Temporally transferable crop mapping with temporal encoding and deep learning augmentations

被引:5
|
作者
Pham, Vu-Dong [1 ,2 ]
Tetteh, Gideon [3 ]
Thiel, Fabian [1 ]
Erasmi, Stefan [3 ]
Schwieder, Marcel [3 ,4 ]
Frantz, David [5 ]
van der Linden, Sebastian [1 ,2 ]
机构
[1] Univ Greifswald, Inst Geog & Geol, Earth Observat & Geoinformat Sci Lab, Friedrich-Ludwig-Jahn-Str 16, D-17489 Greifswald, Germany
[2] Univ Greifswald, Interdisciplinary Ctr Balt Sea Reg Res IFZO, D-17489 Greifswald, Germany
[3] Thunen Inst Farm Econ, Bundesallee 63, D-38116 Braunschweig, Germany
[4] Humboldt Univ, Geog Dept, Unter Linden 6, D-10099 Berlin, Germany
[5] Trier Univ, Geoinformat Spatial Data Sci, Behringstr 21, D-54296 Trier, Germany
关键词
Annual crop mapping; Transferability; Temporal encoding; Data augmentations; 1D-CNN; Transformer; DOMAIN ADAPTATION; RANDOM FOREST; LANDSAT; CLASSIFICATION; REFLECTANCE; SENTINEL-2; ATTENTION;
D O I
10.1016/j.jag.2024.103867
中图分类号
TP7 [遥感技术];
学科分类号
081102 ; 0816 ; 081602 ; 083002 ; 1404 ;
摘要
Detailed maps on the spatial and temporal distribution of crops are key for a better understanding of agricultural practices and for food security management. Multi-temporal remote sensing data and deep learning (DL) have been extensively studied for deriving accurate crop maps. However, strategies to solve the problem of transferring crop classification models over time, e.g., training the model with data for a recent year and mapping back to the past, have not been fully explored. This is due to the lack of a generalized method for aggregating optical data with regard to the irregularity in annual clear sky observations and the scarcity of multi-annual crop reference data to support a more generalized DL model. In this study, we tackled these challenges by introducing a method namely Temporal Encoding (TE) to capture the irregular phenological information. Subsequently, we adapted and integrated two methods, i.e., Random Observations Selection (ROS) and Random Day Shifting (RDS) to simulate the variability of temporal sparsity as well as the shifts of crop phenology over different years. We tested this approach with a 1-dimensional Convolutional Neural Network (1D-CNN) and a Transformer Network models. Our results for both classifiers showed that models trained with crop reference data from 2018 and a dense time series of Landsat 7/8 and Sentinel-2 A/B data can be transferred with little decreases in accuracy to map 12 consecutive years from 2010 to 2021. The Transformer Network was slightly more accurate, while the 1D-CNN was much three times faster. Furthermore, the proposed models could achieve similar performances in the same years with and without fully available satellite information. The TE with ROS and RDS appears well suited for improving temporal transferability to support long term historic crop mapping.
引用
收藏
页数:15
相关论文
共 50 条
  • [41] Deep learning based crop-type mapping using SAR and optical data fusion
    Hamidi, Masoumeh
    Homayouni, Saeid
    Safari, Abdolreza
    Hasani, Hadiseh
    INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION, 2024, 129
  • [42] Deep Learning Application for Crop Classification via Multi-Temporal Remote Sensing Images
    Li, Qianjing
    Tian, Jia
    Tian, Qingjiu
    AGRICULTURE-BASEL, 2023, 13 (04):
  • [43] Enhancing Resilience of Deep Learning Networks By Means of Transferable Adversaries
    Seiler, Moritz
    Trautmann, Heike
    Kerschke, Pascal
    2020 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS (IJCNN), 2020,
  • [44] Transferable aircraft trajectory prediction with generative deep imitation learning
    Kravaris, Theocharis
    Vouros, George A.
    INTERNATIONAL JOURNAL OF DATA SCIENCE AND ANALYTICS, 2024,
  • [45] Versatile Deep Learning Pipeline for Transferable Chemical Data Extraction
    Alshehri, Abdulelah S.
    Horstmann, Kai A.
    You, Fengqi
    JOURNAL OF CHEMICAL INFORMATION AND MODELING, 2024, 64 (15) : 5888 - 5899
  • [46] TRANSFERABLE DEEP LEARNING FOR IN-SITU TOOL WEAR DIAGNOSIS
    Russell, Matthew
    Wang, Peng
    PROCEEDINGS OF THE 2020 INTERNATIONAL SYMPOSIUM ON FLEXIBLE AUTOMATION (ISFA2020), 2020,
  • [47] Information Optimization and Transferable State Abstractions in Deep Reinforcement Learning
    Gomez, Diego
    Quijano, Nicanor
    Giraldo, Luis Felipe
    IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2023, 45 (04) : 4782 - 4793
  • [48] Exploring Novel Word Learning via Fast Mapping and Explicit Encoding in Persons with Temporal Lobe Epilepsy
    Mohan, P. Manju
    Menon, Ramshekhar N.
    Goswami, Satyapal Puri
    Thomas, Sanjeev, V
    Cherian, Ajith
    Radhakrishnan, Ashalatha
    ANNALS OF INDIAN ACADEMY OF NEUROLOGY, 2022, 25 (06) : 1080 - 1086
  • [49] Learning Event Representations by Encoding the Temporal Context
    Dias, Catarina
    Dimiccoli, Mariella
    COMPUTER VISION - ECCV 2018 WORKSHOPS, PT III, 2019, 11131 : 587 - 596
  • [50] Encoding temporal information in deep convolution neural network
    Singh, Avinash Kumar
    Bianchi, Luigi
    FRONTIERS IN NEUROERGONOMICS, 2024, 5