This Python script takes the metadata from the USGS Landsat Bulk Metadata Service in XML format and finds coincidence dates for water quality station samples

parse_landsat_xmlMetadata.py

import datetime
import xmltodict
import numpy as np
import pandas as pd

"""
XML metata data file is take from https://landsat.usgs.gov/landsat-bulk-metadata-service
"""

# Define input files
lsmetaData = r'~/metadata_XXXX.xml'
inDateFile = r'~/wq_dataTable.csv'

statData = pd.read_csv(inDateFile)

# Get input water quality station and sampling dates
statID = np.array(statData.StatID,dtype=str)
sampleDates = np.array(statData.SDate,dtype=str)

# Get unique stations
stations = np.unique(statID)

# Create empty lists to populate
statList = ['StationID']
sDateList = ['SampleDate']
lsIdList = ['SceneID']
lDateList = ['SceneDate']

for i in range(stations.size):
    stat = stations[i]
    
    idx = np.where(statID==stat)
    
    tmpDates = sampleDates[idx]

    try:
        with open(lsmetaData.replace('XXXX',stat)) as fd:
            doc = xmltodict.parse(fd.read())
    except IOError:
        print 'Passing station {0}'.format(stat)
        continue
        
    metaData = doc['searchResponse']['metaData']
    
    for j in range(len(metaData)):
        dDate = metaData[j]['acquisitionDate'].split('-')
        
        lsDate = datetime.date(int(dDate[0]),int(dDate[1]),int(dDate[2]))
        
        lsId = str(metaData[j]['sceneID'])
        
        for k in range(tmpDates.size):
            # Update how to grab date based on input data
            d = tmpDates[k].split('/')
            
            # Make a date variable
            # Expects the date string format to be MM/dd/YYYY
            sampDate = datetime.date(int(d[2]),int(d[0]),int(d[1]))
            
            deltaT = abs(sampDate - lsDate)
            
            if deltaT.days < 2:
                statList.append(stat)
                sDateList.append(sampDate.isoformat())
                lsIdList.append(lsId)
                lDateList.append(lsDate.isoformat())
            
outfile = r'~/out_Landsat_coincidence.csv'

df = pd.DataFrame(zip(*[statList,sDateList,lsIdList,lDateList]))

df.to_csv(outfile, sep=',', encoding='utf-8')

Example format for the inDateFile CSV:

StatID	SDate
H010500	18/08/2005
H010500	10/11/2005
H010500	14/07/2007

StatID is the station unique identifier and SDate is a date that a sample was take for the station.