PratyushTripathy

client side

1. ssh your remote server: ssh username@serverip;

server side

2. source a Python environment which has the Jupyter installed;
3. execute:

   $ jupyter notebook --no-browser --port=8889 --NotebookApp.allow_origin="*" --ip=serverip --NotebookApp.token='' 

PratyushTripathy

import pandas as pd

# define the link to the CSV file on Microsoft's GitHub repo
csv_path = r'https://raw.githubusercontent.com/microsoft/GlobalMLBuildingFootprints/main/dataset-links.csv'

# read the file
df = pd.read_csv(csv_path)

# check basic details
print('Columns in CSV:', df.columns)

PratyushTripathy

// define a function to pull Landsat data for a given footprint
var band_names = ['Blue', 'Green', 'Red', 'NIR', 'SWIR1', 'Thermal', 'SWIR2'];
function get_landsat_for_aoi(aoi){
  return ee.ImageCollection("LANDSAT/LT05/C02/T1_L2").filterBounds(aoi)
    // fetching scenes for Jan-May of 2010 and 2011
    .filter(ee.Filter.calendarRange(2010, 2011,'year')).filter(ee.Filter.calendarRange(1,5,'month'))
    .select(['SR_B1', 'SR_B2', 'SR_B3', 'SR_B4', 'SR_B5', 'ST_B6', 'SR_B7'],
      band_names)
    .filter(ee.Filter.lte('CLOUD_COVER', 30))
    .reduce(ee.Reducer.percentile([25]))

PratyushTripathy

// extract the unique row numbers from the landsat footprint collection
var unique_paths = landsat_footprints.aggregate_array('path').distinct();
print('Unique paths:', unique_paths)

// seggregate the landsat footprints by path 
var path_wise_footprints = unique_paths.map(function a(element){
  return landsat_footprints.filter(ee.Filter.eq('path', element));
});
print('Path wise footprints', path_wise_footprints)

PratyushTripathy

// perform negative buffer of 80 km to the polygons
landsat_footprints = landsat_footprints.map(function (x){return x.buffer(-8e4, 5)})

Map.addLayer(landsat_footprints, {}, 'landsat_footprints')

PratyushTripathy

// pull landsat footprint and filter for your area of interest
var landsat_footprints = ee.FeatureCollection("users/pratyusht/public_assets/landsat_footprints")
.filter(ee.Filter.and(
  ee.Filter.lte('path', 148), ee.Filter.gte('path', 140),
  ee.Filter.lte('row', 46), ee.Filter.gte('row', 42)
))

Map.addLayer(landsat_footprints, {}, 'landsat_footprints')
Map.centerObject(landsat_footprints, 6)

PratyushTripathy

// call the function to match the histogram of slave
var hist = require('users/pratyusht/PublicFunctions:seamless_mosaic/histogram');

var matched_slave = hist.match_histogram(master, slave, overlap, band_names, 30);

Map.addLayer(matched_slave, l8_vis_params, 'Matched Slave');

PratyushTripathy

// Get footprint of the landsat scenes
var master_fp = master.geometry();
var slave_fp = slave.geometry();

// Perform intersection to extarct overlap area
var overlap = master_fp.intersection(slave_fp);

// Add all the three layers to the map
Map.addLayer(master_fp, {}, 'Master Footprint');
Map.addLayer(slave_fp, {}, 'Slave Footprint');

PratyushTripathy

// query landsat 8 collection for given AOI
var band_names = ['Blue', 'Green', 'Red', 'NIR', 'SWIR1', 'Thermal', 'SWIR2'];
var landsat8 = ee.ImageCollection("LANDSAT/LC08/C02/T1_L2")
                      .filterBounds(geometry)
                      .filterDate('2020-03-15', '2020-03-25')
                      .filter(ee.Filter.lte('CLOUD_COVER', 5))
                      .select(
                        ['SR_B2', 'SR_B3', 'SR_B4', 'SR_B5', 'SR_B6', 'ST_B10', 'SR_B7'],
                        band_names
                        );

PratyushTripathy

# define the output file names
mx_outfile = mx_file.replace('.tif', '_byte_deflate.tif')
sb_outfile = sb_file.replace('.tif', '_byte_deflate.tif')

# export the files
raster.export(arr_mx, ds_mx, filename=mx_outfile, dtype='uint8', compress='DEFLATE')
raster.export(arr_sb, ds_sb, filename=sb_outfile, dtype='uint8', compress='DEFLATE')