{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "a333afcc",
   "metadata": {},
   "source": [
    "# PartB - Fig. XX  (a, b; portrait): glacier model differences in steady-state mass and in climate scenario sensitivity\n",
    "\n",
    "Difference of every glacier model to the median model in every RGI region (Global on top, then the regions sorted by their 2020 glacier mass as in GMIP3-comm Fig. 2), for the temperature bin 2.0°C to 4.0°C. One row per glacier model: **dot = median** over the 31 climate scenarios, **line = likely range** (17th to 83rd percentile of the 31 climate scenarios). The figure is made for a width of 8.3cm\n",
    "- `a`: difference of the steady-state mass to the median model\n",
    "- `b`: change of that difference relative to the climate of 1995-2014 (climate scenario sensitivity)\n",
    "\n",
    "We also looked at the other temperature bins, but don't show them in the manuscript."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "aa10452e",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import os\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "a86395fe",
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.rcParams[\"pdf.use14corefonts\"] = False  # don't convert text to path\n",
    "plt.rcParams['pdf.fonttype'] = 42  # keep the text editable in the pdf"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1563c2ed",
   "metadata": {},
   "source": [
    "## Settings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "37930367",
   "metadata": {},
   "outputs": [],
   "source": [
    "fn_steady_state = 'PartB_data/steady_state_relative_volume_change.csv'\n",
    "temp_bin = [2.0, 4.0]            # global warming bin in °C\n",
    "# temp_bin = [4.0, 7.0]            # global warming bin in °C\n",
    "# temp_bin = [-0.1, 2.0]            # global warming bin in °C\n",
    "\n",
    "likely_range = [17, 83]          # percentiles over the climate scenarios\n",
    "ref_period = '1995-2014_hist'    # reference climate for the climate scenario sensitivity (Fig. 2)\n",
    "\n",
    "# panels from left to right: 'a' (steady-state mass), 'b' (climate scenario sensitivity); ['b] \n",
    "panel_order = ['a', 'b']\n",
    "\n",
    "# difference to which median?\n",
    "#  'all_models': median of all glacier models in the region (= the existing 'delta ... to median model' columns, as in the first versions of Fig. 1 and 2)\n",
    "#  'global4': median of the four global models only (the same models in every region)\n",
    "reference = 'all_models'\n",
    "\n",
    "fig_width_cm, fig_height_cm = 8.3, 17   # one-column figure (before: 11.5, 19 )\n",
    "x_lims = {'a': (-40, 65), 'b': (-65, 50)}      # % points\n",
    "x_ticks = {'a': np.arange(-40, 61, 20), 'b': np.arange(-60, 41, 20)}\n",
    "fs = 6                                   # smallest font size in the figure\n",
    "\n",
    "fig_dir = 'figures_partB/glacier_model_ss_differences/'\n",
    "os.makedirs(fig_dir, exist_ok=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6a0fab0d",
   "metadata": {},
   "source": [
    "## Names, colours and order of the regions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "b2ab1e70",
   "metadata": {},
   "outputs": [],
   "source": [
    "reg_names = {'All': 'Global', '01': 'Alaska', '02': 'W Canada & US', '03': 'Arctic Canada N', '04': 'Arctic Canada S',\n",
    "             '05': 'Greenland Periphery', '06': 'Iceland', '07': 'Svalbard & Jan Mayen', '08': 'Scandinavia',\n",
    "             '09': 'Russian Arctic', '10': 'North Asia', '11': 'Central Europe', '12': 'Caucasus & Middle East',\n",
    "             '13': 'Central Asia', '14': 'South Asia W', '15': 'South Asia E', '16': 'Low Latitudes',\n",
    "             '17': 'Southern Andes', '18': 'New Zealand', '19': 'Subantarctic & Antarctic Islands'}\n",
    "# names with a line break where a name is long\n",
    "reg_names_lines = dict(reg_names)\n",
    "reg_names_lines.update({'05': 'Greenland\\nPeriphery', '07': 'Svalbard &\\nJan Mayen', '12': 'Caucasus &\\nMiddle East',\n",
    "                        '19': 'Subantarctic &\\nAntarctic Islands'})\n",
    "\n",
    "# Global on top, then sorted by the 2020 glacier mass (descending) as in Fig. 2 of GMIP3-comm\n",
    "reg_order = ['All', '19', '03', '01', '05', '09', '04', '07', '17', '06', '13', '14', '02', '15', '08', '10', '11', '16', '18', '12']\n",
    "\n",
    "global_models = ['PyGEM-OGGM', 'GloGEMflow', 'OGGM', 'GLIMB']\n",
    "regional_models = ['GloGEMflow3D', 'Kraaijenbrink', 'GO', 'CISM2']\n",
    "model_names_csv = {'PyGEM-OGGM_v13': 'PyGEM-OGGM', 'OGGM_v16': 'OGGM'}   # names in the csv file, all others are the same\n",
    "dict_model_col = {'PyGEM-OGGM': '#5fb0e1', 'GloGEMflow': '#d18c14', 'OGGM': '#ca79bc', 'GLIMB': '#fab1e5',\n",
    "                  'GloGEMflow3D': '#0f9373', 'Kraaijenbrink': '#c46011', 'GO': '#e5d73d', 'CISM2': '#562e10'}"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "84d59223",
   "metadata": {},
   "source": [
    "## 1. Prepare the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "b4eb6474",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "a (96, 3) | range of the likely ranges: -33.6 to 60.7\n",
      "b (96, 3) | range of the likely ranges: -60.5 to 47.0\n"
     ]
    }
   ],
   "source": [
    "pd_ss = pd.read_csv(fn_steady_state, index_col=0)\n",
    "pd_ss['model'] = pd_ss.model_author.replace(model_names_csv)\n",
    "pd_ss['reg'] = pd_ss.rgi_reg.replace({'Global': 'All'})\n",
    "\n",
    "rc = 'relative volume change (in %)'\n",
    "dc = 'delta relative volume change (in %, to median model)'\n",
    "dc2 = 'delta relative volume change (in %, to median model, relative to ref period )'\n",
    "\n",
    "# difference of every model to the reference (a) and its change relative to the reference climate (b)\n",
    "if reference == 'all_models_stored':\n",
    "    # OLD: stored delta columns of PartB_3 (median of the absolute volumes divided by the median 2020 volume,\n",
    "    # i.e. not the median of the plotted relative values)\n",
    "    pd_ss['a'] = pd_ss[dc]\n",
    "    pd_ss['b'] = pd_ss[dc2]\n",
    "else:\n",
    "    if reference == 'all_models':\n",
    "        # median of the relative values (% of each glacier model's own 2020 mass) of all glacier models in the region\n",
    "        ref = pd_ss.groupby(['reg', 'gcm', 'period_scenario'])[rc].median()\n",
    "    else:   # 'global4'\n",
    "        ref = pd_ss.loc[pd_ss.model.isin(global_models)].groupby(['reg', 'gcm', 'period_scenario'])[rc].median()\n",
    "    pd_ss = pd_ss.join(ref.rename('ref'), on=['reg', 'gcm', 'period_scenario'])\n",
    "    pd_ss['a'] = pd_ss[rc] - pd_ss['ref']\n",
    "    d_ref_period = pd_ss.loc[pd_ss.period_scenario == ref_period].set_index(['reg', 'model', 'gcm'])['a']\n",
    "    pd_ss = pd_ss.join(d_ref_period.rename('a_ref_period'), on=['reg', 'model', 'gcm'])\n",
    "    pd_ss['b'] = pd_ss['a'] - pd_ss['a_ref_period']\n",
    "\n",
    "pd_bin = pd_ss.loc[(pd_ss.temp_ch_ipcc >= temp_bin[0]) & (pd_ss.temp_ch_ipcc <= temp_bin[1])]\n",
    "n_scen = len(pd_bin.groupby(['gcm', 'period_scenario']))\n",
    "if temp_bin[1] == 4.0:\n",
    "    np.testing.assert_allclose(len(pd_bin.groupby(['gcm', 'period_scenario'])), 31)\n",
    "\n",
    "\n",
    "# median and likely range over the climate scenarios\n",
    "pd_sums = {}\n",
    "for p in ['a', 'b']:\n",
    "    g = pd_bin.groupby(['reg', 'model'])[p]\n",
    "    pd_sums[p] = pd.DataFrame({'med': g.median(), 'lo': g.quantile(likely_range[0] / 100),\n",
    "                               'hi': g.quantile(likely_range[1] / 100)})\n",
    "    print(p, pd_sums[p].shape, '| range of the likely ranges:', np.round(pd_sums[p].lo.min(),1),\n",
    "          'to', np.round(pd_sums[p].hi.max(),1))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d2d399bb",
   "metadata": {},
   "source": [
    "**Check:** the order of the regions. The 2020 volume in the csv file (volume divided by relative volume change) gives the same order as GMIP3-comm Fig. 2:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "2d1e978b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "order from the csv file  : ['19', '03', '01', '05', '09', '04', '07', '17', '06', '13', '14', '02', '15', '08', '10', '11', '16', '18', '12']\n",
      "order used in the figure : ['19', '03', '01', '05', '09', '04', '07', '17', '06', '13', '14', '02', '15', '08', '10', '11', '16', '18', '12']\n",
      "share of the global volume (%): {'19': 30.1, '03': 18.1, '01': 11.8, '05': 9.8, '09': 9.4, '04': 5.2, '07': 4.8, '17': 3.2, '06': 2.3, '13': 2.0, '14': 1.8, '02': 0.6, '15': 0.5, '08': 0.2, '10': 0.1, '11': 0.1, '16': 0.1, '18': 0.0, '12': 0.0}\n"
     ]
    }
   ],
   "source": [
    "pd_chk = pd_ss.loc[pd_ss.reg != 'All'].copy()\n",
    "pd_chk['v0'] = pd_chk.volume_m3 / (pd_chk[rc] / 100)\n",
    "v0_reg = pd_chk.groupby('reg').v0.median()\n",
    "print('order from the csv file  :', list(v0_reg.sort_values(ascending=False).index))\n",
    "print('order used in the figure :', reg_order[1:])\n",
    "print('share of the global volume (%):', (100 * v0_reg / v0_reg.sum()).round(1).reindex(reg_order[1:]).to_dict())"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "559c3971",
   "metadata": {},
   "source": [
    "## 2. The figure"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "98443ff6-fee0-4f32-8bf7-b4172485dfe5",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'a': array([-40, -20,   0,  20,  40,  60]),\n",
       " 'b': array([-60, -40, -20,   0,  20,  40])}"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_ticks"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "a14417f7",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "31 climate scenarios between 2.0°C and 4.0°C global warming\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 326.772x669.291 with 3 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.patheffects as pe      # thin dark outline around the likely-range lines\n",
    "\n",
    "n_panels = len(panel_order)\n",
    "if n_panels == 1:\n",
    "    boxes = [(0.285, 0.70)]\n",
    "else:\n",
    "    # CL: in cm: column of the region names, gap between the panels, right margin\n",
    "    lab_w_cm, gap_cm, right_cm = 1.9, 0.35, 0.25\n",
    "    panel_w_cm = (fig_width_cm - lab_w_cm - gap_cm - right_cm) / 2\n",
    "    boxes = [(lab_w_cm / fig_width_cm, panel_w_cm / fig_width_cm),\n",
    "             ((lab_w_cm + panel_w_cm + gap_cm) / fig_width_cm, panel_w_cm / fig_width_cm)]      # (left, width) of the panels\n",
    "ax_bottom, ax_height = 0.225, 0.715              # CL: in figure fractions (room for the legend and the 4-line x labels)\n",
    "lw_rng, lw_out, ms_o, ms_d = 0.8, 1.3, 2.4, 2.1    # CL: printed line widths / marker sizes (rows are ~3 pt apart)\n",
    "fig = plt.figure(figsize=(fig_width_cm / 2.54, fig_height_cm / 2.54))\n",
    "\n",
    "# one row per glacier model, the four global models first and then the regional models, with a gap between the regions\n",
    "gap = 0.5      # CL: 0.5 instead of 0.7 rows between the regions (height)\n",
    "rows = []\n",
    "y = 0\n",
    "reg_centres = []\n",
    "reg_spans = []\n",
    "for i, reg in enumerate(reg_order):\n",
    "    models_reg = [m for m in global_models + regional_models if (reg, m) in pd_sums['a'].index]\n",
    "    reg_spans.append((y - 0.5 - gap / 2, y + len(models_reg) - 0.5 + gap / 2, i))\n",
    "    for k, m in enumerate(models_reg):\n",
    "        rows.append((reg, m, y + k))\n",
    "    reg_centres.append((y + (len(models_reg) - 1) / 2, reg))\n",
    "    y = y + len(models_reg) + gap\n",
    "y_max = y - gap - 0.5\n",
    "\n",
    "titles = {'a': 'Steady-state mass', 'b': 'Climate scenario\\nsensitivity'}     # CL: line break, panel (b) is only 2.9 cm wide\n",
    "if reference == 'all_models':\n",
    "    xlabels = {'a': 'difference to the\\nmulti-model median of\\nall glacier models\\n(% of 2020 glacier mass)',      # CL: wrapped\n",
    "               'b': 'change of that difference\\ncompared with the\\n1995–2014 climate scenario\\n(percentage points)'}\n",
    "else:\n",
    "    xlabels = {'a': 'difference to the\\nmulti-model median\\n(% of 2020 glacier mass)',\n",
    "               'b': 'change of that difference\\ncompared with the\\n1995–2014 climate scenario\\n(percentage points)'}\n",
    "\n",
    "axes = []\n",
    "for i_p, p in enumerate(panel_order):\n",
    "    #ax = fig.add_axes([boxes[i_p][0], 0.185, boxes[i_p][1], 0.745])\n",
    "    ax = fig.add_axes([boxes[i_p][0], ax_bottom, boxes[i_p][1], ax_height])\n",
    "    axes.append(ax)\n",
    "    for y_lo, y_hi, i in reg_spans:    # light background for every second region\n",
    "        if i % 2 == 1:\n",
    "            ax.axhspan(y_lo, y_hi, color='0.955', lw=0, zorder=0)\n",
    "    for reg, m, y_row in rows:\n",
    "        r = pd_sums[p].loc[(reg, m)]\n",
    "        marker = 'o' if m in global_models else 'D'\n",
    "        ax.plot([r.lo, r.hi], [y_row] * 2, color=dict_model_col[m], lw=lw_rng, solid_capstyle='butt', zorder=2,\n",
    "                path_effects=[pe.Stroke(linewidth=lw_out, foreground='0.25'), pe.Normal()])\n",
    "        ax.plot([r.med], [y_row], marker, ms=ms_o if marker == 'o' else ms_d,\n",
    "                mfc=dict_model_col[m], mec='0.15', mew=0.35, zorder=3)\n",
    "    ax.set_ylim(y_max, -0.5)\n",
    "    ax.set_xlim(*x_lims[p])\n",
    "    ax.axvline(0, color='0.15', lw=0.9, zorder=1)\n",
    "    ax.set_yticks([])\n",
    "    ax.set_xticks(x_ticks[p])\n",
    "    ax.tick_params(axis='x', labelsize=fs, length=2, pad=1.5)\n",
    "    ax.grid(axis='x', color='0.85', lw=0.4, zorder=0)\n",
    "    for sp in ['top', 'right', 'left']:\n",
    "        ax.spines[sp].set_visible(False)\n",
    "    ax.set_xlabel(xlabels[p], fontsize=fs, labelpad=2, linespacing=1.0)      # CL: fs instead of fs + 0.5 (space)\n",
    "    if n_panels > 1:\n",
    "        ax.set_title(f'({chr(97 + i_p)}) {titles[p]}', fontsize=fs + 0.5, fontweight='bold', pad=3)\n",
    "\n",
    "# names of the regions (without the RGI region numbers)\n",
    "for y_c, reg in reg_centres:\n",
    "    name = reg_names_lines[reg]\n",
    "    axes[0].text(-0.02, y_c, name, transform=axes[0].get_yaxis_transform(), ha='right', va='center',\n",
    "                 fontsize=fs, fontweight='bold' if reg == 'All' else 'normal', linespacing=0.95)\n",
    "\n",
    "# which climate scenarios?\n",
    "#fig.text(0.5, 0.993, f'{n_scen} climate scenarios between {temp_bin[0]:.1f}\\u00b0C and {temp_bin[1]:.1f}\\u00b0C global warming',\n",
    "#         fontsize=fs + 0.5, ha='center', va='top')\n",
    "print( f'{n_scen} climate scenarios between {temp_bin[0]:.1f}\\u00b0C and {temp_bin[1]:.1f}\\u00b0C global warming')\n",
    "\n",
    "# legend: dummy row with arrows, and the model names with their group as subtitle\n",
    "# CL: rearranged for the 8.3 cm width, x in cm, all text at fs\n",
    "ax_leg = fig.add_axes([0.0, 0.0, 1.0, 0.12])\n",
    "ax_leg.set_xlim(0, fig_width_cm)\n",
    "ax_leg.set_ylim(0, 1)\n",
    "ax_leg.axis('off')\n",
    "arrow = dict(arrowstyle='-|>', color='k', lw=0.6, shrinkA=0, shrinkB=0, mutation_scale=5)\n",
    "outline = [pe.Stroke(linewidth=lw_out, foreground='0.25'), pe.Normal()]\n",
    "\n",
    "x0_l, x1_l, y_l, x_med = 0.45, 2.45, 0.64, 1.45     # dummy likely range and median (cm)\n",
    "ax_leg.plot([x0_l, x1_l], [y_l, y_l], color='0.75', lw=lw_rng, solid_capstyle='butt', path_effects=outline)\n",
    "ax_leg.plot([x_med], [y_l], 'o', ms=3.0, mfc='0.75', mec='0.15', mew=0.35)\n",
    "ax_leg.text(0.15, 0.99, f'likely range ({likely_range[0]}th to\\n{likely_range[1]}rd percentile)', fontsize=fs, ha='left',\n",
    "            va='top', linespacing=1.0)\n",
    "ax_leg.annotate('', xy=(2.1, y_l + 0.05), xytext=(2.1, 0.78), arrowprops=arrow)\n",
    "ax_leg.text(x_med + 0.25, 0.48, 'median', fontsize=fs, ha='left', va='center')\n",
    "ax_leg.annotate('', xy=(x_med, y_l - 0.05), xytext=(x_med + 0.22, 0.48), arrowprops=arrow)\n",
    "# CL: space between number and unit (Copernicus figure guidelines)\n",
    "t0_str, t1_str = (f'{t:.1f}'.replace('-', '\\u2212') for t in temp_bin)     # real minus sign\n",
    "ax_leg.text(0.15, 0.01, f'(distribution over the {n_scen} climate\\nscenarios within {t0_str} to {t1_str} \\u00b0C\\nglobal warming)',\n",
    "            fontsize=fs, ha='left', va='bottom', linespacing=1.0)\n",
    "\n",
    "x_g, x_r, y_0, dy = 3.85, 5.95, 0.72, 0.19         # CL: start of the two model columns (cm)\n",
    "ax_leg.text(x_g, 0.93, 'Global models', fontsize=fs, fontweight='bold', va='center')\n",
    "ax_leg.text(x_r, 0.93, 'Regional models', fontsize=fs, fontweight='bold', va='center')\n",
    "for k, m in enumerate(global_models):\n",
    "    ax_leg.plot([x_g + 0.1], [y_0 - dy * k], 'o', ms=3.0, mfc=dict_model_col[m], mec='0.15', mew=0.35)\n",
    "    ax_leg.text(x_g + 0.3, y_0 - dy * k, m, fontsize=fs, va='center')\n",
    "for k, m in enumerate(regional_models):\n",
    "    ax_leg.plot([x_r + 0.1], [y_0 - dy * k], 'D', ms=2.7, mfc=dict_model_col[m], mec='0.15', mew=0.35)\n",
    "    ax_leg.text(x_r + 0.3, y_0 - dy * k, m, fontsize=fs, va='center')\n",
    "\n",
    "fn_out = f'{fig_dir}/manuscript_v0_steady_state_glacier_model_differences_portrait_{\"_\".join(panel_order)}_{reference}_temp_bin_{temp_bin[0]}_{temp_bin[1]}'\n",
    "plt.savefig(fn_out + '.png', dpi=300)\n",
    "plt.savefig(fn_out + '.pdf')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "30de58fd",
   "metadata": {},
   "source": [
    "## Caption in LaTeX)\n",
    "\n",
    "```latex\n",
    "\\begin{figure}[ht]\n",
    "    \\centering\n",
    "    \\includegraphics[width=8.3cm]{fig/manuscript_v0_steady_state_glacier_model_differences_portrait_a_b_all_models_temp_bin_2.0_4.0.png}\n",
    "    \\caption{Glacier model differences in steady-state glacier mass for climate scenarios with a global warming of 2.0 to 4.0\\,$^{\\circ}$C (31 climate scenarios), globally and for the 19 RGI regions. \\textbf{(a)} Difference of each glacier model to the multi-model median of all glacier models available in a region (in \\% of the 2020 glacier mass). \\textbf{(b)} Climate scenario sensitivity, i.e.\\ the difference in (a) minus the difference of the same glacier model under the 1995--2014 climate scenario of the same climate model (in percentage points; see Sect.~\\ref{sec:differences}). Negative values in (b) indicate that a glacier model loses more mass than the multi-model median between the 1995--2014 climate scenario and a given climate scenario; values close to zero indicate that its difference to the multi-model median persists across climate scenarios. Dots (global glacier models) and diamonds (regional glacier models) show the median, and lines the likely range (17th to 83rd percentile), over the 31 climate scenarios. Global (top) is the sum over all regions of the four global glacier models; the regions are sorted by their 2020 glacier mass.}\n",
    "\\label{fig:app_gmip3_steady_state_model_differences}\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "adbf9255-aede-4807-910b-d0c8325e737b",
   "metadata": {},
   "source": [
    "## 3. Numbers in the manuscript text, Sect. 3.1.2 (model-specific differences in steady-state mass)\n",
    "\n",
    "Recomputes every number of the text from `pd_ss` (same differences as in the figure). Each line prints `ok` if the value, rounded as in the text, equals the text, otherwise `!!`. Independent of `temp_bin` in the settings (the three warming bins are defined here), but needs `reference = 'all_models'`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "ddd3b0d6-50f1-4879-9526-a680e51d02e2",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "coldest bin: 40 climate scenarios\n",
      "intermediate bin: 31 climate scenarios\n",
      "warmest bin: 9 climate scenarios\n",
      "\n",
      "--- paragraph 1: global glacier models with a consistent direction\n",
      "   PyGEM-OGGM: less in 15, more in 2, itself the multi-model median in 2 regions\n",
      "ok | PyGEM-OGGM less / more / is the median (regions): computed [15, 2, 2] | text [15, 2, 2]\n",
      "ok | PyGEM-OGGM: less by more than 5 % (regions): computed 9 | text 9\n",
      "ok | sign test PyGEM-OGGM, p (3 decimals): computed 0.002 | text 0.002\n",
      "   GLIMB: less in 5, more in 14, itself the multi-model median in 0 regions\n",
      "ok | GLIMB more (regions): computed 14 | text 14\n",
      "ok | GLIMB: more by more than 5 % (regions): computed 10 | text 10\n",
      "ok | sign test GLIMB, p (2 decimals): computed 0.06 | text 0.06\n",
      "ok | Global, 2.0-4.0 degC: GloGEMflow, PyGEM-OGGM, GLIMB, OGGM (%): computed [18, -12, 4, -4] | text [18, -12, 4, -4]\n",
      "ok | all climate scenarios: number / GloGEMflow most global mass / PyGEM-OGGM least: computed [80, 72, 76] | text [80, 72, 76]\n",
      "\n",
      "--- paragraph 2: region-specific deviations of GloGEMflow and OGGM\n",
      "ok | GloGEMflow: at most 5 % from the multi-model median (regions): computed 13 | text 13\n",
      "ok | OGGM: at most 5 % from the multi-model median (regions): computed 13 | text 13\n",
      "ok | GloGEMflow: Svalbard & Jan Mayen, Arctic Canada N, Subantarctic & Antarctic Islands (%): computed [32, 30, 45] | text [32, 30, 45]\n",
      "ok | the two regions with the largest 2020 glacier mass: computed ['19', '03'] | text ['19', '03']\n",
      "ok | OGGM: South Asia W, Subantarctic & Antarctic Islands (%): computed [-16, -10] | text [-16, -10]\n",
      "   (OGGM South Asia W unrounded: -16.497)\n",
      "ok | PyGEM-OGGM: Subantarctic & Antarctic Islands also less (%): computed -9 | text -9\n",
      "ok | Southern Andes: model with the least mass: computed GloGEMflow | text GloGEMflow\n",
      "ok | Southern Andes: GloGEMflow, PyGEM-OGGM (%): computed [-17, -4] | text [-17, -4]\n",
      "\n",
      "--- paragraph 3: regional glacier models\n",
      "ok | Kraaijenbrink: available in / more mass in (regions): computed [3, 3] | text [3, 3]\n",
      "ok | Kraaijenbrink: Central Asia, South Asia W (%): computed [30, 27] | text [30, 27]\n",
      "ok | Kraaijenbrink more than GLIMB in Central Asia and South Asia W: computed True | text True\n",
      "ok | GO, Iceland (%): computed 21 | text 21\n",
      "ok | Central Europe: CISM2, PyGEM-OGGM (%): computed [-2, -2] | text [-2, -2]\n",
      "ok | GloGEMflow3D minus GloGEMflow: regions with both / more / less: computed [11, 4, 7] | text [11, 4, 7]\n",
      "ok | GloGEMflow3D minus GloGEMflow: largest |difference| without Iceland, Iceland (%): computed [2, 8] | text [2, 8]\n",
      "\n",
      "--- paragraph 4: climate scenario sensitivity\n",
      "ok | Global: largest |sensitivity| of the four global models (percentage points): computed 4 | text 4\n",
      "ok | GO, Iceland: sensitivity (percentage points): computed 0 | text 0\n",
      "ok | coldest bin: PyGEM-OGGM less / GLIMB more (regions): computed [16, 11] | text [16, 11]\n",
      "ok | warmest bin: PyGEM-OGGM less / GLIMB more (regions): computed [13, 13] | text [13, 13]\n",
      "ok | W Canada & US, 1995-2014: GloGEMflow, GloGEMflow3D retain (%): computed [10, 10] | text [10, 10]\n",
      "ok | W Canada & US, 1995-2014: other glacier models retain, min / max (%): computed [49, 80] | text [49, 80]\n",
      "   North Asia, 1995-2014 (% retained): GloGEMflow 37, GloGEMflow3D 38, others 57 to 73 ('much less')\n",
      "ok | North Asia: GloGEMflow and GloGEMflow3D retain less than all other glacier models (1995-2014): computed True | text True\n",
      "   sensitivity (percentage points), W Canada & US and North Asia:\n",
      "model  GloGEMflow  GloGEMflow3D\n",
      "reg                            \n",
      "02           33.7          33.2\n",
      "10           21.1          19.1\n",
      "ok | W Canada & US and North Asia: GloGEMflow(3D) lose less mass (all > 0), largest value (\"up to\"): computed [True, 34] | text [True, 34]\n",
      "ok | GloGEMflow3D: Iceland, Central Asia, South Asia W, Low Latitudes lose more mass (all < 0), largest |value|: computed [True, 20] | text [True, 20]\n",
      "   1995-2014, GloGEMflow3D minus GloGEMflow (% of the 2020 glacier mass): {'06': 25.2, '13': 14.7, '14': 1.4, '16': 29.2}\n",
      "ok | \"mostly where it retains more mass than GloGEMflow\": regions with at least 15 % more (rounded): computed 3 | text 3\n",
      "ok | Kraaijenbrink, Central Asia and South Asia W: difference under 1995-2014 (%), i.e. sign change: computed [-8, -2] | text [-8, -2]\n",
      "ok | Kraaijenbrink: sensitivity (percentage points), \"more than 30\": computed [37, 31] | text [37, 31]\n",
      "\n",
      "--- paragraph 5: interpretation\n",
      "ok | GloGEMflow3D loses more mass than GloGEMflow in Iceland, Central Asia and Low Latitudes: computed True | text True\n",
      "ok | ... where it retains more mass under 1995-2014, min / max (%): computed [15, 29] | text [15, 29]\n"
     ]
    }
   ],
   "source": [
    "# Numbers used in the manuscript text, Sect. 3.1.2 (model-specific differences in steady-state mass)\n",
    "# Everything is recomputed from pd_ss (section \"1. Prepare the data\"), i.e. from the same differences as in the figure:\n",
    "#   'a': difference to the multi-model median of all glacier models in the region (% of the 2020 glacier mass)\n",
    "#   'b': climate scenario sensitivity = 'a' minus 'a' under the 1995-2014 climate scenario of the same climate model\n",
    "#        (percentage points)\n",
    "# Medians are taken over the climate scenarios of a warming bin (as the dots in the figure).\n",
    "# Every line prints 'ok' if the value(s), rounded as in the text, equal the number(s) given in the text, otherwise '!!'.\n",
    "from math import comb\n",
    "\n",
    "assert reference == 'all_models', 'the text uses the multi-model median of all glacier models in a region'\n",
    "\n",
    "def rnd(x):\n",
    "    \"\"\"round half away from zero, as in the text (np.round(16.5) would give 16)\"\"\"\n",
    "    x = np.asarray(x, dtype=float)\n",
    "    out = np.sign(x) * np.floor(np.abs(x) + 0.5)\n",
    "    return int(out) if out.ndim == 0 else [int(v) for v in out]\n",
    "\n",
    "def check(statement, computed, text_value):\n",
    "    ok = np.array_equal(np.asarray(computed, dtype=object), np.asarray(text_value, dtype=object))\n",
    "    print(f\"{'ok' if ok else '!!'} | {statement}: computed {computed} | text {text_value}\")\n",
    "\n",
    "def sign_test(n_pos, n_neg):\n",
    "    \"\"\"two-sided sign test (exact binomial test with p = 0.5); regions with a difference of exactly 0 are excluded\"\"\"\n",
    "    n, k = n_pos + n_neg, max(n_pos, n_neg)\n",
    "    return min(1.0, 2 * sum(comb(n, i) for i in range(k, n + 1)) / 2 ** n)\n",
    "\n",
    "# warming bins (coldest bin: all climate scenarios below 2.0 degC, i.e. including the two scenarios slightly below -0.1 degC)\n",
    "scen_t = pd_ss.groupby(['gcm', 'period_scenario']).temp_ch_ipcc.first()\n",
    "bins = {'coldest': (scen_t.min(), 2.0), 'intermediate': (2.0, 4.0), 'warmest': (4.0, scen_t.max())}\n",
    "med = {}\n",
    "for name, (t0, t1) in bins.items():\n",
    "    _sel = pd_ss.loc[(pd_ss.temp_ch_ipcc >= t0) & (pd_ss.temp_ch_ipcc <= t1)]\n",
    "    print(f\"{name} bin: {len(_sel.groupby(['gcm', 'period_scenario']))} climate scenarios\")\n",
    "    med[name] = _sel.groupby(['reg', 'model'])[['a', 'b']].median()\n",
    "A = med['intermediate']['a'].unstack('model')      # rows: regions ('All' = Global), columns: glacier models\n",
    "B = med['intermediate']['b'].unstack('model')\n",
    "regs = [r for r in reg_order if r != 'All']          # the 19 RGI regions\n",
    "Ar = A.loc[regs]\n",
    "is_median = np.isclose(Ar, 0, atol=1e-9)             # regions in which a glacier model is itself the multi-model median\n",
    "# 'a' under the 1995-2014 climate scenarios and the remaining mass itself (median over the five climate models)\n",
    "_ref = pd_ss.loc[pd_ss.period_scenario == ref_period].groupby(['reg', 'model'])\n",
    "A_ref = _ref['a'].median().unstack('model')\n",
    "M_ref = _ref[rc].median().unstack('model')           # % of the 2020 glacier mass\n",
    "print()\n",
    "\n",
    "print('--- paragraph 1: global glacier models with a consistent direction')\n",
    "for m, direction in [('PyGEM-OGGM', 'less'), ('GLIMB', 'more')]:\n",
    "    x = Ar[m]\n",
    "    n_less, n_more, n_med = int((x < 0).sum()), int((x > 0).sum()), int(np.isclose(x, 0, atol=1e-9).sum())\n",
    "    print(f'   {m}: less in {n_less}, more in {n_more}, itself the multi-model median in {n_med} regions')\n",
    "    if m == 'PyGEM-OGGM':\n",
    "        check('PyGEM-OGGM less / more / is the median (regions)', [n_less, n_more, n_med], [15, 2, 2])\n",
    "        check('PyGEM-OGGM: less by more than 5 % (regions)', int((x < -5).sum()), 9)\n",
    "        check('sign test PyGEM-OGGM, p (3 decimals)', round(sign_test(n_more, n_less), 3), 0.002)\n",
    "    else:\n",
    "        check('GLIMB more (regions)', n_more, 14)\n",
    "        check('GLIMB: more by more than 5 % (regions)', int((x > 5).sum()), 10)\n",
    "        check('sign test GLIMB, p (2 decimals)', round(sign_test(n_more, n_less), 2), 0.06)\n",
    "check('Global, 2.0-4.0 degC: GloGEMflow, PyGEM-OGGM, GLIMB, OGGM (%)',\n",
    "      rnd(A.loc['All', ['GloGEMflow', 'PyGEM-OGGM', 'GLIMB', 'OGGM']]), [18, -12, 4, -4])\n",
    "_glob = pd_ss.loc[pd_ss.reg == 'All'].pivot_table(index=['gcm', 'period_scenario'], columns='model', values=rc)\n",
    "check('all climate scenarios: number / GloGEMflow most global mass / PyGEM-OGGM least',\n",
    "      [len(_glob), int((_glob.idxmax(axis=1) == 'GloGEMflow').sum()), int((_glob.idxmin(axis=1) == 'PyGEM-OGGM').sum())],\n",
    "      [80, 72, 76])\n",
    "print()\n",
    "\n",
    "print('--- paragraph 2: region-specific deviations of GloGEMflow and OGGM')\n",
    "for m in ['GloGEMflow', 'OGGM']:\n",
    "    check(f'{m}: at most 5 % from the multi-model median (regions)', int((Ar[m].abs() <= 5).sum()), 13)\n",
    "check('GloGEMflow: Svalbard & Jan Mayen, Arctic Canada N, Subantarctic & Antarctic Islands (%)',\n",
    "      rnd(Ar.loc[['07', '03', '19'], 'GloGEMflow']), [32, 30, 45])\n",
    "_chk = pd_ss.loc[pd_ss.reg != 'All'].copy()\n",
    "_chk['v0'] = _chk.volume_m3 / (_chk[rc] / 100)       # 2020 glacier mass (as in the check of the region order above)\n",
    "check('the two regions with the largest 2020 glacier mass', list(_chk.groupby('reg').v0.median().nlargest(2).index), ['19', '03'])\n",
    "# CL: the text said 17 % for South Asia W; the figure data give -16.497 %, i.e. 16 % (17 % came from the old stored delta column)\n",
    "check('OGGM: South Asia W, Subantarctic & Antarctic Islands (%)', rnd(Ar.loc[['14', '19'], 'OGGM']), [-16, -10])\n",
    "print(f\"   (OGGM South Asia W unrounded: {Ar.loc['14', 'OGGM']:.3f})\")\n",
    "check('PyGEM-OGGM: Subantarctic & Antarctic Islands also less (%)', rnd(Ar.loc['19', 'PyGEM-OGGM']), -9)\n",
    "check('Southern Andes: model with the least mass', Ar.loc['17'].idxmin(), 'GloGEMflow')\n",
    "check('Southern Andes: GloGEMflow, PyGEM-OGGM (%)', rnd(Ar.loc['17', ['GloGEMflow', 'PyGEM-OGGM']]), [-17, -4])\n",
    "print()\n",
    "\n",
    "print('--- paragraph 3: regional glacier models')\n",
    "_k = Ar['Kraaijenbrink'].dropna()\n",
    "check('Kraaijenbrink: available in / more mass in (regions)', [len(_k), int((_k > 0).sum())], [3, 3])\n",
    "check('Kraaijenbrink: Central Asia, South Asia W (%)', rnd(Ar.loc[['13', '14'], 'Kraaijenbrink']), [30, 27])\n",
    "check('Kraaijenbrink more than GLIMB in Central Asia and South Asia W',\n",
    "      bool((Ar.loc[['13', '14'], 'Kraaijenbrink'] > Ar.loc[['13', '14'], 'GLIMB']).all()), True)\n",
    "check('GO, Iceland (%)', rnd(Ar.loc['06', 'GO']), 21)\n",
    "check('Central Europe: CISM2, PyGEM-OGGM (%)', rnd(Ar.loc['11', ['CISM2', 'PyGEM-OGGM']]), [-2, -2])\n",
    "_d3 = (Ar['GloGEMflow3D'] - Ar['GloGEMflow']).dropna()\n",
    "check('GloGEMflow3D minus GloGEMflow: regions with both / more / less',\n",
    "      [len(_d3), int((_d3 > 0).sum()), int((_d3 < 0).sum())], [11, 4, 7])\n",
    "check('GloGEMflow3D minus GloGEMflow: largest |difference| without Iceland, Iceland (%)',\n",
    "      [rnd(_d3.drop('06').abs().max()), rnd(_d3['06'])], [2, 8])\n",
    "print()\n",
    "\n",
    "print('--- paragraph 4: climate scenario sensitivity')\n",
    "check('Global: largest |sensitivity| of the four global models (percentage points)',\n",
    "      rnd(B.loc['All', global_models].abs().max()), 4)\n",
    "check('GO, Iceland: sensitivity (percentage points)', rnd(B.loc['06', 'GO']), 0)\n",
    "for name in ['coldest', 'warmest']:\n",
    "    _a = med[name]['a'].unstack('model').loc[regs]\n",
    "    check(f'{name} bin: PyGEM-OGGM less / GLIMB more (regions)',\n",
    "          [int((_a['PyGEM-OGGM'] < 0).sum()), int((_a['GLIMB'] > 0).sum())], {'coldest': [16, 11], 'warmest': [13, 13]}[name])\n",
    "_others = ['PyGEM-OGGM', 'OGGM', 'GLIMB']\n",
    "check('W Canada & US, 1995-2014: GloGEMflow, GloGEMflow3D retain (%)', rnd(M_ref.loc['02', ['GloGEMflow', 'GloGEMflow3D']]), [10, 10])\n",
    "check('W Canada & US, 1995-2014: other glacier models retain, min / max (%)',\n",
    "      [rnd(M_ref.loc['02', _others].min()), rnd(M_ref.loc['02', _others].max())], [49, 80])\n",
    "print(f\"   North Asia, 1995-2014 (% retained): GloGEMflow {M_ref.loc['10', 'GloGEMflow']:.0f}, \"\n",
    "      f\"GloGEMflow3D {M_ref.loc['10', 'GloGEMflow3D']:.0f}, others {M_ref.loc['10', _others].min():.0f} to \"\n",
    "      f\"{M_ref.loc['10', _others].max():.0f} ('much less')\")\n",
    "check('North Asia: GloGEMflow and GloGEMflow3D retain less than all other glacier models (1995-2014)',\n",
    "      bool(M_ref.loc['10', ['GloGEMflow', 'GloGEMflow3D']].max() < M_ref.loc['10', _others].min()), True)\n",
    "_bw = B.loc[['02', '10'], ['GloGEMflow', 'GloGEMflow3D']]\n",
    "print(f'   sensitivity (percentage points), W Canada & US and North Asia:\\n{_bw.round(1)}')\n",
    "check('W Canada & US and North Asia: GloGEMflow(3D) lose less mass (all > 0), largest value (\"up to\")',\n",
    "      [bool((_bw > 0).all().all()), rnd(_bw.max().max())], [True, 34])\n",
    "_r3 = ['06', '13', '14', '16']                       # Iceland, Central Asia, South Asia W, Low Latitudes\n",
    "check('GloGEMflow3D: Iceland, Central Asia, South Asia W, Low Latitudes lose more mass (all < 0), largest |value|',\n",
    "      [bool((B.loc[_r3, 'GloGEMflow3D'] < 0).all()), rnd(B.loc[_r3, 'GloGEMflow3D'].abs().max())], [True, 20])\n",
    "_dm = M_ref.loc[_r3, 'GloGEMflow3D'] - M_ref.loc[_r3, 'GloGEMflow']\n",
    "print(f'   1995-2014, GloGEMflow3D minus GloGEMflow (% of the 2020 glacier mass): {_dm.round(1).to_dict()}')\n",
    "check('\"mostly where it retains more mass than GloGEMflow\": regions with at least 15 % more (rounded)', sum(v >= 15 for v in rnd(_dm)), 3)\n",
    "check('Kraaijenbrink, Central Asia and South Asia W: difference under 1995-2014 (%), i.e. sign change',\n",
    "      rnd(A_ref.loc[['13', '14'], 'Kraaijenbrink']), [-8, -2])\n",
    "check('Kraaijenbrink: sensitivity (percentage points), \"more than 30\"', [rnd(v) for v in B.loc[['13', '14'], 'Kraaijenbrink']], [37, 31])\n",
    "print()\n",
    "\n",
    "print('--- paragraph 5: interpretation')\n",
    "check('GloGEMflow3D loses more mass than GloGEMflow in Iceland, Central Asia and Low Latitudes',\n",
    "      bool((B.loc[['06', '13', '16'], 'GloGEMflow3D'] < B.loc[['06', '13', '16'], 'GloGEMflow']).all()), True)\n",
    "check('... where it retains more mass under 1995-2014, min / max (%)', [rnd(_dm[['06', '13', '16']].min()), rnd(_dm[['06', '13', '16']].max())], [15, 29])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5a52a4ea-c410-460a-98fe-46c367dfd3d5",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:oggm_gmip3_working]",
   "language": "python",
   "name": "conda-env-oggm_gmip3_working-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": "3.10.14"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}