Kossuth Iowa Corn Yield USDA NASS 1990-2022
收藏DataCite Commons2026-05-07 更新2026-05-07 收录
下载链接:
https://zenodo.org/doi/10.5281/zenodo.20061514
下载链接
链接失效反馈官方服务:
资源简介:
Corn Belt county yield prediction dataset: crop progress, condition, and detrended target for a single high-production county in the US Corn Belt.
Kossuth County, Iowa is one of the top corn-producing counties in the United States. In 2022 it produced ~62.8 million bushels (~1.59 million metric tons) — approximately 0.46% of total US corn production, equivalent to ~41.8% of total Polish corn production. This makes it an excellent single-location benchmark for crop-yield ML.
The dataset combines three USDA NASS data streams (1990–2022):
Annual corn grain yield at county level (Kossuth, Iowa)
Weekly Crop Progress reports at state level (Iowa)
Weekly Crop Condition reports at state level (Iowa)
Raw API data is provided alongside a processed, model-ready dataset (n=33 years × 15 columns) with engineered features: phenology timing (week of 50% planted/silking/mature/harvested), condition aggregates (Good+Excellent and Poor+VeryPoor across full season, silking window, and Jul–Aug grain fill period), linear technology trend, and detrended yield target.
Recommended ML target: yield_detrended = yield − linear trend. Isolates weather/environmental signal from long-term technology gains (~+2 BU/ACRE/year reflecting hybrid and agronomic improvements).
Suggested tasks: K-NN regression with Leave-One-Out CV (small n=33), correlation analysis between condition index and yield, phenology trend analysis, anomaly detection (1993 flood: −61 BU; 2012 drought: −17 BU).
Source: USDA NASS QuickStats API (public domain). Compiled at: FarmerPlanner / Demeter AgTech for educational and benchmarking use.
This dataset is also mirrored on Kaggle: kossuth-corn-yield-iowa.
美国玉米带县域产量预测数据集:涵盖美国玉米带单个高产能县域的作物生长进度、健康状况及去趋势化目标产量数据。
艾奥瓦州科苏特县(Kossuth County)是美国玉米产量顶尖县域之一,2022年玉米总产量约6280万蒲式耳(约159万公吨),占美国玉米总产量的0.46%,相当于波兰全国玉米总产量的41.8%左右,是用于作物产量机器学习建模的优质单站点基准数据集。
本数据集整合了1990年至2022年的三类美国农业部国家农业统计服务中心(USDA NASS)数据:
1. 艾奥瓦州科苏特县年度玉米籽粒产量数据
2. 艾奥瓦州级每周作物生长进度报告
3. 艾奥瓦州级每周作物健康状况报告
除原始API接口数据外,本数据集还提供经预处理后可直接用于模型训练的标准化数据集(样本量n=33年 × 15列),包含经特征工程生成的特征:物候时序特征(50%植株完成播种、抽丝、成熟、收获的对应周数)、健康状况聚合特征(覆盖全生育期、抽丝期、7-8月籽粒灌浆期的优良级与极差级占比总和)、线性技术进步趋势项,以及去趋势化后的产量目标变量。
推荐机器学习建模目标为:去趋势化产量(yield_detrended)= 原始产量 - 线性技术趋势项。该目标可分离出长期技术进步(约每年每英亩+2蒲式耳,对应杂交品种与农艺技术改进)之外的气象与环境信号。
建议开展的建模与分析任务包括:采用留一法交叉验证(Leave-One-Out CV)的K近邻(K-NN)回归(样本量较小n=33)、作物健康指数与产量的相关性分析、物候时序趋势分析、异常值检测(如1993年洪灾导致单产减少61蒲式耳/英亩,2012年旱灾导致单产减少17蒲式耳/英亩)。
数据来源:美国农业部国家农业统计服务中心快速统计(USDA NASS QuickStats)API(公有领域)。数据集由FarmerPlanner / Demeter AgTech整理,仅用于教育与基准测试用途。
本数据集同时镜像至Kaggle平台,数据集名称为:kossuth-corn-yield-iowa。
提供机构:
FarmerPlanner
创建时间:
2026-05-07



