Urban Health Surveillance with Environmental and Geospatial Data

// Urban Health Surveillance: Multi-Sensor Environmental Exposure Mapping
// Target Region: London Metropolitan Area
var urbanPoint = ee.Geometry.Point([-0.1278, 51.5074]);
var roi = urbanPoint.buffer(25000).bounds();

// Time window for summer season
var startDate = '2023-06-01';
var endDate = '2023-08-31';

// 1. Sentinel-5P Nitrogen Dioxide (NO2) Exposure
var s5p = ee.ImageCollection('COPERNICUS/S5P/OFFL/L3_NO2')
  .filterBounds(roi)
  .filterDate(startDate, endDate)
  .select('tropospheric_NO2_column_number_density');
var meanNO2 = s5p.mean().clip(roi);

// 2. MODIS Land Surface Temperature (LST) Exposure
var modis = ee.ImageCollection('MODIS/061/MOD11A2')
  .filterBounds(roi)
  .filterDate(startDate, endDate)
  .select('LST_Day_1km');

var convertToCelsius = function(img) {
  var celsius = img.multiply(0.02).subtract(273.15);
  return celsius.copyProperties(img, ['system:time_start']);
};
var meanLST = modis.map(convertToCelsius).mean().clip(roi);

// 3. GHSL Built-up Intensity
var ghsl = ee.Image('JRC/GHSL/P2023A/GHS_BUILT_S/2020')
  .select('built_surface')
  .clip(roi);

// Center Map
Map.centerObject(roi, 10);

// Visualization Parameters
var vizNO2 = {min: 0.00003, max: 0.00015, palette: ['blue', 'yellow', 'red']};
var vizLST = {min: 18, max: 35, palette: ['blue', 'green', 'yellow', 'orange', 'red']};
var vizBuilt = {min: 0, max: 100, palette: ['white', 'gray', 'black']};

// Add Layers
Map.addLayer(meanLST, vizLST, 'Land Surface Temp (C)');
Map.addLayer(meanNO2, vizNO2, 'Tropospheric NO2 Density');
Map.addLayer(ghsl, vizBuilt, 'GHSL Built-Up Surface');

// Continuation: Min-Max Normalization and GeoAI Modeling

// Helper function for Min-Max Normalization
var normalize = function(img, geometry) {
  var minMax = img.reduceRegion({
    reducer: ee.Reducer.minMax(),
    geometry: geometry,
    scale: 1000,
    maxPixels: 1e9
  });
  var keys = minMax.keys();
  var minVal = ee.Number(minMax.get(keys.get(0)));
  var maxVal = ee.Number(minMax.get(keys.get(1)));
  return img.subtract(minVal).divide(maxVal.subtract(minVal)).rename('normalized');
};

var lstNorm = normalize(meanLST, roi);
var no2Norm = normalize(meanNO2, roi);
var builtNorm = normalize(ghsl, roi);

// Compute Composite Health Vulnerability Index (UVI)
var uvi = lstNorm.multiply(0.4)
  .add(no2Norm.multiply(0.4))
  .add(builtNorm.multiply(0.2))
  .rename('UVI');

// Load Elevation Covariate
var elevation = ee.Image('USGS/SRTMGL1_003').select('elevation').clip(roi);

// Stack features for GeoAI Modeling
var featureStack = meanLST.rename('LST')
  .addBands(meanNO2.rename('NO2'))
  .addBands(ghsl.rename('Built'))
  .addBands(elevation);

// Generate Stratified Random Training Sample
var samples = featureStack.sample({
  region: roi,
  scale: 1000,
  numPixels: 1000,
  geometries: true
});

// Train Random Forest Regressor to predict Thermal Risk (LST)
var rfModel = ee.Classifier.smileRandomForest(50).setOutputMode('REGRESSION');
var trainedModel = rfModel.train({
  features: samples,
  classProperty: 'LST',
  inputProperties: ['NO2', 'Built', 'elevation']
});

var predictedLST = featureStack.classify(trainedModel).rename('Predicted_LST');

// Visualization
var vizUVI = {min: 0, max: 0.8, palette: ['navy', 'cyan', 'lime', 'yellow', 'red']};
Map.addLayer(uvi, vizUVI, 'Urban Health Vulnerability Index (UVI)');
Map.addLayer(predictedLST, vizLST, 'GeoAI Predicted LST');

// Export Urban Health Vulnerability Index Raster to Google Drive
Export.image.toDrive({
  image: uvi,
  description: 'Urban_Health_Vulnerability_Index_2023',
  folder: 'GEE_Urban_Health',
  fileNamePrefix: 'UVI_London_2023',
  scale: 1000,
  region: roi,
  maxPixels: 1e9,
  fileFormat: 'GeoTIFF'
});