360.nc\n",
      "Load future\n",
      "ssp126\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "ssp245\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "ssp370\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "ssp585\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "\n",
      "\n",
      "MRI-ESM2-0 (9/10)\n",
      "Load past\n",
      "1.125°x1.112°\n",
      "Annual\n",
      "Reuse existing file: bilinear_160x320_180x360.nc\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Reuse existing file: bilinear_160x320_180x360.nc\n",
      "JJAS\n",
      "Reuse existing file: bilinear_160x320_180x360.nc\n",
      "Load future\n",
      "ssp126\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "ssp245\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "ssp370\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "ssp585\n",
      "Annual\n",
      "DJFMA\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data/mlalande/miniconda3/envs/phd_v3/lib/python3.8/site-packages/xarray/core/common.py:1123: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.\n",
      "The new arguments that you should use are 'offset' or 'origin'.\n",
      "\n",
      ">>> df.resample(freq=\"3s\", base=2)\n",
      "\n",
      "becomes:\n",
      "\n",
      ">>> df.resample(freq=\"3s\", offset=\"2s\")\n",
      "\n",
      "  grouper = pd.Grouper(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "JJAS\n",
      "\n",
      "\n",
      "UKESM1-0-LL (10/10)\n",
      "Load past\n",
      "1.875°x1.25°\n",
      "Annual\n",
      "Reuse existing file: bilinear_144x192_180x360.nc\n",
      "DJFMA\n",
      "Reuse existing file: bilinear_144x192_180x360.nc\n",
      "JJAS\n",
      "Reuse existing file: bilinear_144x192_180x360.nc\n",
      "Load future\n",
      "ssp126\n",
      "Annual\n",
      "DJFMA\n",
      "JJAS\n",
      "ssp245\n",
      "Annual\n",
      "DJFMA\n",
      "JJAS\n",
      "ssp370\n",
      "Annual\n",
      "DJFMA\n",
      "JJAS\n",
      "ssp585\n",
      "Annual\n",
      "DJFMA\n",
      "JJAS\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "list_models = u.get_model_names_projections()\n",
    "# list_models = list_models[:1]\n",
    "# list_models = ['EC-Earth3']\n",
    "\n",
    "resolution = [None]*len(list_models)\n",
    "\n",
    "da_past_ym_list = [[None] * (len(list_models)) for i in range(len(seasons))]\n",
    "\n",
    "scenarios = ['ssp126', 'ssp245', 'ssp370', 'ssp585']\n",
    "da_ssp126_ym_list = [[None] * (len(list_models)) for i in range(len(seasons))]\n",
    "da_ssp245_ym_list = [[None] * (len(list_models)) for i in range(len(seasons))]\n",
    "da_ssp370_ym_list = [[None] * (len(list_models)) for i in range(len(seasons))]\n",
    "da_ssp585_ym_list = [[None] * (len(list_models)) for i in range(len(seasons))]\n",
    "\n",
    "da_future_ym_list = [da_ssp126_ym_list, da_ssp245_ym_list, da_ssp370_ym_list, da_ssp585_ym_list]\n",
    "\n",
    "\n",
    "for i, model in enumerate(list_models):\n",
    "    print(model + ' ('+str(i+1)+'/'+str(len(list_models))+')')\n",
    "    institute, grid, realization, calendar = u.get_model_infos(model, var)\n",
    "    \n",
    "    #################\n",
    "    ### Load past ###\n",
    "    #################\n",
    "    print('Load past')\n",
    "    da_past = xr.open_mfdataset(\n",
    "        '/bdd/CMIP6/CMIP/'+institute+'/'+model+'/historical/'+realization+'/'+table+'/'+var+'/'+grid+'/latest/*.nc', \n",
    "        combine='by_coords').sel(time=period_past)[var].load()\n",
    "    \n",
    "    np.testing.assert_equal((int(period_past.stop) - int(period_past.start) + 1)*12, da_past.time.size)\n",
    "    \n",
    "    # pr (mm/s -> mm/day)\n",
    "    #################################################################################\n",
    "    \n",
    "    da_past = da_past * 86400\n",
    "    \n",
    "    #################################################################################\n",
    "    \n",
    "    # Print resolution and save it\n",
    "    lat = (da_past.lat[1] - da_past.lat[0]).values; lon = (da_past.lon[1] - da_past.lon[0]).values\n",
    "    print(str(lon.round(3))+'°x'+str(lat.round(3))+'°')\n",
    "    resolution[i] = str(lon.round(1))+'°x'+str(lat.round(1))+'°'\n",
    "    \n",
    "    for k, season in enumerate(seasons):\n",
    "        print(season)\n",
    "        # Make year_mean\n",
    "        da_past_ym = u.year_mean(da_past, calendar=calendar, season=season)\n",
    "\n",
    "        # Regrid\n",
    "        regridder = xe.Regridder(da_past, elevation, 'bilinear', periodic=False, reuse_weights=True)\n",
    "        da_past_ym_regrid = regridder(da_past_ym)\n",
    "\n",
    "        # Remove last year for Annual and summer to fit to winter\n",
    "        if season in ['Annual', 'JJAS']:\n",
    "            da_past_ym_regrid = da_past_ym_regrid.isel(year=slice(0,-1))\n",
    "        \n",
    "        # Store result\n",
    "        da_past_ym_list[k][i] = da_past_ym_regrid\n",
    "    \n",
    "    \n",
    "    \n",
    "    \n",
    "    ###################\n",
    "    ### Load future ###\n",
    "    ###################\n",
    "    print('Load future')\n",
    "    for j, scenario in enumerate(scenarios):\n",
    "        print(scenario)\n",
    "        if model in ['IPSL-CM6A-LR', 'MRI-ESM2-0']:\n",
    "            da_future = xr.open_mfdataset(\n",
    "                '/bdd/CMIP6/ScenarioMIP/'+institute+'/'+model+'/'+scenario+'/'+realization+'/'+table+'/'+var+'/'+grid+'/latest/*210012.nc', \n",
    "                combine='by_coords').sel(time=period_future)[var].load()\n",
    "        else:\n",
    "            da_future = xr.open_mfdataset(\n",
    "                '/bdd/CMIP6/ScenarioMIP/'+institute+'/'+model+'/'+scenario+'/'+realization+'/'+table+'/'+var+'/'+grid+'/latest/*.nc', \n",
    "                combine='by_coords').sel(time=period_future)[var].load()\n",
    "\n",
    "        np.testing.assert_equal((int(period_future.stop) - int(period_future.start) + 1)*12, da_future.time.size)\n",
    "        \n",
    "        # pr (mm/s -> mm/day)\n",
    "        #################################################################################\n",
    "\n",
    "        da_future = da_future * 86400\n",
    "\n",
    "        #################################################################################\n",
    "        \n",
    "        for k, season in enumerate(seasons):\n",
    "            print(season)\n",
    "            # Make year_mean\n",
    "            da_future_ym = u.year_mean(da_future, calendar=calendar, season=season)\n",
    "\n",
    "            # Regrid\n",
    "            da_future_ym_regrid = regridder(da_future_ym)\n",
    "            \n",
    "            # Remove last year for Annual and summer to fit to winter\n",
    "            if season in ['Annual', 'JJAS']:\n",
    "                da_future_ym_regrid = da_future_ym_regrid.isel(year=slice(0,-1))\n",
    "\n",
    "            # Store result\n",
    "            da_future_ym_list[j][k][i] = da_future_ym_regrid\n",
    "        \n",
    "    print('\\n')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# '/bdd/CMIP6/CMIP/'+institute+'/'+model+'/historical/'+realization+'/'+table+'/'+var+'/'+grid+'/latest'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# '/bdd/CMIP6/ScenarioMIP/'+institute+'/'+model+'/'+scenario+'/'+realization+'/'+table+'/'+var+'/'+grid+'/latest/'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Concatenate results by zones and scenarios"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Model past\n",
    "temp = [None]*len(seasons)\n",
    "for k, season in enumerate(seasons):\n",
    "    temp[k] = xr.concat(\n",
    "        da_past_ym_list[k][:], pd.Index(list_models, name='model'), coords='minimal', compat='override'\n",
    "    )\n",
    "mm_past_ym = xr.concat(temp, pd.Index(seasons, name=\"season\"))\n",
    "\n",
    "temp = [None]*len(zones) \n",
    "for i, zone in enumerate(zones):\n",
    "    if zone in ['HMA', 'HK', 'HM', 'TP']:\n",
    "        temp[i] = u.spatial_average(\n",
    "            mm_past_ym.sel(lat=zones_df.latlim[zone], lon=zones_df.lonlim[zone]).where(elevation > 2500)\n",
    "        )\n",
    "    elif zone == 'GLOB':\n",
    "        temp[i] = u.spatial_average(mm_past_ym)\n",
    "        \n",
    "    elif zone == 'GLOB-land':\n",
    "        temp[i] = u.spatial_average(mm_past_ym.where(mask == 1))\n",
    "        \n",
    "    elif zone in ['NH-land', 'NH-land-60']:\n",
    "        temp[i] = u.spatial_average(mm_past_ym.where(mask == 1).sel(lat=zones_df.latlim[zone], lon=zones_df.lonlim[zone]))\n",
    "        \n",
    "mm_past_ym_zones = xr.concat(temp, pd.Index(zones, name=\"zone\")).load()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# mm_past_ym_zones"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Model future\n",
    "temp_scenarios = [None]*len(scenarios)\n",
    "\n",
    "for j, scenario in enumerate(scenarios):\n",
    "    temp = [None]*len(seasons)\n",
    "    for k, season in enumerate(seasons):\n",
    "        temp[k] = xr.concat(\n",
    "            da_future_ym_list[j][k][:], pd.Index(list_models, name='model'), coords='minimal', compat='override'\n",
    "        )        \n",
    "    mm_future_ym = xr.concat(temp, pd.Index(seasons, name=\"season\")).load()\n",
    "\n",
    "    temp = [None]*len(zones) \n",
    "    for i, zone in enumerate(zones):\n",
    "        if zone in ['HMA', 'HK', 'HM', 'TP']:\n",
    "            temp[i] = u.spatial_average(\n",
    "                mm_future_ym.sel(lat=zones_df.latlim[zone], lon=zones_df.lonlim[zone]).where(elevation > 2500)\n",
    "            )\n",
    "        elif zone == 'GLOB':\n",
    "            temp[i] = u.spatial_average(mm_future_ym)\n",
    "            \n",
    "        elif zone == 'GLOB-land':\n",
    "            temp[i] = u.spatial_average(mm_future_ym.where(mask == 1))\n",
    "        \n",
    "        elif zone in ['NH-land', 'NH-land-60']:\n",
    "            temp[i] = u.spatial_average(mm_future_ym.where(mask == 1).sel(lat=zones_df.latlim[zone], lon=zones_df.lonlim[zone]))\n",
    "        \n",
    "        \n",
    "    temp_scenarios[j] = xr.concat(temp, pd.Index(zones, name=\"zone\")).load()\n",
    "    \n",
    "mm_future_ym_zones_scenarios = xr.concat(temp_scenarios, pd.Index(scenarios, name=\"scenario\")).load()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# mm_future_ym_zones_scenarios"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Save results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'pr'"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "var"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "mm_past_ym_zones.to_netcdf('results/projections_GLOB-HMA_'+var+'_'+period_past.start+'-'+period_past.start+'_mm_past_ym_zones.nc')\n",
    "mm_future_ym_zones_scenarios.to_netcdf('results/projections_GLOB-HMA_'+var+'_'+period_future.start+'-'+period_future.start+'_mm_future_ym_zones_scenarios.nc')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "list_models = u.get_model_names_projections()\n",
    "scenarios = ['ssp126', 'ssp245', 'ssp370', 'ssp585']\n",
    "\n",
    "mm_past_ym_zones = xr.open_dataarray('results/projections_GLOB-HMA_'+var+'_'+period_past.start+'-'+period_past.start+'_mm_past_ym_zones.nc')\n",
    "mm_future_ym_zones_scenarios = xr.open_dataarray('results/projections_GLOB-HMA_'+var+'_'+period_future.start+'-'+period_future.start+'_mm_future_ym_zones_scenarios.nc')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Plot"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# mm_future_ym_zones_scenarios\n",
    "# mm_past_ym_zones"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'tas'"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "var"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 320.833x316.528 with 1 Axes>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 339,
       "width": 310
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "season='Annual'\n",
    "colors = ['indigo9', 'blue5', 'orange5', 'red6']\n",
    "clim_ref = mm_past_ym_zones.sel(season=season, year=slice('1995','2014')).mean('year')\n",
    "\n",
    "fig, axs = plot.subplots()\n",
    "\n",
    "for i, scenario in enumerate(scenarios):\n",
    "    x = []\n",
    "    y = []\n",
    "    \n",
    "    y_start = 2015\n",
    "    y_end = 0\n",
    "    delta = 10\n",
    "    \n",
    "    while y_end < 2100-delta:\n",
    "        y_end = y_start + 20 - 1\n",
    "        \n",
    "        for model in list_models:\n",
    "            temp = (mm_future_ym_zones_scenarios.sel(season=season, year=slice(str(y_start),str(y_end)), scenario=scenario, model=model).mean('year') - clim_ref.sel(model=model)).squeeze()\n",
    "\n",
    "            x.append(temp.sel(zone='HMA').values)\n",
    "            y.append(temp.sel(zone='GLOB').values)\n",
    "        \n",
    "        y_start += delta\n",
    "    \n",
    "    # Plot\n",
    "    axs[0].scatter(x, y, color=colors[i], label=scenario, zorder=-i, marker='+')\n",
    "\n",
    "lim = (0,10)\n",
    "axs.format(\n",
    "    xlabel='HMA anomalies [°C]',\n",
    "    ylabel='Global anomalies [°C]',\n",
    "    xlim=lim, ylim=lim,\n",
    "    suptitle='Near-Surface Air Temperature anomalies\\n(relative to 1995–2014 average)'\n",
    ")\n",
    "\n",
    "axs.legend(loc='ul', frame=True, ncols=2)\n",
    "\n",
    "filename = 'projections/'+var+'_HMA_global'\n",
    "# fig.save('img/'+filename+'.jpg'); fig.save('img/'+filename+'.png'); fig.save('img/'+filename+'.pdf')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:phd_v3] *",
   "language": "python",
   "name": "conda-env-phd_v3-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}