Exporting NDVI time series from random points in a region

Question

I am wondering how to export a time series of NDVI across a 20 year time series (1990 - 2010) for each point in my region. To be explicit: There are 1,000 random points in this arbitrary region. I need to export time vs NDVI for each image in this Landsat Collection for each point. I am getting stuck with the method of extraction, because I do not want to reduce an image stack at each point. I have also seen individual point time series chart, but I would like this equivalent for each point. This is what I have so far:

//Mask clouds in Landsat 5 imagery.
var maskClouds = function(image) {
  var scored = ee.Algorithms.Landsat.simpleCloudScore(image);
  return image.updateMask(scored.select(['cloud']).lt(20));
};
//Mask clouds and adds quality bands to Landsat 8 images.
var addQualityBand = function(image) {
  return maskClouds(image)
    //NDVI
    .addBands(image.normalizedDifference(['B4', 'B3']))
    //time in days
    .addBands(image.metadata('system:time_start'));
};
var images= ee.ImageCollection('LANDSAT/LT05/C01/T1_TOA')
  .filterDate('1990-01-01','2011-01-01');

var collection= images.map(addQualityBand);

var randomPoints = ee.FeatureCollection.randomPoints(geometry,1000);

// Add to map
Map.centerObject(randomPoints);
Map.addLayer(randomPoints, {}, 'random points',true);

Here is a link to the full script with the geometry import: https://code.earthengine.google.com/47c441094aef35980e528e3fd9bb9b0a

Answer 1

Here is a script that you might find useful:

/**
 * @license
 * Copyright 2020 Google LLC.
 * SPDX-License-Identifier: Apache-2.0
 * 
 * @description
 * This script calculates image statistics from the 9 pixels surrounding points
 * in a feature collection for all images in a Landsat surface reflectance
 * collection (TM, ETM+, OLI). Images are cloud masked using CFmask. All images
 * dates are included. The result is a featureCollection; essentially a table
 * with one row per unique observation defined by combinations of image and
 * point. Columns include the image statistic per band, point ID, and image
 * properties (ID, date, satellite).
 */


// #############################################################################
// ### POINT FEATURE COLLECTION SETUP ###
// #############################################################################

// `pts` is a FeatureCollection import from drawing tools.
pts = pts
  // Add a buffer to each point (result is a 90x90m square around each point).
  .map(function(pt){
    return pt.buffer(45).bounds();
  });

// Display the points to the Map.
print(pts.limit(5));
Map.centerObject(pts);
Map.addLayer(pts);


// #############################################################################
// ### DEFINE FUNCTIONS TO PREPARE THE LANDSAT IMAGE COLLECTION  ###
// #############################################################################

// Define function to mask cloud and shadow pixels out of images using CFmask.
function fmask(img) {
  var cloudShadowBitMask = 1 << 3;
  var cloudsBitMask = 1 << 5;
  var qa = img.select('pixel_qa');
  var mask = qa.bitwiseAnd(cloudShadowBitMask).eq(0)
    .and(qa.bitwiseAnd(cloudsBitMask).eq(0));
  return img.updateMask(mask)}

// Function to get and rename bands of interest from OLI.
function renameOli(img) {
  return img.select(
    ['B2', 'B3', 'B4', 'B5', 'B6', 'B7'],
    ['Blue', 'Green', 'Red', 'NIR', 'SWIR1', 'SWIR2'])}

// Function to get and rename bands of interest from TM/ETM+.
function renameEtm(img) {
  return img.select(
    ['B1', 'B2', 'B3', 'B4', 'B5', 'B7'],
    ['Blue', 'Green', 'Red', 'NIR', 'SWIR1', 'SWIR2'])}

// Define function to prepare (cloud mask and rename) OLI images.
function prepOli(img) {
  img = fmask(img);
  img = renameOli(img);
  return img}

// Define function to prepare (cloud mask and rename) TM/ETM+ images.
function prepEtm(img) {
  img = fmask(img);
  img = renameEtm(img);
  return img}


// #############################################################################
// ### FILTER AND PREPARE LANDSAT IMAGE COLLECTIONS ###
// #############################################################################

// Get Landsat surface reflectance collections for OLI, ETM+ and TM sensors.
// filter them by the bounds of the point feature collection and apply the
// relevant image preparation function to each image.
var LC08col = ee.ImageCollection('LANDSAT/LC08/C01/T1_SR')
  .filterBounds(pts)
  .map(prepOli);
var LE07col = ee.ImageCollection('LANDSAT/LE07/C01/T1_SR')
  .filterBounds(pts)
  .map(prepEtm);
var LT05col = ee.ImageCollection('LANDSAT/LT05/C01/T1_SR')
  .filterBounds(pts)
  .map(prepEtm);

// Merge the prepared sensor collections.
var col = LC08col.merge(LE07col).merge(LT05col);

// #############################################################################
// ### REDUCE REGION: FOR EACH IMAGE AROUND EACH POINT ###
// #############################################################################

// Map over the image collection; for each image find the intersecting points
// and do a region reduction at each point.
var statCol = col.map(function(img) {
  // Get metadata for the image.
  var imgDate = img.date().format('YYYY-MM-dd'); 
  var imgId = img.getString('LANDSAT_ID');
  var imgSat = img.getString('SATELLITE');
  // Filter the points to those that intersect the image.
  var thesePts = pts.filterBounds(img.geometry());

  // Reduce the image by mean for the points that intersect it. 
  var bandStats = img.reduceRegions({
    collection: thesePts,
    reducer: ee.Reducer.mean(),
    scale: 30,
    crs: 'EPSG:5070'
  })
  // Add metadata to each point.
  .map(function(pt) {
    return pt.set({
      'imgDate': imgDate,
      'imgId': imgId,
      'imgSat': imgSat,
      'plotId': pt.getString('PlotID')
    });
  });

  // Return the featureCollection.
  return bandStats;
});

// Flatten the collection of collections; result is a featureCollection. Think
// of it as a table with one row per unique observation defined by combinations
// of image and point. Columns include the image region statistic per band and
// image properties.
statCol = statCol.flatten();

// Filter out observations where bands stats are null (point fell on masked
// pixels);
statCol = statCol.filter(
  ee.Filter.notNull(['Blue', 'Green', 'Red', 'NIR', 'SWIR1', 'SWIR2']));

// Print a sample of the points.
print(statCol.limit(10));


// #############################################################################
// ### BATCH TASK - EXPORT AS ASSET OR GOOGLE DRIVE FILE ###
// #############################################################################

// It's quite likely that for point feature collections that
// cover a large area or contain many points, you'll need to complete this
// operation as a batch task by either exporting the final feature collection as
// an asset or as a CSV/Shapefile/GeoJSON to Google Drive. If your browser
// times out, definitely try exporting the results.

// Export as asset option.
Export.table.toAsset({
  collection: statCol,
  description: 'point_summary_table_asset',
  assetId: 'point_summary_table'
});

// Export to Google Drive option.
Export.table.toDrive({
  collection: statCol,
  description: 'point_summary_table_gdrive',
  fileFormat: 'CSV'});

Code Editor link