Watching the Landscape Tell Its Story Over Time

Your first chart: NDVI across a year

Let's chart the seasonal rise and fall of vegetation greenness for a forest point in north-central Florida over the year 2023. We'll see the whole seasonal rhythm in one picture.

// Cloud masking function for Landsat 8 SR
function maskL8sr(image) {
  var qaMask = image.select('QA_PIXEL').bitwiseAnd(1 << 3).eq(0)
      .and(image.select('QA_PIXEL').bitwiseAnd(1 << 4).eq(0));
  return image.updateMask(qaMask)
      .multiply(0.0000275).add(-0.2);
}

// Define a forest point
var forest = ee.Geometry.Point([-82.4, 29.7]);

// Load Landsat 8 for 2023, add NDVI band
var collection = ee.ImageCollection('LANDSAT/LC08/C02/T1_L2')
  .filterDate('2023-01-01', '2023-12-31')
  .filterBounds(forest)
  .filter(ee.Filter.lt('CLOUD_COVER', 30))
  .map(maskL8sr)
  .map(function(img) {
    var ndvi = img.normalizedDifference(['SR_B5', 'SR_B4']).rename('NDVI');
    return img.addBands(ndvi);
  });

// Create chart
var chart = ui.Chart.image.series({
  imageCollection: collection.select('NDVI'),
  region: forest,
  reducer: ee.Reducer.mean(),
  scale: 30
}).setOptions({
  title: 'NDVI Over 2023 - Florida Forest',
  vAxis: {title: 'NDVI'},
  hAxis: {title: 'Date'},
  lineWidth: 2,
  pointSize: 4
});

print(chart);

Why bother charting over time?

Maps show us spatial patterns. Charts show us temporal patterns. Together, they give us the complete picture. Here are real questions that time series charts can answer:

• Is this forest getting greener or browning? A multi-year NDVI chart reveals the long-term trajectory.
• When does the growing season start? The date when NDVI crosses a threshold tells us the onset of spring.
• How fast did vegetation recover after a fire? Charting NDVI from the fire year forward shows the recovery curve.
• How does cropland differ from forest? Side-by-side charts reveal fundamentally different seasonal signatures.

Here's the cool part: GEE provides built-in charting functions that plot values directly from image collections. No exporting required.

How the charting function works

The ui.Chart.image.series() function extracts values from every image in a collection at a specified location and plots them over time. Let's break down the key parameters:

• imageCollection: The collection of images (select only the band you want to plot).
• region: The point or polygon where values are extracted.
• reducer: How to summarize pixel values within the region (use ee.Reducer.mean() for a point).
• scale: The resolution for the extraction (30 for Landsat).

// Basic chart structure
var chart = ui.Chart.image.series({
  imageCollection: collection.select('NDVI'),
  region: forest,
  reducer: ee.Reducer.mean(),
  scale: 30
});

// Customize appearance
chart = chart.setOptions({
  title: 'NDVI Time Series',
  vAxis: {title: 'NDVI', minValue: 0, maxValue: 1},
  hAxis: {title: 'Date'},
  lineWidth: 2,
  pointSize: 4,
  series: {0: {color: 'green'}}
});

print(chart);

Reading the chart: Each point represents one cloud-free Landsat image. Gaps in the chart mean no clear image was available on that date. Spikes downward? Those often indicate residual cloud contamination that the mask didn't catch.

Comparing places side by side

Here's where it gets interesting. The ui.Chart.image.seriesByRegion() function plots multiple locations on the same chart. This is incredibly powerful for comparing how different land cover types respond to seasons and disturbances.

// Define three land cover points
var forest = ee.Geometry.Point([-82.4, 29.7]);
var urban = ee.Geometry.Point([-82.32, 29.65]);
var cropland = ee.Geometry.Point([-82.5, 29.5]);

// Create a FeatureCollection with labels
var regions = ee.FeatureCollection([
  ee.Feature(forest, {'label': 'Forest'}),
  ee.Feature(urban, {'label': 'Urban'}),
  ee.Feature(cropland, {'label': 'Cropland'})
]);

// Build the multi-region chart
var regionChart = ui.Chart.image.seriesByRegion({
  imageCollection: collection.select('NDVI'),
  regions: regions,
  reducer: ee.Reducer.mean(),
  band: 'NDVI',
  scale: 30,
  seriesProperty: 'label'
}).setOptions({
  title: 'NDVI by Land Cover Type (2023)',
  vAxis: {title: 'NDVI'},
  hAxis: {title: 'Date'},
  lineWidth: 2,
  pointSize: 4
});

print(regionChart);

Is this place greening or browning?

A chart shows us the pattern visually, but how do we quantify whether a location is actually greening or browning? The ee.Reducer.linearFit() reducer fits a straight line through the data and returns two values:

• Slope: The rate of change per unit of time. Positive slope means NDVI is increasing (greening). Negative slope means NDVI is decreasing (browning).
• Offset: The y-intercept, representing the baseline value.

To use linearFit(), we need to give the collection a time band. GEE uses the system:time_start property as the x-axis automatically when we add a time band to each image. Don't worry, we'll walk through it step by step.

// Build multi-year collection (2020-2023)
var longCollection = ee.ImageCollection('LANDSAT/LC08/C02/T1_L2')
  .filterDate('2020-01-01', '2023-12-31')
  .filterBounds(forest)
  .filter(ee.Filter.lt('CLOUD_COVER', 30))
  .map(maskL8sr)
  .map(function(img) {
    var ndvi = img.normalizedDifference(['SR_B5', 'SR_B4']).rename('NDVI');
    // Add time band (fractional years)
    var timeImage = ee.Image.constant(
      img.date().difference('2020-01-01', 'year')
    ).float().rename('time');
    return ndvi.addBands(timeImage);
  });

// Apply linear regression
// linearFit expects bands in order: [independent (time), dependent (NDVI)]
var trend = longCollection.select(['time', 'NDVI'])
  .reduce(ee.Reducer.linearFit());

// Print the trend at the forest point
print('Trend at forest:', trend.reduceRegion({
  reducer: ee.Reducer.mean(),
  geometry: forest,
  scale: 30,
  maxPixels: 1e9
}));

Interpreting the result: A slope of 0.01 means NDVI increases by 0.01 per year. Over four years, that's a total increase of 0.04 in NDVI. A slope near zero means no significant change. Simple as that.

Let's see the whole landscape at once

Here's where remote sensing gets powerful. We can apply linearFit() to every pixel at once, not just a single point. The result is a trend map where each pixel's color tells us whether that location is greening, stable, or browning.

// The 'trend' image from the previous step has two bands:
// 'scale' = slope, 'offset' = intercept

// Visualize the slope band with a diverging palette
Map.centerObject(forest, 10);
Map.addLayer(trend.select('scale'), {
  min: -0.05, max: 0.05,
  palette: ['red', 'orange', 'white', 'lightgreen', 'darkgreen']
}, 'NDVI Trend (Slope)');

// Add a reference point
Map.addLayer(forest, {color: 'blue'}, 'Forest Point');

What's next?

• Change Detection - Compare images from two dates to detect environmental change.
• Lab 22: Time Series Charting - Build multi-region charts and trend maps independently.