基于GEE使用Landsat 8和Landsat 5影像计算RSEI

一、使用2018年landsat 8影像数据计算遥感生态指数RSEI

//could masking，GEE上有现有的去云函数
function maskL8sr(image) {
  // Bit 0 - Fill
  // Bit 1 - Dilated Cloud
  // Bit 2 - Cirrus
  // Bit 3 - Cloud
  // Bit 4 - Cloud Shadow
  var qaMask = image.select('QA_PIXEL').bitwiseAnd(parseInt('11111', 2)).eq(0);
  var saturationMask = image.select('QA_RADSAT').eq(0);

  // Apply the scaling factors to the appropriate bands.
  var opticalBands = image.select('SR_B.').multiply(0.0000275).add(-0.2);
  var thermalBands = image.select('ST_B.*').multiply(0.00341802).add(149.0);

  // Replace the original bands with the scaled ones and apply the masks.
  return image.addBands(opticalBands, null, true)
      .addBands(thermalBands, null, true)
      .updateMask(qaMask)
      .updateMask(saturationMask);
}

var dataset1 = ee.ImageCollection('LANDSAT/LC08/C02/T1_L2')
    .filterDate('2018-x-x', '2018-x-x')
    .filterBounds(roi)
    .filter(ee.Filter.lte('CLOUD_COVER',5))
    .sort('CLOUD_COVER')
    .map(maskL8sr)
    //.mean()
print(dataset1)

var image1=dataset1.mosaic().clip(roi)
print(image1,'maskL8sr')
//Map.addLayer(image1,{bands:["SR_B4","SR_B3","SR_B2"],gamma:1.3,max:108,min:15},'Image1');
//Map.addLayer(roi,{color:'yellow',fillColor: "00000000", width: 1},'roi');

var jrc = jrcwater18.select('b1')
var image = image1.updateMask(jrc.eq(1))
//Map.addLayer(image,{bands:["SR_B4","SR_B3","SR_B2"],gamma:1.3,max:108,min:15},'Image');
print(image,'water and cloud')

//计算SI

function SI_cal(img) {
 var blue = img.select("SR_B2");
 var red = img.select("SR_B4");
 var nir = img.select("SR_B5");
 var swir1 = img.select("SR_B6");
 var swir2 = img.select("SR_B7");
 var SI_temp =((swir1.add(red)).subtract(blue.add(nir)))
              .divide((swir1.add(red)).add(blue.add(nir)))
 //print(SI_temp);
 return SI_temp;
}

//计算IBI
function IBI_cal(img) {
 var green = img.select("SR_B3");
 var red = img.select("SR_B4");
 var nir = img.select("SR_B5");
 var swir1 = img.select("SR_B6");
 var IBI_temp =(((swir1.multiply(2.0)).divide(swir1.add(nir)))
               .subtract((nir.divide(nir.add(red))).add(green.divide(green.add(swir1)))))
               .divide((((swir1.multiply(2.0)).divide(swir1.add(nir)))
               .add((nir.divide(nir.add(red))).add(green.divide(green.add(swir1))))))
 //print(IBI_temp);
 return IBI_temp;
}

//计算湿度指数Wet

function Wet_cal(img) {
 var blue = img.select("SR_B2");
 var green = img.select("SR_B3");
 var red = img.select("SR_B4");
 var nir = img.select("SR_B5");
 var swir1 = img.select("SR_B6");
 var swir2 = img.select("SR_B7");
 var wet_temp =blue.multiply(0.1511)
              .add(green.multiply(0.1973))
              .add(red.multiply(0.3283))
              .add(nir.multiply(0.3407))
              .add(swir1.multiply(-0.7117))
              .add(swir2.multiply(-0.4559))
 //print(wet_temp);
 return wet_temp;
}

//外部导入的NDVI重采样

var NDVI=ndvi.reproject('EPSG:4326',null,1000);  

//normalization function
function img_normalize(img){
      var minMax = img.reduceRegion({
            reducer:ee.Reducer.minMax(),
            geometry: roi,
            scale: 1000,
            maxPixels: 10e13,
        })
      var year = img.get('year')
      var normalize  = ee.ImageCollection.fromImages(
            img.bandNames().map(function(name){
                  name = ee.String(name);
                  var band = img.select(name);
                  return band.unitScale(ee.Number(minMax.get(name.cat('_min'))), ee.Number(minMax.get(name.cat('_max'))));
              })
        ).toBands().rename(img.bandNames());
        return normalize;
} 

//使用MODIS的地表温度数据

var dataset2 = ee.ImageCollection('MODIS/061/MOD11A2')
    .filterDate('2018-x-x', '2018-x-x')
    .mean()
//var dataset_no_water = dataset.updateMask(NDWI_mask).clip(table);
var dataset2_no_water = dataset2.updateMask(jrc.gte(0.1)).clip(roi);

//将地表温度单位转换为摄氏度
var lst = dataset2_no_water.expression(
    'a*0.02-273.15',
    {
        a:dataset2_no_water.select('LST_Day_1km'), 
    });

var SI = SI_cal(image);
print(SI,'SI');
var IBI = IBI_cal(image);
var NDBSI =(SI.add(IBI)).divide(2.0);
var wet = Wet_cal(image);

var unit_ndvi = img_normalize(NDVI);
image=image.addBands(unit_ndvi.rename('NDVI').toFloat())
var unit_NDBSI = img_normalize(NDBSI);
image=image.addBands(unit_NDBSI.rename('NDBSI').toFloat())
var unit_Wet = img_normalize(wet);
image=image.addBands(unit_Wet.rename('Wet').toFloat())
var unit_lst = img_normalize(lst);
image=image.addBands(unit_lst.rename('LST').toFloat())
print(image)
var bands = ["Wet","NDVI","NDBSI","LST"]
var sentImage =image.select(bands)

var region = roi;
var image2=  sentImage.select(bands);
var scale = 1000;
var bandNames = image2.bandNames();
// 图像波段重命名函数
var getNewBandNames = function(prefix) {
    var seq = ee.List.sequence(1, bandNames.length());
    return seq.map(function(b) {
      return ee.String(prefix).cat(ee.Number(b).int());
    });
  };
//数据平均
var meanDict = image2.reduceRegion({
    reducer: ee.Reducer.mean(),
    geometry: region,
    scale: scale,
    maxPixels: 1e9
});
var means = ee.Image.constant(meanDict.values(bandNames));
var centered = image2.subtract(means);

//主成分分析函数，GEE上有
var getPrincipalComponents = function(centered, scale, region) {
    // 图像转为一维数组
    var arrays = centered.toArray();
    // 计算协方差矩阵
    var covar = arrays.reduceRegion({
      reducer: ee.Reducer.centeredCovariance(),
      geometry: region,
      scale: scale,
      maxPixels: 1e9
    });
    // 获取“数组”协方差结果并转换为数组。波段与波段之间的协方差
    var covarArray = ee.Array(covar.get('array'));
    // 执行特征分析，并分割值和向量。
    var eigens = covarArray.eigen();
    // 特征值的P向量长度
    var eigenValues = eigens.slice(1, 0, 1);

    //计算主成分载荷
    var eigenValuesList = eigenValues.toList().flatten()
    var total = eigenValuesList.reduce(ee.Reducer.sum())
    var percentageVariance = eigenValuesList.map(function(item) {
      return (ee.Number(item).divide(total)).multiply(100).format('%.2f')
    })
    print('特征值',eigenValues )
    print("贡献率", percentageVariance)  

    // PxP矩阵，其特征向量为行。
    var eigenVectors = eigens.slice(1, 1);
    // 将图像转换为二维阵列
    var arrayImage = arrays.toArray(1);

    //使用特征向量矩阵左乘图像阵列
    var principalComponents = ee.Image(eigenVectors).matrixMultiply(arrayImage);

    // 将特征值的平方根转换为P波段图像。
    var sdImage = ee.Image(eigenValues.sqrt())
      .arrayProject([0]).arrayFlatten([getNewBandNames('sd')]);

    //将PC转换为P波段图像，通过SD标准化。
    principalComponents=principalComponents
      // 抛出一个不需要的维度，[[]]->[]。
      .arrayProject([0])
      // 使单波段阵列映像成为多波段映像，[]->image。
      .arrayFlatten([getNewBandNames('pc')])
      // 通过SDs使PC正常化。
      .divide(sdImage);
    return principalComponents
  };
//进行主成分分析，获得分析结果

var pcImage = getPrincipalComponents(centered, scale, region);
var rsei_un_unit = pcImage.expression(
  'constant -pc1',
  {
             constant: 1,
             pc1: pcImage.select('pc1')
         });
print(rsei_un_unit,'rsei_un_unit')
var rsei = img_normalize(rsei_un_unit);

//Map.addLayer(rsei, {}, 'PCA');
//Map.addLayer(lst, visParams1, "LST");
//Map.addLayer(NDBSI, {}, "NDBSI");
//Map.addLayer(ndvi, visParams, "NDVI");
//Map.addLayer(wet, visParams2, "Wet");
//print(wet)
print(rsei);
Map.addLayer(rsei, {}, "rsei");

//导出结果

Export.image.toDrive({
        image:  rsei,//分类结果
        description: 'rsei',//文件名
        folder: 'test',
        scale: 30,//分辨率
        region: roi,//区域
        maxPixels:34e10//此处值设置大一些，防止溢出
      });

二、使用2009年landsat 5影像数据计算遥感生态指数RSEI

//could masking
function maskL5sr(image) {
  // Bit 0 - Fill
  // Bit 1 - Dilated Cloud
  // Bit 2 - Unused
  // Bit 3 - Cloud
  // Bit 4 - Cloud Shadow
  var qaMask = image.select('QA_PIXEL').bitwiseAnd(parseInt('11111', 2)).eq(0);
  var saturationMask = image.select('QA_RADSAT').eq(0);

  // Apply the scaling factors to the appropriate bands.
  var opticalBands = image.select('SR_B.').multiply(0.0000275).add(-0.2);
  var thermalBand = image.select('ST_B6').multiply(0.00341802).add(149.0);

  // Replace the original bands with the scaled ones and apply the masks.
  return image.addBands(opticalBands, null, true)
      .addBands(thermalBand, null, true)
      .updateMask(qaMask)
      .updateMask(saturationMask);
}

var dataset1 = ee.ImageCollection('LANDSAT/LT05/C02/T1_L2')
    .filterDate('2009-x-x', '2009-x-x')
    .filterBounds(roi)
    .filter(ee.Filter.lte('CLOUD_COVER',5))
    .sort('CLOUD_COVER')
    .map(maskL5sr)
    //.mean()
print(dataset1)

var image1=dataset1.mosaic().clip(roi)
print(image1,'maskL5sr')
//Map.addLayer(image1,{bands:["SR_B3","SR_B2","SR_B1"],gamma:1.3,max:108,min:15},'image1');

var jrc = jrcwater09.select('b1')
var image = image1.updateMask(jrc.eq(1))
Map.addLayer(image,{bands:["SR_B3","SR_B2","SR_B1"],gamma:1.3,max:108,min:15},'Imagewc');
print(image,'water and cloud')

function SI_cal(img) {
 var blue = img.select("SR_B1");
 var red = img.select("SR_B3");
 var nir = img.select("SR_B4");
 var swir1 = img.select("SR_B5");
 var swir2 = img.select("SR_B7");
 var SI_temp =((swir1.add(red)).subtract(blue.add(nir)))
              .divide((swir1.add(red)).add(blue.add(nir)))
 //print(SI_temp);
 return SI_temp;
}
function IBI_cal(img) {
 var green = img.select("SR_B2");
 var red = img.select("SR_B3");
 var nir = img.select("SR_B4");
 var swir1 = img.select("SR_B5");
 var IBI_temp =(((swir1.multiply(2.0)).divide(swir1.add(nir)))
               .subtract((nir.divide(nir.add(red))).add(green.divide(green.add(swir1)))))
               .divide((((swir1.multiply(2.0)).divide(swir1.add(nir)))
               .add((nir.divide(nir.add(red))).add(green.divide(green.add(swir1))))))
 //print(IBI_temp);
 return IBI_temp;
}

function Wet_cal(img) {
 var blue = img.select("SR_B1");
 var green = img.select("SR_B2");
 var red = img.select("SR_B3");
 var nir = img.select("SR_B4");
 var swir1 = img.select("SR_B5");
 var swir2 = img.select("SR_B7");
 var wet_temp =blue.multiply(0.0315)
              .add(green.multiply(0.2021))
              .add(red.multiply(0.3102))
              .add(nir.multiply(0.1594))
              .add(swir1.multiply(-0.6806))
              .add(swir2.multiply(-0.6109))
 //print(wet_temp);
 return wet_temp;
}

var NDVI=ndvi2009.reproject('EPSG:4326',null,1000);  
//var img_mean=lst.mean().reproject('EPSG:4326',null,1000);

//normalization function
function img_normalize(img){
      var minMax = img.reduceRegion({
            reducer:ee.Reducer.minMax(),
            geometry: roi,
            scale: 1000,
            maxPixels: 10e13,
        })
      var year = img.get('year')
      var normalize  = ee.ImageCollection.fromImages(
            img.bandNames().map(function(name){
                  name = ee.String(name);
                  var band = img.select(name);
                  return band.unitScale(ee.Number(minMax.get(name.cat('_min'))), ee.Number(minMax.get(name.cat('_max'))));
              })
        ).toBands().rename(img.bandNames());
        return normalize;
} 

var dataset2 = ee.ImageCollection('MODIS/061/MOD11A2')
    .filterDate('2009-X-X', '2009-X-X')
    .mean()
//var dataset_no_water = dataset.updateMask(NDWI_mask).clip(table);
var dataset2_no_water = dataset2.updateMask(jrc.gte(0.1)).clip(roi);

var lst = dataset2_no_water.expression(
    'a*0.02-273.15',
    {
        a:dataset2_no_water.select('LST_Day_1km'), 
    });

var SI = SI_cal(image);
print(SI,'SI');
var IBI = IBI_cal(image);
var NDBSI =(SI.add(IBI)).divide(2.0);
var wet = Wet_cal(image);

var unit_ndvi = img_normalize(NDVI);
image=image.addBands(unit_ndvi.rename('NDVI').toFloat())
var unit_NDBSI = img_normalize(NDBSI);
image=image.addBands(unit_NDBSI.rename('NDBSI').toFloat())
var unit_Wet = img_normalize(wet);
image=image.addBands(unit_Wet.rename('Wet').toFloat())
var unit_lst = img_normalize(lst);
image=image.addBands(unit_lst.rename('LST').toFloat())
print(image)
var bands = ["Wet","NDVI","NDBSI","LST"]
var sentImage =image.select(bands)

var region = roi;
var image2=  sentImage.select(bands);
var scale = 1000;
var bandNames = image2.bandNames();
// 图像波段重命名函数
var getNewBandNames = function(prefix) {
    var seq = ee.List.sequence(1, bandNames.length());
    return seq.map(function(b) {
      return ee.String(prefix).cat(ee.Number(b).int());
    });
  };
//数据平均
var meanDict = image2.reduceRegion({
    reducer: ee.Reducer.mean(),
    geometry: region,
    scale: scale,
    maxPixels: 1e9
});
var means = ee.Image.constant(meanDict.values(bandNames));
var centered = image2.subtract(means);

//主成分分析函数
var getPrincipalComponents = function(centered, scale, region) {
    // 图像转为一维数组
    var arrays = centered.toArray();
    // 计算协方差矩阵
    var covar = arrays.reduceRegion({
      reducer: ee.Reducer.centeredCovariance(),
      geometry: region,
      scale: scale,
      maxPixels: 1e9
    });
    // 获取“数组”协方差结果并转换为数组。波段与波段之间的协方差
    var covarArray = ee.Array(covar.get('array'));
    // 执行特征分析，并分割值和向量。
    var eigens = covarArray.eigen();
    // 特征值的P向量长度
    var eigenValues = eigens.slice(1, 0, 1);

    //计算主成分载荷
    var eigenValuesList = eigenValues.toList().flatten()
    var total = eigenValuesList.reduce(ee.Reducer.sum())
    var percentageVariance = eigenValuesList.map(function(item) {
      return (ee.Number(item).divide(total)).multiply(100).format('%.2f')
    })
    print('特征值',eigenValues )
    print("贡献率", percentageVariance)  

    // PxP矩阵，其特征向量为行。
    var eigenVectors = eigens.slice(1, 1);
    // 将图像转换为二维阵列
    var arrayImage = arrays.toArray(1);

    //使用特征向量矩阵左乘图像阵列
    var principalComponents = ee.Image(eigenVectors).matrixMultiply(arrayImage);

    // 将特征值的平方根转换为P波段图像。
    var sdImage = ee.Image(eigenValues.sqrt())
      .arrayProject([0]).arrayFlatten([getNewBandNames('sd')]);

    //将PC转换为P波段图像，通过SD标准化。
    principalComponents=principalComponents
      // 抛出一个不需要的维度，[[]]->[]。
      .arrayProject([0])
      // 使单波段阵列映像成为多波段映像，[]->image。
      .arrayFlatten([getNewBandNames('pc')])
      // 通过SDs使PC正常化。
      .divide(sdImage);
    return principalComponents
  };
//进行主成分分析，获得分析结果

var pcImage = getPrincipalComponents(centered, scale, region);
var rsei_un_unit = pcImage.expression(
  'constant - pc1*0.605-pc2*0.395' , 
  {
             constant: 1,
             pc1:pcImage.select('pc1'),
         });
print(rsei_un_unit,'rsei_un_unit')
var rsei = img_normalize(rsei_un_unit);

//Map.addLayer(rsei, {}, 'PCA');
//Map.addLayer(lst, visParams1, "LST");
//Map.addLayer(NDBSI, {}, "NDBSI");
//Map.addLayer(ndvi, visParams, "NDVI");
//Map.addLayer(wet, visParams2, "Wet");
//print(wet)
print(rsei);
Map.addLayer(rsei, {}, "rsei");

Export.image.toDrive({
        image:  rsei,//分类结果
        description: 'rsei',//文件名
        folder: 'test',
        scale: 30,//分辨率
        region: roi,//区域
        maxPixels:34e10//此处值设置大一些，防止溢出
      });