Contents lists available at ScienceDirect ISPRS Journal of Photogrammetry and Remote Sensing journal homepage: www.elsevier.com/locate/isprsjprs Automatic canola mapping using time series of sentinel 2 images Davoud Ashourloo a, ⁎ , Hamid Salehi Shahrabi b , Mohsen Azadbakht a , Hossein Aghighi a , Hamed Nematollahi b , Abbas Alimohammadi c ,AliAkbarMatkan a a Remote Sensing and GIS Research Center, Faculty of Earth Sciences, Shahid Beheshti University, Tehran 653641255, Iran b Applied Remote Sensing Department, Iranian Space Research Center, Tehran, Iran c GIS Engineering Department, Faculty of Geodesy and Geomatic Engineering, K.N. Toosi University of Technology, Tehran 19967-15433, Iran ARTICLEINFO Keywords: Precision agriculture Canola Flowering date Automatic crop mapping Spectral index Sentinel-2 time-series ABSTRACT Differenttechniquesutilizedformappingvariouscropsaremainlybasedonusingtrainingdataset.But,dueto difficulties of access to a well-represented training data, development of automatic methods for detection of crops is an important need which has not been considered as it deserves. Therefore, main objective of present study was to propose a new automatic method for canola (Brassica napus L.) mapping based on Sentinel 2 satellitetimeseriesdata.TimeseriesdataofthreestudysitesinIran(Moghan,Gorgan,Qazvin)andonesitein USA: (Oklahoma), were used. Then, spectral reflectance values of canola in various spectral bands were com- pared with those of the other crops during the growing season. NDVI, Red and Green spectral bands were successfully applied for automatic identification of canola flowering date using the threshold values. Examination of the fisher function indicated that multiplication of the near-infrared (NIR) band by the sum of redandgreenbandsduringthefloweringdateisanefficientindextodifferentiatecanolafromtheothercrops. TheKappaandoverallaccuracy(OA)forthefourstudysitesweremorethan0.75and88%,respectively.Results of this research demonstrated the potential of the proposed approach for canola mapping using time series of remotely sensed data. 1. Introduction Food security is an important issue for growing world population (Ingladaetal.,2015;Mattonetal.,2015).Detailedspatialandtemporal informationofagriculturalfieldsisanurgentrequirementformanagers and planners to guarantee food security. Hence, operational agri- cultural monitoring systems using remote sensing data have been de- veloped in several countries (Thenkabail and Wu, 2012). Remote sen- sing data have been frequently adopted for various applications in agriculture such as the yield prediction (Becker-Reshef et al., 2010; Cheng et al., 2016; Johnson, 2016; Peralta et al., 2016), biochemical and biophysical parameter estimation (Nguy-Robertson et al., 2014) andstudyingcarbonandwatercyclemechanisms(Zhongetal.,2016). While all these applications are based on the availability of local field maps (Sulik and Long, 2016); these maps are not available in most agricultural regions and crop map should be precedently provided. Moreover, in areas where different crops are cultivated, local maps usually vary from year to year. Therefore, in-season crop mapping is essential,andlackofsuchmapscanadverselyaffectmostoftherelated applicationsandstudies(Becker-Reshefetal.,2010;BoltonandFriedl, 2013;SulikandLong,2016). Remotesensingtechniquesprovidecropmapsatlocalandregional scales, due to the broad coverage and regular acquisition of remotely sensed imageries (Forkuor et al., 2015). These techniques are very di- versewithvaryingcomplexitiesrangingfromclustering(Arangoetal., 2016),decisiontrees(Kontgisetal.,2015;PalandMather,2003;Peña- Barragánetal.,2011),object-oriented(Castillejo-Gonzálezetal.,2009; Conradetal.,2010;Lietal.,2016;Peña-Barragánetal.,2011;Vaudour et al., 2015; Vieira et al., 2012; Watts et al., 2009), knowledge-based (CohenandShoshany,2002;Lucasetal.,2007;Villaetal.,2015;Wang et al., 2017) to phenology-based (Bargiel, 2017; Dong et al., 2016; Foerster et al., 2012; Massey et al., 2017; Zhong et al., 2014; Zhong etal.,2011;Zhongetal.,2016)andmachinelearningmethods(Inglada etal.,2015;Löwetal.,2013;Valeroetal.,2016;Vaudouretal.,2015). Synergyofthesetechniqueshavealsobeenconsideredandreportedin some studies (Caietal.,2018;Qiuetal.,2017a). Most of the previous techniques for production of crop maps are mainly based on the availability of the well represented training data samples(Foersteretal.,2012).Howeverprovisionofsuchtrainingdata is usually time-consuming, costly and labor-intensive (Zhong et al., https://doi.org/10.1016/j.isprsjprs.2019.08.007 Received10January2019;Receivedinrevisedform30July2019;Accepted5August2019 ⁎ Corresponding author. E-mail address: d_ashourloo@sbu.ac.ir (D. Ashourloo). ISPRS Journal of Photogrammetry and Remote Sensing 156 (2019) 63–76 0924-2716/ © 2019 International Society for Photogrammetry and Remote Sensing, Inc. (ISPRS). Published by Elsevier B.V. All rights reserved. T