利用多传感器Landsat数据和深度学习模型，首次完成了1985年至2023年中国东北地区水稻的年度分布图绘制。采用年度结果增强法（ARE），该方法能处理深度学习模型在不同阶段产生的类别概率差异。这种方法有助于缓解大规模跨传感器水稻制图时训练样本有限的影响。与传统水稻制图方法相比，ARE法获得的结果F1分数提高了6%。", "ds_source": "
数据来源于https://doi.org/10.6084/m9.figshare.27604839.v1 。", "ds_process_way": "
采用全分辨率网络（FR-Net）深度学习模型对多传感器Landsat影像进行初步分类，并创新性地提出年度结果增强（ARE）方法，通过融合水稻不同关键物候期的类别概率差异，有效解决了跨传感器、长时序条件下训练样本不足的难题，最终将制图结果的F1分数提升了6%，总体精度达到0.93。", "ds_quality": "
本数据集经10万余个独立地面样本严格验证，采用年度结果增强（ARE）方法后，总体F1分数达0.93，较传统方法提升6%。具体评价指标为：用户精度0.92、制图精度0.95、马修斯相关系数0.81。这些量化指标共同证明了数据集在识别准确性和空间一致性方面具有高可靠性，能够准确反映近40年东北水稻田的时空动态变化。", "ds_acq_start_time": "1985-01-01 00:00:00", "ds_acq_end_time": "2023-12-31 00:00:00", "ds_acq_place": "中国东北地区", "ds_acq_lon_east": null, "ds_acq_lat_south": null, "ds_acq_lon_west": null, "ds_acq_lat_north": null, "ds_acq_alt_low": null, "ds_acq_alt_high": null, "ds_share_type": "login-access", "ds_total_size": 2160750584, "ds_files_count": 39, "ds_format": "*.tif", "ds_space_res": "", "ds_time_res": "", "ds_coordinate": "无", "ds_projection": "", "ds_thumbnail": "17367d4c-d431-4b2e-953f-1305f7fd6cf7.png", "ds_thumb_from": 2, "ds_ref_way": "", "paper_ref_way": "", "ds_ref_instruction": "", "ds_from_station": null, "organization_id": "a4dd5849-78f2-44c5-b0f1-3450e952b2a2", "ds_serv_man": "敏玉芳", "ds_serv_phone": "0931-4967596", "ds_serv_mail": "ncdc@lzb.ac.cn", "doi_value": "", "subject_codes": [ "170.45" ], "quality_level": 3, "publish_time": "2025-11-27 15:50:59", "last_updated": "2026-01-12 11:37:40", "protected": false, "protected_to": null, "lang": "zh", "cstr": "11738.11.NCDC.FIGSHARE.DB7020.2025", "i18n": { "en": { "title": "Long history paddy rice mapping across Northeast China with deep learning and annual result enhancement method", "ds_format": "", "ds_source": "
The data is sourced from https://doi.org/10.6084/m9.figshare.27604839.v1 .", "ds_quality": "
This dataset has been rigorously validated with over 100000 independent ground samples, and after using the Annual Results Augmentation (ARE) method, the overall F1 score reached 0.93, an increase of 6% compared to traditional methods. The specific evaluation indicators are: user accuracy 0.92, mapping accuracy 0.95, and Matthews correlation coefficient 0.81. These quantitative indicators collectively demonstrate that the dataset has high reliability in recognition accuracy and spatial consistency, and can accurately reflect the spatiotemporal dynamic changes of rice fields in Northeast China over the past 40 years.", "ds_ref_way": "", "ds_abstract": "
The annual distribution map of rice in Northeast China from 1985 to 2023 was completed for the first time using multi-sensor Landsat data and deep learning models. Adopting the Annual Results Enhancement (ARE) method, which can handle the category probability differences generated by deep learning models at different stages. This method helps alleviate the impact of limited training samples during large-scale cross sensor rice mapping. Compared with traditional rice mapping methods, the ARE method achieved a 6% increase in F1 score.
", "ds_time_res": "", "ds_acq_place": "Northeast China", "ds_space_res": "", "ds_projection": "", "ds_process_way": "The full resolution network (FR Net) deep learning model was used to preliminarily classify multi-sensor Landsat images, and an innovative annual result enhancement (ARE) method was proposed. By integrating the probability differences of different key phenological stages of rice, the problem of insufficient training samples under cross sensor and long-term conditions was effectively solved. Finally, the F1 score of the mapping results was improved by 6%, and the overall accuracy reached 0.93.", "ds_ref_instruction": "" } }, "submit_center_id": "ncdc", "data_level": 0, "license_type": "CC BY 4.0", "ds_topic_tags": [ "深度学习", "中国东北", "水稻" ], "ds_subject_tags": [ "地理学" ], "ds_class_tags": [], "ds_locus_tags": [ "中国东北地区" ], "ds_time_tags": [ 1985, 1986, 1987, 1988, 1989, 1990, 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019, 2020, 2021, 2022, 2023 ], "ds_contributors": [ { "true_name": "杨鹏", "email": "yangpeng@caas.cn", "work_for": "中国农业科学院农业资源与农业区划研究所", "country": "中国" } ], "ds_meta_authors": [ { "true_name": "杨鹏", "email": "yangpeng@caas.cn", "work_for": "中国农业科学院农业资源与农业区划研究所", "country": "中国" } ], "ds_managers": [ { "true_name": "杨鹏", "email": "yangpeng@caas.cn", "work_for": "中国农业科学院农业资源与农业区划研究所", "country": "中国" } ], "category": "水文" }