Tanzanian_Veg_EE.ipynb

{
 "metadata": {
  "name": "",
  "signature": "sha256:53cfe6f6637eed4a1e15d962507f2092343748227579c807e5e169efca476068"
 },
 "nbformat": 3,
 "nbformat_minor": 0,
 "worksheets": [
  {
   "cells": [
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "Obtaining EVI data by land type within a Polygon using Google's Earth Engine"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "The first portions of this notebook are entirely from Tyler Erikson's Earth Engine Notebooks.\n",
      "Additionally, thanks to Noel Gorelick, Tyler Erikson (Google), and Lukasz Tracewski, for helpful comments.\n",
      "\n",
      "This code is desgined to create masks for a given range of land-types (broadly catagorised as Forest and grassland), then apply these masks to large volumes of Landsat and MODIS Enhanced Vegitation Index (EVI) data, and extract statistic from within a speicifc study area (a region that broadly encomasses coastal Tanzania to approx 200km inland).\n",
      "\n",
      "These time-series data are then converted into a Pandas dataframe, and saved as a CSV file (one value per time-step, of a mean, stdev, and count to indicate the EVI over the grass-type and forest-type EVI pixels. \n",
      "\n",
      "The user can specify which data collection to call from, however it is set up now to use LandSat 5 and 7 30m 8-day, and MODIS Terra 250m 16-day data.\n",
      "\n",
      "Processing each image may take around a second, so beware when requesting large volumes of data. Fortunatley though, there will be no time-out from the Google servers, as requests for each image are sent individually (and consecutivley), and take around a second each.\n",
      "\n",
      "Coded by Dr. Benjamin A. Laken (email: benlaken@gmail.com    site: www.benlaken.com)"
     ]
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "Load modules (requires EE access)"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import ee\n",
      "import os\n",
      "import datetime\n",
      "import matplotlib.pyplot as plt\n",
      "import matplotlib.colors as mcolors\n",
      "\n",
      "from IPython.html import widgets\n",
      "from IPython.display import display\n",
      "from IPython.utils import traitlets\n",
      "from IPython.core.display import Javascript\n",
      "\n",
      "ee.Initialize()  # Initlize Earth Engine, the part that requires credentials"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 3
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "Map Widget and JS (by Tyler Erickson)"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "class GoogleMapsWidget(widgets.DOMWidget):\n",
      "    _view_name = traitlets.Unicode('GoogleMapsView', sync=True)\n",
      "    value = traitlets.Unicode(sync=True)\n",
      "    description = traitlets.Unicode(sync=True)\n",
      "    lat = traitlets.CFloat(0, help=\"Center latitude, -90 to 90\", sync=True)\n",
      "    lng = traitlets.CFloat(0, help=\"Center longitude, -180 to 180\", sync=True)\n",
      "    zoom = traitlets.CInt(0, help=\"Zoom level, 0 to ~25\", sync=True)\n",
      "    bounds = traitlets.List([], help=\"Visible bounds, [W, S, E, N]\", sync=True)\n",
      "    \n",
      "    def __init__(self, lng=0.0, lat=0.0, zoom=2):\n",
      "        self.lng = lng\n",
      "        self.lat = lat\n",
      "        self.zoom = zoom\n",
      "        \n",
      "    def addLayer(self, image, vis_params=None, name=None, visible=True):\n",
      "        mapid = image.getMapId(vis_params)\n",
      "        self.send({'command':'addLayer', 'mapid': mapid['mapid'], 'token': mapid['token'], 'name': name, 'visible': visible})\n",
      "        \n",
      "    def center(self, lng, lat, zoom=None):\n",
      "        self.send({'command': 'center', 'lng': lng, 'lat': lat, 'zoom': zoom})"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 4
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "%%javascript\n",
      "\n",
      "require([\"widgets/js/widget\"], function(WidgetManager){\n",
      "    var maps = [];\n",
      "    \n",
      "    // Define the GoogleMapsView\n",
      "    var GoogleMapsView = IPython.DOMWidgetView.extend({\n",
      "        \n",
      "        render: function() {\n",
      "            // Resize widget element to be 100% wide\n",
      "            this.$el.css('width', '100%');\n",
      "\n",
      "            // iframe source;  just enough to load Google Maps and let us poll whether initialization is complete\n",
      "            var src='<html style=\"height:100%\"><head>' +\n",
      "                '<scr'+'ipt src=\"http://maps.googleapis.com/maps/api/js?sensor=false\"></scr'+'ipt>' +\n",
      "                '<scr'+'ipt>google.maps.event.addDomListener(window,\"load\",function(){ready=true});</scr'+'ipt>' +\n",
      "                '</head>' +\n",
      "                '<body style=\"height:100%; margin:0px; padding:0px\"></body></html>';\n",
      "            \n",
      "            // Create the Google Maps container element.\n",
      "            this.$iframe = $('<iframe />')\n",
      "                .css('width', '100%')\n",
      "                .css('height', '500px')\n",
      "                .attr('srcdoc', src)\n",
      "                .appendTo(this.$el);\n",
      "                        \n",
      "            var self = this; // hold onto this for initMapWhenReady\n",
      "\n",
      "            // Wait until maps library has finished loading in iframe, then create map\n",
      "            function initMapWhenReady() {\n",
      "                // Iframe document and window\n",
      "                var doc = self.$iframe[0].contentDocument;\n",
      "                var win = self.$iframe[0].contentWindow;\n",
      "                if (!win || !win.ready) {\n",
      "                    // Maps library not yet loaded;  try again soon\n",
      "                    setTimeout(initMapWhenReady, 20);\n",
      "                    return;\n",
      "                }\n",
      "\n",
      "                // Maps library finished loading.  Build map now.\n",
      "                var mapOptions = {\n",
      "                    center: new win.google.maps.LatLng(self.model.get('lat'), self.model.get('lng')),\n",
      "                    zoom: self.model.get('zoom')\n",
      "                };\n",
      "                var mapDiv = $('<div />')\n",
      "                    .css('width', '100%')\n",
      "                    .css('height', '100%')\n",
      "                    .appendTo($(doc.body));\n",
      "                self.map = new win.google.maps.Map(mapDiv[0], mapOptions);\n",
      "                \n",
      "                \n",
      "                // Add an event listeners for user panning, zooming, and resizing map\n",
      "                // TODO(rsargent): Bind self across all methods, and save some plumbing here\n",
      "                win.google.maps.event.addListener(self.map, 'bounds_changed', function () {\n",
      "                    self.handleBoundsChanged();\n",
      "                });\n",
      "                \n",
      "                self.initializeLayersControl();\n",
      "            }\n",
      "            initMapWhenReady();\n",
      "        },\n",
      "        \n",
      "        LayersControl: function(widget, controlDiv, map) {\n",
      "            var win = widget.$iframe[0].contentWindow;\n",
      "            var chicago = new win.google.maps.LatLng(41.850033, -87.6500523);\n",
      "\n",
      "            // Set CSS styles for the DIV containing the control\n",
      "            // Setting padding to 5 px will offset the control\n",
      "            // from the edge of the map.\n",
      "            controlDiv.style.padding = '5px';\n",
      "\n",
      "            // Set CSS for the control border.\n",
      "            var $controlUI = $('<div />')\n",
      "                .css('backgroundColor', 'white')\n",
      "                .css('borderStyle', 'solid')\n",
      "                .css('borderWidth', '1px')\n",
      "                .css('cursor', 'pointer')\n",
      "                .css('textAlign', 'center')\n",
      "                .appendTo($(controlDiv));\n",
      "            \n",
      "            // Set CSS for the control interior.\n",
      "            var $controlContents = $('<div />')\n",
      "                .css('fontFamily', 'Arial,sans-serif')\n",
      "                .css('fontSize', '12px')\n",
      "                .css('paddingLeft', '4px')\n",
      "                .css('paddingRight', '4px')\n",
      "                .css('paddingTop', '0px')\n",
      "                .css('paddingBottom', '0px')\n",
      "                .appendTo($controlUI);\n",
      "            \n",
      "            this.$controlTable = $('<table />')\n",
      "                .append($('<tr><td colspan=2>Layers</td></tr>'))\n",
      "                .appendTo($controlContents);\n",
      "        },\n",
      "\n",
      "        initializeLayersControl: function() {\n",
      "            var doc = this.$iframe[0].contentDocument;\n",
      "            var win = this.$iframe[0].contentWindow;\n",
      "\n",
      "            // Create the DIV to hold the control and call the LayersControl() constructor\n",
      "            // passing in this DIV.\n",
      "    \n",
      "            var layersControlDiv = document.createElement('div');\n",
      "            this.layersControl = new this.LayersControl(this, layersControlDiv, this.map);\n",
      "\n",
      "            layersControlDiv.index = 1;\n",
      "            this.map.controls[win.google.maps.ControlPosition.TOP_RIGHT].push(layersControlDiv);\n",
      "        },\n",
      "        \n",
      "        // Map geometry changed (pan, zoom, resize)\n",
      "        handleBoundsChanged: function() {\n",
      "            this.model.set('lng', this.map.getCenter().lng());\n",
      "            this.model.set('lat', this.map.getCenter().lat());\n",
      "            this.model.set('zoom', this.map.getZoom());\n",
      "            var bounds = this.map.getBounds();\n",
      "            var playgroundCompatible = [bounds.getSouthWest().lng(), bounds.getSouthWest().lat(),\n",
      "                                        bounds.getNorthEast().lng(), bounds.getNorthEast().lat()];\n",
      "            this.model.set('bounds', playgroundCompatible);\n",
      "            this.touch();\n",
      "        },\n",
      "        \n",
      "        // Receive custom messages from Python backend\n",
      "        on_msg: function(msg) {\n",
      "            var win = this.$iframe[0].contentWindow;\n",
      "            if (msg.command == 'addLayer') {\n",
      "                this.addLayer(msg.mapid, msg.token, msg.name, msg.visible);\n",
      "            } else if (msg.command == 'center') {\n",
      "                this.map.setCenter(new win.google.maps.LatLng(msg.lat, msg.lng));\n",
      "                if (msg.zoom !== null) {\n",
      "                    this.map.setZoom(msg.zoom);\n",
      "                }\n",
      "            }\n",
      "        },\n",
      "        \n",
      "        // Add an Earth Engine layer\n",
      "        addLayer: function(mapid, token, name, visible) {\n",
      "            var win = this.$iframe[0].contentWindow;\n",
      "            var eeMapOptions = {\n",
      "                getTileUrl: function(tile, zoom) {\n",
      "                    var url = ['https://earthengine.googleapis.com/map',\n",
      "                               mapid, zoom, tile.x, tile.y].join(\"/\");\n",
      "                    url += '?token=' + token\n",
      "                    return url;\n",
      "                },\n",
      "                tileSize: new win.google.maps.Size(256, 256),\n",
      "                opacity: visible ? 1.0 : 0.0,\n",
      "            };\n",
      "            \n",
      "            // Create the overlay map type\n",
      "            var mapType = new win.google.maps.ImageMapType(eeMapOptions);\n",
      "                \n",
      "            // Overlay the Earth Engine generated layer\n",
      "            this.map.overlayMapTypes.push(mapType);\n",
      "\n",
      "            // Update layer visibility control\n",
      "            var maxSlider = 100;\n",
      "            \n",
      "            function updateOpacity() {\n",
      "                mapType.setOpacity($checkbox[0].checked ? $slider[0].value / 100.0 : 0);\n",
      "            }\n",
      "            \n",
      "            var $checkbox = $('<input type=\"checkbox\">')\n",
      "                .prop('checked', visible)\n",
      "                .change(updateOpacity);\n",
      "            \n",
      "            var $slider = $('<input type=\"range\" />')\n",
      "                .prop('min', 0)\n",
      "                .prop('max', maxSlider)\n",
      "                .prop('value', maxSlider)\n",
      "                .css('width', '60px')\n",
      "                .on('input', updateOpacity);\n",
      "\n",
      "            // If user doesn't specify a layer name, create a default\n",
      "            if (name === null) {\n",
      "                name = 'Layer ' + this.map.overlayMapTypes.length;\n",
      "            }\n",
      "            \n",
      "            var $row = $('<tr />');\n",
      "            $('<td align=\"left\" />').append($checkbox).append(name).appendTo($row);\n",
      "            $('<td />').append($slider).appendTo($row);\n",
      "\n",
      "            this.layersControl.$controlTable.append($row);\n",
      "        }\n",
      "    });\n",
      "    \n",
      "    // Register the GoogleMapsView with the widget manager.\n",
      "    WidgetManager.register_widget_view('GoogleMapsView', GoogleMapsView);\n",
      "});"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "javascript": [
        "\n",
        "require([\"widgets/js/widget\"], function(WidgetManager){\n",
        "    var maps = [];\n",
        "    \n",
        "    // Define the GoogleMapsView\n",
        "    var GoogleMapsView = IPython.DOMWidgetView.extend({\n",
        "        \n",
        "        render: function() {\n",
        "            // Resize widget element to be 100% wide\n",
        "            this.$el.css('width', '100%');\n",
        "\n",
        "            // iframe source;  just enough to load Google Maps and let us poll whether initialization is complete\n",
        "            var src='<html style=\"height:100%\"><head>' +\n",
        "                '<scr'+'ipt src=\"http://maps.googleapis.com/maps/api/js?sensor=false\"></scr'+'ipt>' +\n",
        "                '<scr'+'ipt>google.maps.event.addDomListener(window,\"load\",function(){ready=true});</scr'+'ipt>' +\n",
        "                '</head>' +\n",
        "                '<body style=\"height:100%; margin:0px; padding:0px\"></body></html>';\n",
        "            \n",
        "            // Create the Google Maps container element.\n",
        "            this.$iframe = $('<iframe />')\n",
        "                .css('width', '100%')\n",
        "                .css('height', '500px')\n",
        "                .attr('srcdoc', src)\n",
        "                .appendTo(this.$el);\n",
        "                        \n",
        "            var self = this; // hold onto this for initMapWhenReady\n",
        "\n",
        "            // Wait until maps library has finished loading in iframe, then create map\n",
        "            function initMapWhenReady() {\n",
        "                // Iframe document and window\n",
        "                var doc = self.$iframe[0].contentDocument;\n",
        "                var win = self.$iframe[0].contentWindow;\n",
        "                if (!win || !win.ready) {\n",
        "                    // Maps library not yet loaded;  try again soon\n",
        "                    setTimeout(initMapWhenReady, 20);\n",
        "                    return;\n",
        "                }\n",
        "\n",
        "                // Maps library finished loading.  Build map now.\n",
        "                var mapOptions = {\n",
        "                    center: new win.google.maps.LatLng(self.model.get('lat'), self.model.get('lng')),\n",
        "                    zoom: self.model.get('zoom')\n",
        "                };\n",
        "                var mapDiv = $('<div />')\n",
        "                    .css('width', '100%')\n",
        "                    .css('height', '100%')\n",
        "                    .appendTo($(doc.body));\n",
        "                self.map = new win.google.maps.Map(mapDiv[0], mapOptions);\n",
        "                \n",
        "                \n",
        "                // Add an event listeners for user panning, zooming, and resizing map\n",
        "                // TODO(rsargent): Bind self across all methods, and save some plumbing here\n",
        "                win.google.maps.event.addListener(self.map, 'bounds_changed', function () {\n",
        "                    self.handleBoundsChanged();\n",
        "                });\n",
        "                \n",
        "                self.initializeLayersControl();\n",
        "            }\n",
        "            initMapWhenReady();\n",
        "        },\n",
        "        \n",
        "        LayersControl: function(widget, controlDiv, map) {\n",
        "            var win = widget.$iframe[0].contentWindow;\n",
        "            var chicago = new win.google.maps.LatLng(41.850033, -87.6500523);\n",
        "\n",
        "            // Set CSS styles for the DIV containing the control\n",
        "            // Setting padding to 5 px will offset the control\n",
        "            // from the edge of the map.\n",
        "            controlDiv.style.padding = '5px';\n",
        "\n",
        "            // Set CSS for the control border.\n",
        "            var $controlUI = $('<div />')\n",
        "                .css('backgroundColor', 'white')\n",
        "                .css('borderStyle', 'solid')\n",
        "                .css('borderWidth', '1px')\n",
        "                .css('cursor', 'pointer')\n",
        "                .css('textAlign', 'center')\n",
        "                .appendTo($(controlDiv));\n",
        "            \n",
        "            // Set CSS for the control interior.\n",
        "            var $controlContents = $('<div />')\n",
        "                .css('fontFamily', 'Arial,sans-serif')\n",
        "                .css('fontSize', '12px')\n",
        "                .css('paddingLeft', '4px')\n",
        "                .css('paddingRight', '4px')\n",
        "                .css('paddingTop', '0px')\n",
        "                .css('paddingBottom', '0px')\n",
        "                .appendTo($controlUI);\n",
        "            \n",
        "            this.$controlTable = $('<table />')\n",
        "                .append($('<tr><td colspan=2>Layers</td></tr>'))\n",
        "                .appendTo($controlContents);\n",
        "        },\n",
        "\n",
        "        initializeLayersControl: function() {\n",
        "            var doc = this.$iframe[0].contentDocument;\n",
        "            var win = this.$iframe[0].contentWindow;\n",
        "\n",
        "            // Create the DIV to hold the control and call the LayersControl() constructor\n",
        "            // passing in this DIV.\n",
        "    \n",
        "            var layersControlDiv = document.createElement('div');\n",
        "            this.layersControl = new this.LayersControl(this, layersControlDiv, this.map);\n",
        "\n",
        "            layersControlDiv.index = 1;\n",
        "            this.map.controls[win.google.maps.ControlPosition.TOP_RIGHT].push(layersControlDiv);\n",
        "        },\n",
        "        \n",
        "        // Map geometry changed (pan, zoom, resize)\n",
        "        handleBoundsChanged: function() {\n",
        "            this.model.set('lng', this.map.getCenter().lng());\n",
        "            this.model.set('lat', this.map.getCenter().lat());\n",
        "            this.model.set('zoom', this.map.getZoom());\n",
        "            var bounds = this.map.getBounds();\n",
        "            var playgroundCompatible = [bounds.getSouthWest().lng(), bounds.getSouthWest().lat(),\n",
        "                                        bounds.getNorthEast().lng(), bounds.getNorthEast().lat()];\n",
        "            this.model.set('bounds', playgroundCompatible);\n",
        "            this.touch();\n",
        "        },\n",
        "        \n",
        "        // Receive custom messages from Python backend\n",
        "        on_msg: function(msg) {\n",
        "            var win = this.$iframe[0].contentWindow;\n",
        "            if (msg.command == 'addLayer') {\n",
        "                this.addLayer(msg.mapid, msg.token, msg.name, msg.visible);\n",
        "            } else if (msg.command == 'center') {\n",
        "                this.map.setCenter(new win.google.maps.LatLng(msg.lat, msg.lng));\n",
        "                if (msg.zoom !== null) {\n",
        "                    this.map.setZoom(msg.zoom);\n",
        "                }\n",
        "            }\n",
        "        },\n",
        "        \n",
        "        // Add an Earth Engine layer\n",
        "        addLayer: function(mapid, token, name, visible) {\n",
        "            var win = this.$iframe[0].contentWindow;\n",
        "            var eeMapOptions = {\n",
        "                getTileUrl: function(tile, zoom) {\n",
        "                    var url = ['https://earthengine.googleapis.com/map',\n",
        "                               mapid, zoom, tile.x, tile.y].join(\"/\");\n",
        "                    url += '?token=' + token\n",
        "                    return url;\n",
        "                },\n",
        "                tileSize: new win.google.maps.Size(256, 256),\n",
        "                opacity: visible ? 1.0 : 0.0,\n",
        "            };\n",
        "            \n",
        "            // Create the overlay map type\n",
        "            var mapType = new win.google.maps.ImageMapType(eeMapOptions);\n",
        "                \n",
        "            // Overlay the Earth Engine generated layer\n",
        "            this.map.overlayMapTypes.push(mapType);\n",
        "\n",
        "            // Update layer visibility control\n",
        "            var maxSlider = 100;\n",
        "            \n",
        "            function updateOpacity() {\n",
        "                mapType.setOpacity($checkbox[0].checked ? $slider[0].value / 100.0 : 0);\n",
        "            }\n",
        "            \n",
        "            var $checkbox = $('<input type=\"checkbox\">')\n",
        "                .prop('checked', visible)\n",
        "                .change(updateOpacity);\n",
        "            \n",
        "            var $slider = $('<input type=\"range\" />')\n",
        "                .prop('min', 0)\n",
        "                .prop('max', maxSlider)\n",
        "                .prop('value', maxSlider)\n",
        "                .css('width', '60px')\n",
        "                .on('input', updateOpacity);\n",
        "\n",
        "            // If user doesn't specify a layer name, create a default\n",
        "            if (name === null) {\n",
        "                name = 'Layer ' + this.map.overlayMapTypes.length;\n",
        "            }\n",
        "            \n",
        "            var $row = $('<tr />');\n",
        "            $('<td align=\"left\" />').append($checkbox).append(name).appendTo($row);\n",
        "            $('<td />').append($slider).appendTo($row);\n",
        "\n",
        "            this.layersControl.$controlTable.append($row);\n",
        "        }\n",
        "    });\n",
        "    \n",
        "    // Register the GoogleMapsView with the widget manager.\n",
        "    WidgetManager.register_widget_view('GoogleMapsView', GoogleMapsView);\n",
        "});"
       ],
       "metadata": {},
       "output_type": "display_data",
       "text": [
        "<IPython.core.display.Javascript at 0x105e5c710>"
       ]
      }
     ],
     "prompt_number": 5
    },
    {
     "cell_type": "heading",
     "level": 3,
     "metadata": {},
     "source": [
      "JS map widget to display images (result of above code)"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "map_item = GoogleMapsWidget(lat=0, lng=0.0, zoom=3)\n",
      "display(map_item)     # add items to this map later..."
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 6
    },
    {
     "cell_type": "heading",
     "level": 3,
     "metadata": {},
     "source": [
      "Function to handel retrying EE code if a warning (memory allocation issue) is recieved"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "#Taken from http://www.saltycrane.com/blog/2009/11/trying-out-retry-decorator-python/\n",
      "\n",
      "import time\n",
      "from functools import wraps\n",
      "\n",
      "def retry(ExceptionToCheck, tries=8, delay=3, backoff=2, default=None, logger=None):\n",
      "    \"\"\"Retry calling the decorated function using an exponential backoff.\n",
      "    http://www.saltycrane.com/blog/2009/11/trying-out-retry-decorator-python/\n",
      "    original from: http://wiki.python.org/moin/PythonDecoratorLibrary#Retry\n",
      "    :param ExceptionToCheck: the exception to check. may be a tuple of\n",
      "        exceptions to check\n",
      "    :type ExceptionToCheck: Exception or tuple\n",
      "    :param tries: number of times to try (not retry) before giving up\n",
      "    :type tries: int\n",
      "    :param delay: initial delay between retries in seconds\n",
      "    :type delay: int\n",
      "    :param backoff: backoff multiplier e.g. value of 2 will double the delay\n",
      "        each retry\n",
      "    :type backoff: int\n",
      "    :param logger: logger to use. If None, print\n",
      "    :type logger: logging.Logger instance\n",
      "    \"\"\"\n",
      "    def deco_retry(f):\n",
      "\n",
      "        @wraps(f)\n",
      "        def f_retry(*args, **kwargs):\n",
      "            mtries, mdelay = tries, delay\n",
      "            while mtries > 1:\n",
      "                try:\n",
      "                    return f(*args, **kwargs)\n",
      "                except ExceptionToCheck, e:\n",
      "                    msg = \"%s, Retrying in %d seconds...\" % (str(e), mdelay)\n",
      "                    if logger:\n",
      "                        logger.warning(msg)\n",
      "                    else:\n",
      "                        print msg\n",
      "                    time.sleep(mdelay)\n",
      "                    mtries -= 1\n",
      "                    mdelay *= backoff\n",
      "            if default is not None:\n",
      "                return default\n",
      "            return f(*args, **kwargs)\n",
      "\n",
      "        return f_retry  # true decorator\n",
      "\n",
      "    return deco_retry"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 20
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "heading",
     "level": 1,
     "metadata": {},
     "source": [
      "Main program"
     ]
    },
    {
     "cell_type": "heading",
     "level": 3,
     "metadata": {},
     "source": [
      "1. Identify collections and generate a list of the file names to process"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "# Call the data image and collections\n",
      "cover = ee.Image('MCD12Q1/MCD12Q1_005_2007_01_01').select('Land_Cover_Type_1')\n",
      "                                     \n",
      "base_MODIS_EVI = ee.ImageCollection('MODIS/MCD43A4_EVI') # 16-day, 250m, 2000 - 2014\n",
      "#base_MODIS_NDVI = ee.ImageCollection('MCD43A4_NDVI')\n",
      "\n",
      "base_LS5_EVI = ee.ImageCollection('LANDSAT/LT5_L1T_8DAY_EVI') # 8-day, 30m, Jan 1984 - May 2012\n",
      "base_LS7_EVI = ee.ImageCollection('LANDSAT/LE7_L1T_8DAY_EVI') # 8-day, 30m, Jan 1991 - 2014\n",
      "\n",
      "testDateStart = datetime.datetime(2001, 1, 1)          # Define a time interval.\n",
      "testDateEnd = datetime.datetime(2001, 5, 1)          # YYYY,MM,DD\n",
      "\n",
      "\n",
      "# Target collection: CHANGE BELOW TO ALTER THE COLLECTION TO PROCESS\n",
      "call_collection = ((base_MODIS_EVI).filterDate(testDateStart, testDateEnd))\n",
      "cvar = 'EVI'  # Set string of variable which will be called from collection:\n",
      "\n",
      "info = call_collection.getInfo()    # Extract collection info to get img names \n",
      "# The image names from info pointer above are the only things used to call the images"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 8
    },
    {
     "cell_type": "heading",
     "level": 3,
     "metadata": {},
     "source": [
      "2. Set up for AreaReducer function and masking"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "# Set up for processing\n",
      "\n",
      "# Polygon over costal Tanzania area (to approx 200km inland)\n",
      "study_area = ee.Geometry.Polygon([[37.3535,-3.6889],[37.0459,-7.6674],\n",
      "    [37.7490,-11.4370],\n",
      "    [40.6934,-10.4014],\n",
      "    [41.3965,-4.7407]])\n",
      "\n",
      "#meanReducer = ee.Reducer.mean()\n",
      "#sigmaReducer = ee.Reducer.stdDev()           # Type of Reducers to apply\n",
      "#countReducer = ee.Reducer.count()            # Actually, better to do this with a combined \n",
      "                                              # reducer below for speed up.  \n",
      "reducer_group = ee.Reducer.count().combine(\n",
      "    ee.Reducer.mean(),outputPrefix='',sharedInputs=True).combine(\n",
      "    ee.Reducer.stdDev(),outputPrefix='',sharedInputs=True);\n",
      "\n",
      "scale = 250                                  # Resolution to perform the reduction (m)\n",
      "vget = 'Land_Cover_Type_1'                   # Veg data will be mapped to this var name\n",
      "\n",
      "                                             # Create mask using ee.Image.logic functions \n",
      "output_ft = cover.gte(1).And(cover.lte(7))   # Create a Forest Type (_ft) Mask\n",
      "mask_ft = output_ft.mask(output_ft)\n",
      "output_gt = cover.gte(8).And(cover.lte(10))  # Create a Grass Type (_ft) Mask\n",
      "mask_gt = output_gt.mask(output_gt)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 9
    },
    {
     "cell_type": "heading",
     "level": 3,
     "metadata": {},
     "source": [
      "For curiosity, plot the classification masks to the JS map from earlier"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "# Plot the forest mask, grass mask, and study area onto the JS map above for evaluation:\n",
      "map_item.addLayer(\n",
      "    image=mask_ft,          \n",
      "    name=\"Forest Type Mask\",\n",
      "    vis_params={'palette':\"387616\", 'opacity':0.8},\n",
      "    visible=True)\n",
      "\n",
      "map_item.addLayer(\n",
      "    image=mask_gt,\n",
      "    vis_params={'palette':\"d29e69\", 'opacity':0.8},\n",
      "    name=\"Grass Type Mask\",\n",
      "    visible=True)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 21
    },
    {
     "cell_type": "heading",
     "level": 3,
     "metadata": {},
     "source": [
      "EE-functions with an exception decorator"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "# Bewlow are the two EE-functions which I need to call from within the loop\n",
      "# They are placed here as the decorator will keep repeating the call if an\n",
      "# exception was raised (importantly the bad memory allocation exception)\n",
      "\n",
      "@retry(Exception)\n",
      "def reduce_masked(masked_image,reducer_group, study_area,scale):\n",
      "    return masked_image.reduceRegion(reducer_group, study_area,scale)\n",
      "\n",
      "@retry(Exception)\n",
      "def get_dic(reducer_return):\n",
      "    return reducer_return.getInfo() "
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 23
    },
    {
     "cell_type": "heading",
     "level": 3,
     "metadata": {},
     "source": [
      "3. Main loop to extract the masked regional stats. of all images specified"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "# A single process to call images one after the other, and ReduceRegion for some stats.\n",
      "# It takes about 1 second per image, which is a bit slow. But only needs to be run once,\n",
      "# as the output is saved as a CSV file below.\n",
      "\n",
      "Im_fnm = []                                        # Lists to hold the Image file name\n",
      "GT_mn = []                                         # and also the Grass Type (GT) \n",
      "GT_std =[]                                         # and Forest Type (FT) Means,\n",
      "GT_cnt =[]                                         # STdevs, and pixel counts\n",
      "FT_mn = []\n",
      "FT_std =[]\n",
      "FT_cnt =[]\n",
      "\n",
      "#for n in arange(0,30):                              # For a test set of n time-steps...\n",
      "for n in arange(0,len(info['features'])):          # For all images of a collection...\n",
      "    tmp_fnm =[]                                     # Wipe pointer\n",
      "    tmp_fnm = info['features'][n]['id']\n",
      "    print '\\r','Processing image:',n,tmp_fnm,            # Print an overwriting status line\n",
      "    #print 'Extracting:',n,tmp_fnm\n",
      "    tmp_img =  ee.Image(tmp_fnm).select([cvar])\n",
      "\n",
      "    interest_ft = []                                # Wipe the pointers in each loop\n",
      "    interest_gt = []                                \n",
      "    interest_ft = mask_ft.multiply(0).add(tmp_img)  # Apply the masks created above\n",
      "    interest_gt = mask_gt.multiply(0).add(tmp_img)  # to the data. Then reduce by a polygon area\n",
      "    \n",
      "    # Bit complicated, but now all the EE-calls are within functions which\n",
      "    # have a Decorator to catch exceptions, and retry.\n",
      "    gt_return = reduce_masked(interest_gt,reducer_group, study_area,scale)\n",
      "    gt_dic = get_dic(gt_return)\n",
      "    \n",
      "    ft_return=reduce_masked(interest_ft,reducer_group, study_area,scale)\n",
      "    ft_dic = get_dic(ft_return)\n",
      "    \n",
      "    GT_cnt.append(gt_dic[vget+'_count'])\n",
      "    GT_mn.append(gt_dic[vget+'_mean'])\n",
      "    GT_std.append(gt_dic[vget+'_stdDev'])\n",
      "\n",
      "    FT_cnt.append(ft_dic[vget+'_count'])\n",
      "    FT_mn.append(ft_dic[vget+'_mean'])\n",
      "    FT_std.append(ft_dic[vget+'_stdDev'])\n",
      "\n",
      "    Im_fnm.append(tmp_fnm)                           # Append the filename to the list\n",
      "\n",
      "print '\\n Finished'\n",
      "# nb. this code currently isnt too fast: currently takes about 1 sec per day"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "Processing image: 0 MCD43A4_EVI/MCD43A4_005_2001_01_01 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "Processing image: 1 MCD43A4_EVI/MCD43A4_005_2001_01_09 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "Processing image: 2 MCD43A4_EVI/MCD43A4_005_2001_01_17 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "Processing image: 3 MCD43A4_EVI/MCD43A4_005_2001_01_25 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "Processing image: 4 MCD43A4_EVI/MCD43A4_005_2001_02_02 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "Processing image: 5 MCD43A4_EVI/MCD43A4_005_2001_02_10 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "Processing image: 6 MCD43A4_EVI/MCD43A4_005_2001_02_18 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "Processing image: 7 MCD43A4_EVI/MCD43A4_005_2001_02_26 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "Processing image: 8 MCD43A4_EVI/MCD43A4_005_2001_03_06 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "Processing image: 9 MCD43A4_EVI/MCD43A4_005_2001_03_14 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "Processing image: 10 MCD43A4_EVI/MCD43A4_005_2001_03_22 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "Processing image: 11 MCD43A4_EVI/MCD43A4_005_2001_03_30 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "Processing image: 12 MCD43A4_EVI/MCD43A4_005_2001_04_07 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "Processing image: 13 MCD43A4_EVI/MCD43A4_005_2001_04_15 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "Processing image: 14 MCD43A4_EVI/MCD43A4_005_2001_04_23 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        " Finished\n"
       ]
      }
     ],
     "prompt_number": 24
    },
    {
     "cell_type": "heading",
     "level": 3,
     "metadata": {},
     "source": [
      "Process the file names into a list of Datetime objects"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "# Break the list of file names into a list of string arrays, and use them to make datetime indexes\n",
      "str_tmp = [str.split(str(n),'_')[3:6] for n in Im_fnm]\n",
      "\n",
      "if((Im_fnm[0][0:3] == 'LT5') | (Im_fnm[0][0:3] == 'LE7')):  # Check if this is Landsat or Modis\n",
      "    print '\\n','Creating date objects from names of LandSat files'\n",
      "    dt_list = [datetime.datetime(int(n[0][4:8]),int(n[0][8:10]),\n",
      "                                 int(n[0][10:12])).date() for n in str_tmp]\n",
      "\n",
      "if(Im_fnm[0][0:7] == 'MCD43A4'):\n",
      "    print '\\n','Creating date objects from names of MODIS files'\n",
      "    dt_list = [datetime.datetime(int(n[0]),int(n[1]),int(n[2])).date() \n",
      "           for n in str_tmp]"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Creating date objects from names of MODIS files\n"
       ]
      }
     ],
     "prompt_number": 25
    },
    {
     "cell_type": "heading",
     "level": 3,
     "metadata": {},
     "source": [
      "4. Convert the data to a Pandas DF object: display, plot, and export data as csv"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import pandas as pd"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 15
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "d = {'GT_mean': GT_mn,'GT_std':GT_std,'GT_cnt':GT_cnt,\n",
      "     'FT_mean': FT_mn,'FT_std':FT_std,'FT_cnt':FT_cnt}\n",
      "df = pd.DataFrame(data=d, index=dt_list)  # Use the Pandas dataframe object"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 26
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "df   # Show the Pandas Dataframe object"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "html": [
        "<div style=\"max-height:1000px;max-width:1500px;overflow:auto;\">\n",
        "<table border=\"1\" class=\"dataframe\">\n",
        "  <thead>\n",
        "    <tr style=\"text-align: right;\">\n",
        "      <th></th>\n",
        "      <th>FT_cnt</th>\n",
        "      <th>FT_mean</th>\n",
        "      <th>FT_std</th>\n",
        "      <th>GT_cnt</th>\n",
        "      <th>GT_mean</th>\n",
        "      <th>GT_std</th>\n",
        "    </tr>\n",
        "  </thead>\n",
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>2001-01-01</th>\n",
        "      <td> 19881</td>\n",
        "      <td> 0.458445</td>\n",
        "      <td> 0.161085</td>\n",
        "      <td>  331777</td>\n",
        "      <td> 0.424951</td>\n",
        "      <td> 0.136074</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>2001-01-09</th>\n",
        "      <td> 24289</td>\n",
        "      <td> 0.458373</td>\n",
        "      <td> 0.137066</td>\n",
        "      <td>  377304</td>\n",
        "      <td> 0.439683</td>\n",
        "      <td> 0.120790</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>2001-01-17</th>\n",
        "      <td> 32019</td>\n",
        "      <td> 0.483665</td>\n",
        "      <td> 0.126115</td>\n",
        "      <td>  493425</td>\n",
        "      <td> 0.463690</td>\n",
        "      <td> 0.109993</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>2001-01-25</th>\n",
        "      <td> 41347</td>\n",
        "      <td> 0.492292</td>\n",
        "      <td> 0.118211</td>\n",
        "      <td>  596911</td>\n",
        "      <td> 0.467676</td>\n",
        "      <td> 0.103682</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>2001-02-02</th>\n",
        "      <td> 42044</td>\n",
        "      <td> 0.502341</td>\n",
        "      <td> 0.129903</td>\n",
        "      <td>  725672</td>\n",
        "      <td> 0.484457</td>\n",
        "      <td> 0.112159</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>2001-02-10</th>\n",
        "      <td> 31569</td>\n",
        "      <td> 0.525633</td>\n",
        "      <td> 0.162721</td>\n",
        "      <td>  968665</td>\n",
        "      <td> 0.531841</td>\n",
        "      <td> 0.128645</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>2001-02-18</th>\n",
        "      <td> 40273</td>\n",
        "      <td> 0.504992</td>\n",
        "      <td> 0.168065</td>\n",
        "      <td> 1053807</td>\n",
        "      <td> 0.504121</td>\n",
        "      <td> 0.152199</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>2001-02-26</th>\n",
        "      <td> 43555</td>\n",
        "      <td> 0.482512</td>\n",
        "      <td> 0.151715</td>\n",
        "      <td>  784045</td>\n",
        "      <td> 0.452918</td>\n",
        "      <td> 0.152385</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>2001-03-06</th>\n",
        "      <td> 31237</td>\n",
        "      <td> 0.484999</td>\n",
        "      <td> 0.170691</td>\n",
        "      <td>  671452</td>\n",
        "      <td> 0.466472</td>\n",
        "      <td> 0.158515</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>2001-03-14</th>\n",
        "      <td> 33987</td>\n",
        "      <td> 0.528097</td>\n",
        "      <td> 0.175400</td>\n",
        "      <td>  975660</td>\n",
        "      <td> 0.518982</td>\n",
        "      <td> 0.142452</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>2001-03-22</th>\n",
        "      <td> 27488</td>\n",
        "      <td> 0.522366</td>\n",
        "      <td> 0.178839</td>\n",
        "      <td>  918749</td>\n",
        "      <td> 0.524023</td>\n",
        "      <td> 0.136521</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>2001-03-30</th>\n",
        "      <td> 19894</td>\n",
        "      <td> 0.547580</td>\n",
        "      <td> 0.168355</td>\n",
        "      <td>  903519</td>\n",
        "      <td> 0.543153</td>\n",
        "      <td> 0.109931</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>2001-04-07</th>\n",
        "      <td> 18800</td>\n",
        "      <td> 0.549720</td>\n",
        "      <td> 0.152068</td>\n",
        "      <td>  954901</td>\n",
        "      <td> 0.541962</td>\n",
        "      <td> 0.097375</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>2001-04-15</th>\n",
        "      <td> 16570</td>\n",
        "      <td> 0.524785</td>\n",
        "      <td> 0.161284</td>\n",
        "      <td>  922199</td>\n",
        "      <td> 0.518054</td>\n",
        "      <td> 0.100356</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>2001-04-23</th>\n",
        "      <td> 30237</td>\n",
        "      <td> 0.552585</td>\n",
        "      <td> 0.137074</td>\n",
        "      <td> 1308932</td>\n",
        "      <td> 0.505311</td>\n",
        "      <td> 0.094231</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
        "</div>"
       ],
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 27,
       "text": [
        "            FT_cnt   FT_mean    FT_std   GT_cnt   GT_mean    GT_std\n",
        "2001-01-01   19881  0.458445  0.161085   331777  0.424951  0.136074\n",
        "2001-01-09   24289  0.458373  0.137066   377304  0.439683  0.120790\n",
        "2001-01-17   32019  0.483665  0.126115   493425  0.463690  0.109993\n",
        "2001-01-25   41347  0.492292  0.118211   596911  0.467676  0.103682\n",
        "2001-02-02   42044  0.502341  0.129903   725672  0.484457  0.112159\n",
        "2001-02-10   31569  0.525633  0.162721   968665  0.531841  0.128645\n",
        "2001-02-18   40273  0.504992  0.168065  1053807  0.504121  0.152199\n",
        "2001-02-26   43555  0.482512  0.151715   784045  0.452918  0.152385\n",
        "2001-03-06   31237  0.484999  0.170691   671452  0.466472  0.158515\n",
        "2001-03-14   33987  0.528097  0.175400   975660  0.518982  0.142452\n",
        "2001-03-22   27488  0.522366  0.178839   918749  0.524023  0.136521\n",
        "2001-03-30   19894  0.547580  0.168355   903519  0.543153  0.109931\n",
        "2001-04-07   18800  0.549720  0.152068   954901  0.541962  0.097375\n",
        "2001-04-15   16570  0.524785  0.161284   922199  0.518054  0.100356\n",
        "2001-04-23   30237  0.552585  0.137074  1308932  0.505311  0.094231"
       ]
      }
     ],
     "prompt_number": 27
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "# Create a plot from the Pandas DF object\n",
      "fig_ts = plt.figure(figsize=[10,5])\n",
      "ax1 = fig_ts.add_subplot(111)\n",
      "#ax1.plot(dt_list,Tz_mn,'b.')\n",
      "ax1.errorbar(df.index,df.FT_mean,yerr=df.FT_std,xerr=None,ls='-',c='b')\n",
      "ax1.errorbar(df.index,df.GT_mean,yerr=df.GT_std,xerr=None,ls='-',c='r')\n",
      "ax1.set_ylabel(r'EVI ($\\mu\\pm1\\sigma$)')\n",
      "ax1.set_xlabel('Date')\n",
      "ax1.set_title('250m 16-day MODIS EVI: Tanzania coastal region Forest-type (blue) & Grass-type (red)')\n",
      "plt.grid(True)\n",
      "plt.show(fig_ts)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAnYAAAFRCAYAAAD5FeDqAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
       "text": [
        "<matplotlib.figure.Figure at 0x10984eb50>"
       ]
      }
     ],
     "prompt_number": 28
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "df.to_csv('Test_MODIS_2001b.csv')"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 54
    }
   ],
   "metadata": {}
  }
 ]
}