Google Earth Engine Tutorial: Estimate Soil Moisture with TVDI

Introduction to Using TVDI for Soil Moisture Estimation in Google Earth Engine

Soil moisture estimation is critical for agriculture, hydrology, and climate studies. The Temperature-Vegetation Dryness Index (TVDI) is a remote sensing technique that combines land surface temperature (LST) and vegetation indices to infer soil moisture levels. In this tutorial, we walk through the steps to calculate TVDI using Google Earth Engine (GEE), leveraging satellite data for spatial analysis.

Step 1: Data Collection

To estimate soil moisture using TVDI, you need satellite data with land surface temperature and vegetation index layers. Common sources include:

• Landsat 8/9 (for complementary LST and NDVI)
• MODIS (daily LST and NDVI products)
• Sentinel-2 (high-resolution vegetation and thermal data)

Begin by importing and selecting the relevant data. For example, use ee.ImageCollection to access Landsat 8 surface reflectance and thermal bands:

var landsat8 = ee.ImageCollection("LANDSAT/LC08/C02/T1_L2")
  .filterDate('2022-06-01', '2022-06-30');

Step 2: Calculate Surface Temperature

Surface temperature (LST) can be derived from thermal infrared bands. For Landsat 8, use the TIRS band 10 or 11:

var lst = landsat8.select('ST_B10'); // or 'ST_B11'

Convert raw digital numbers (DN) to actual temperature using the 'ST_01' and 'ST_02' coefficients:

var lstConverted = lst.map(function(image) {
  var coeff = image.select('ST_01', 'ST_02');
  return image.select('ST_B10').multiply(coeff.select('ST_01')).add(coeff.select('ST_02'));
});

Step 3: Generate Vegetation Index (NDVI)

Normalized Difference Vegetation Index (NDVI) measures vegetation density. Calculate NDVI using the red and near-infrared bands:

var ndvi = landsat8.map(function(image) {
  return image.select(['SR_B4', 'SR_B5']).reduce(ee.Reduce.ndvi());
});

Step 4: Calculate TVDI

TVDI is calculated using the formula: TVDI = (T - T_min) / (T_max - T_min). First, create a temperature-NDVI scatter plot, then fit a regression line to derive T_max and T_min for each NDVI value:

var tvdi = lstConverted.select('ST_B10')
  .map(function(image) {
    var ndviImage = ndvi.select('ndvi').clip(image.geometry());
    var scatter = ee.ImageCollection.fromImages([image.select('ST_B10'), ndviImage]).toBands();
    var regression = scatter.select(['ST_B10', 'ndvi']).reduce(ee.Reduce.linearRegression(2, 1));
    var slope = regression.select('slope');
    var intercept = regression.select('intercept');
    var tmax = intercept.add(slope.multiply(ndviImage));
    var tmin = ...; // Derived from minimum LST at high NDVI values
    return image
      .expression('(T - Tmin) / (Tmax - Tmin)', {
        'T': image.select('ST_B10'),
        'Tmin': tmin,
        'Tmax': tmax
      });
  });

Step 5: Visualize and Export Results

Use the Map.addLayer() function to visualize TVDI results. Apply a color palette to distinguish dry and wet areas:

Map.addLayer(tvdi, {min: 0, max: 1, palette: ['blue', 'green', 'yellow', 'red']}, 'TVDI');

Export the results as a GeoTIFF or CSV for further analysis:

Export.image.toDrive({
  image: tvdi,
  description: 'TVDI_Output',
  folder: 'GEE_Exports',
  fileNamePrefix: 'TVDI',
  region: geometry,
  scale: 30,
  maxPixels: 1e9
});

Step 6: Interpret TVDI Maps

TVDI ranges from 0 (wet soil) to 1 (dry soil). Higher values indicate lower soil moisture, while lower values suggest wetter conditions. Overlay TVDI with land use/land cover data to analyze regional patterns.