Some guideance on regrid testing and regrid UI

regrid_help.ipynb

{
 "metadata": {
  "name": "",
  "signature": "sha256:44ca4f80992b328dbe607f0caf37756bc6f0eea7df01508482f2eacac574e785"
 },
 "nbformat": 3,
 "nbformat_minor": 0,
 "worksheets": [
  {
   "cells": [
    {
     "cell_type": "heading",
     "level": 3,
     "metadata": {},
     "source": [
      "Regridding interface"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "class ESMF(object):\n",
      "    def regridder(self, src_grid, tgt_grid):\n",
      "        _ESMFRegridder(src_grid, tgt_grid, method=self._method, extrapolation='self._extrapolation')\n",
      "\n",
      "\n",
      "class LinearESMF(ESMF):\n",
      "    def __init__(self, extrapolation='linear'):\n",
      "        self._extrapolation = extrapolation\n",
      "        self._method = 'linear'\n",
      "\n",
      "\n",
      "class ConservativeESMF(ESMF):\n",
      "    def __init__(self, extrapolation='linear'):\n",
      "        self._extrapolation = extrapolation\n",
      "        self._method = 'weightedArea'"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "heading",
     "level": 3,
     "metadata": {},
     "source": [
      "Calling a regridder directly (TwoStage, ConservativeESMF, Linear, AreaWeighted, ... etc.)"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "result = source.regrid(target, TwoStage())"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "heading",
     "level": 3,
     "metadata": {},
     "source": [
      "Calling via the decomposition framework"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "operation = ConservativeESMF().regridder(source, target)\n",
      "result = ants.decomposition.decompose(operation, source, target)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "heading",
     "level": 3,
     "metadata": {},
     "source": [
      "Calling the ants general regridder commandline application (choosing TwoStage, ConservativeESMF, Linear, AreaWeighted, ... etc.)"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "$ ancil_generalised_regrid.py source_file --target-grid target_grid_file -o output_file --ants-config /path/to/config.cfg"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "$ cat /path/to/config.cfg:\n",
      "\n",
      "    [decomposition]\n",
      "    target_split = 1, 1, 10, 10\n",
      "    processes = 10\n",
      "    framework = MultiprocessingDomainDecompose\n",
      "\n",
      "    [regridding_horizontal]\n",
      "    scheme = TwoStage"
     ]
    },
    {
     "cell_type": "heading",
     "level": 3,
     "metadata": {},
     "source": [
      "How about a unittest for the regridder"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Example unittest for from the TwoStage regridder:\n",
      "\n",
      "`ants.tests.regrid.test_integration.TestTwoStage.test_api`  \n",
      "https://code.metoffice.gov.uk/trac/ancil/browser/ants/trunk/lib/ants/tests/regrid/test_integration.py?rev=3543#L73\n",
      "\n",
      "This test takes an OSGB source and regrids this to a lat-lon target.  \n",
      "https://code.metoffice.gov.uk/trac/ancil/browser/ants/trunk/lib/ants/tests/results/visual_tests/test_integration.TestTwoStage.test_api.0.png?rev=3543\n",
      "\n",
      "A graphical image is generated upon which future changes will be compared against."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Note: There are numerous useful target grids available, of which many of interest can be found here:\n",
      "        \n",
      "    https://code.metoffice.gov.uk/trac/ancil/browser/data/trunk/namelist/grids\n",
      "        \n",
      "Variable resolution grids:\n",
      "    \n",
      "    https://code.metoffice.gov.uk/trac/ancil/browser/data/trunk/namelist/variable_grids\n",
      "        \n",
      "These files can be loaded easily by `ants.load_grid(filenames)`"
     ]
    },
    {
     "cell_type": "heading",
     "level": 3,
     "metadata": {},
     "source": [
      "How about an integration test which demonstrates end-to-end usage"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "The UI is explained most clearly in the docs on trunk here: https://code.metoffice.gov.uk/trac/ancil/browser/ants/trunk/doc/ants/source/running_applications.rst"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Create an 'application test'\n",
      "\n",
      "$ cat ants/bin/tests/test_generalised_regrid_rotated.py"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "# (C) British Crown Copyright 2017, Met Office\n",
      "# Refer to README.md of this project for license details.\n",
      "#\n",
      "# This file is part of ANTS.\n",
      "import common\n",
      "from ants.config import CONFIG\n",
      "\n",
      "\n",
      "@common.skip_data\n",
      "class TestAll(common.StaticTestCase):\n",
      "    def test_all(self):\n",
      "        CONFIG['regridding']['scheme'] = 'ConservativeESMF'\n",
      "\n",
      "        source_filepath = common.get_data_path('river_routing_preprocessed.nc')\n",
      "        target_grid_path = common.get_data_path(\n",
      "            'ukv_coarse_land_cover_fraction.nc')\n",
      "\n",
      "        self.application_test('ancil_generalised_regrid',\n",
      "                              source_filepath,\n",
      "                              target_grid_path,\n",
      "                              self.temporary_output(identifier='rotated'))\n",
      "\n",
      "\n",
      "if __name__ == '__main__':\n",
      "    common.main()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "If you run this new application, it generates the following new reference files:  \n",
      "\n",
      "   bin/tests/results/ancil_generalised_regrid_rotated.ancil  \n",
      "   bin/tests/results/ancil_generalised_regrid_rotated.cdl  \n",
      "   bin/tests/results/visual_tests/test_generalised_regrid_rotated.TestAll.test_all.0.png  \n",
      "   bin/tests/results/visual_tests/test_generalised_regrid_rotated.TestAll.test_all.1.png  \n",
      "    \n",
      "That is, an xml of the cubes, a dump of the ancillary metadata and lastly the graphical plots.  On future running of these tests,\n",
      "a change to the test results will result in a difference in one or more of these files and cause a failure indicating what is different."
     ]
    }
   ],
   "metadata": {}
  }
 ]
}

Hi @cpelley,
Thanks, good start.

• As mentioned previously, there is a single esmf regridding class that takes care of all methods (linear, area weighted).
• Also there is no extrapolation involved. I'm thinking that should be a separate step -- I wouldn't know in any case what kind of extrapolation should be applied.