Abstract Crop emergence date is a critical input to models of crop development and biomass accumulation. The ability to robustly detect and map emergence date using remote sensing would greatly benefit operational yield estimation and crop monitoring efforts; however, this has proven to be challenging. Previous remote-sensing phenology algorithms showed that crop stages can typically be detected starting only around the V3-V4 (3 to 4 established leaves) vegetative stage. Furthermore, traditional approaches have a strong assumption regarding the temporal evolution of plant growth and normally require a complete growth period of observations to define seasonal changes. Most approaches were not designed for within-season operational mapping, particularly in the early growing season. In the current paper, we describe a new within-season emergence (WISE) approach to mapping crop green-up date using satellite observations available during early growth stages. The approach was first optimized using high spatiotemporal resolution (10 m, 2-day revisit) imagery from the Vegetation and Environment monitoring New MicroSatellite (VENμS) research mission, and assessed using ground observations of early crop growth stages (emergence VE and one leaf V1 stages for corn, and emergence VE and unifoliolate VC stages for soybeans) collected over the Beltsville Agricultural Research Center (BARC) experimental fields in Beltsville, MD during the 2019 growing season. Results show that early crop growth stages can be reliably detected at sub-field scale about two weeks after crop emergence. The remote-sensing green-up dates were about 4–5 days after crop emergence on average. Coefficients of determination (R2) between green-up dates and the mid-point dates of the early growth stages were above 0.90. The mean absolute differences, standard deviations, and root mean square errors comparing to the early growth stage mid-point dates were within six days. The maximum differences were within ±10 days across all fields. The WISE approach was assessed using operational Sentinel-2 data (10 m, 5-day revisit) over BARC. The detected green-up dates from Sentinel-2 were consistent with those from VENμS. Some fields were not detected due to the lack of observations around the emergence dates. For independent evaluation, the WISE approach was applied over an agricultural watershed on the Maryland Eastern Shore using both VENμS and the Harmonized Landsat and Sentinel-2 (HLS) data (30 m, 3–4-day revisit). The detected green-up dates were compared with emergence dates in crop progress reports from the National Agricultural Statistics Service (NASS) at the state-level. The WISE-detected green-up dates at the regional scale are within VE stage ranges but slightly earlier than NASS crop progress reports at the state-level. The WISE approach uses remote-sensing observations during the early crop growth stages and has potential for operational application within the season using Sentinel-2 and HLS data.

中文翻译：

---

使用高时空分辨率图像检测玉米和大豆早期生长阶段的季节性方法

摘要 作物出苗日期是作物发育和生物量积累模型的关键输入。使用遥感可靠地检测和绘制出苗日期的能力将极大地有益于业务产量估算和作物监测工作；然而，这已被证明具有挑战性。以前的遥感物候算法表明，通常只能从 V3-V4（3 到 4 片已形成的叶子）营养阶段开始检测作物阶段。此外，传统方法对植物生长的时间演变有很强的假设，并且通常需要完整的观察生长期来定义季节性变化。大多数方法不是为季节内的操作制图而设计的，特别是在早期的生长季节。在目前的论文中，我们描述了一种新的季节内出苗 (WISE) 方法，使用早期生长阶段可用的卫星观测来绘制作物绿化日期。该方法首先使用来自植被和环境监测新微卫星 (VENμS) 研究任务的高时空分辨率（10 m，2 天重访）图像进行优化，并使用早期作物生长阶段（萌发 VE 和单叶 V1在 2019 年生长季节期间在马里兰州贝尔茨维尔的贝尔茨维尔农业研究中心 (BARC) 试验田收集的玉米生长阶段，以及大豆的出苗 VE 和单叶 VC 阶段。结果表明，在作物出苗约两周后，可以在亚田范围内可靠地检测早期作物生长阶段。遥感绿化时间平均在作物出苗后4-5天左右。绿化日期和早期生长阶段的中点日期之间的决定系数 (R2) 大于 0.90。与早期生长期中点日期相比，平均绝对差异、标准偏差和均方根误差均在六天内。所有田地的最大差异均在 ±10 天内。WISE 方法是使用 BARC 上的操作 Sentinel-2 数据（10 m，5 天重访）评估的。Sentinel-2 检测到的绿化日期与 VENμS 的一致。由于缺乏对出现日期的观察，一些田地没有被检测到。对于独立评估，使用 VENμS 和 Harmonized Landsat 和 Sentinel-2 (HLS) 数据（30 m，3-4 天重访）将 WISE 方法应用于马里兰州东海岸的一个农业流域。检测到的绿化日期与国家农业统计局 (NASS) 在州一级的作物进展报告中的出苗日期进行了比较。WISE 检测到的区域尺度的绿化日期在 VE 阶段范围内，但略早于州级 NASS 作物进展报告。WISE 方法在作物早期生长阶段使用遥感观测，并具有使用 Sentinel-2 和 HLS 数据在季节内进行业务应用的潜力。检测到的绿化日期与国家农业统计局 (NASS) 在州一级的作物进展报告中的出苗日期进行了比较。WISE 检测到的区域尺度的绿化日期在 VE 阶段范围内，但略早于州级 NASS 作物进展报告。WISE 方法在作物早期生长阶段使用遥感观测，并具有使用 Sentinel-2 和 HLS 数据在季节内进行业务应用的潜力。检测到的绿化日期与国家农业统计局 (NASS) 在州一级的作物进展报告中的出苗日期进行了比较。WISE 检测到的区域尺度的绿化日期在 VE 阶段范围内，但略早于州级 NASS 作物进展报告。WISE 方法在作物早期生长阶段使用遥感观测，并具有使用 Sentinel-2 和 HLS 数据在季节内进行业务应用的潜力。