Early Identification of Seed Maize and Common Maize Production Fields Using Sentinel-2 Images

Abstract

<jats:p>Accurate and timely access to the production area of crop seeds allows the seed market and secure seed supply to be monitored. Seed maize and common maize production fields typically share similar phenological development profiles with differences in the planting patterns, which makes it challenging to separate these fields from decametric-resolution satellite images. In this research, we proposed a method to identify seed maize production fields as early as possible in the growing season using a time series of remote sensing images in the Liangzhou district of Gansu province, China. We collected Sentinel-2 and GaoFen-1 (GF-1) images captured from March to September. The feature space for classification consists of four original bands, namely red, green, blue, and near-infrared (nir), and eight vegetation indexes. We analyzed the timeliness of seed maize identification using Sentinel-2 time series of different time spans and identified the earliest time frame for reasonable classification accuracy. Then, the earliest time series that met the requirements of regulatory accuracy were compared and analyzed. Four machine/deep learning algorithms were tested, including K-nearest neighbor (KNN), support vector classification (SVC), random forest (RF), and long short-term memory (LSTM). The results showed that using Sentinel-2 images from March to June, the RF and LSTM algorithms achieve over 88% accuracy, with the LSTM performing the best (90%). In contrast, the accuracy of KNN and SVC was between 82% and 86%. At the end of June, seed maize mapping can be carried out in the experimental area, and the precision can meet the basic requirements of monitoring for the seed industry. The classification using GF-1 images were less accurate and reliable; the accuracy was 85% using images from March to June. To achieve near real-time identification of seed maize fields early in the growing season, we adopted an automated sample generation approach for the current season using only historical samples based on clustering analysis. The classification accuracy using new samples extracted from historical mapping reached 74% by the end of the season (September) and 63% by the end of July. This research provides important insights into the classification of crop fields cultivated with the same crop but different planting patterns using remote sensing images. The approach proposed by this study enables near-real time identification of seed maize production fields within the growing season, which could effectively support large-scale monitoring of the seed supply industry.</jats:p>

Deep Learning for Land Cover Change Detection

“…Qiu et al [25,26] combine a residual convolutional neural network (ResNet) and an RNN for urban land cover classification. LSTM approaches, an extension of RNNs, are also applied in land cover classification [12,27], crop type classification [11,28], and crop area estimation [29]. Rußwurm and Körner [11] rely on an LSTM network with Sentinel-2 data and a GT, including a large number of crop classes.…”

“…The LSTM application of van Duynhoven and Dragićević [27] demonstrates good classification performance even with few available satellite images. The LSTM approach of Ren et al [28] achieves about 90% overall accuracy in a seed maize identification with Sentinel-2 and GaoFen-1 data. The LSTM network outperforms approaches, such as Random Forest.…”

“…Since we present the first land cover classification and change detection study on this GT in Klingenberg, Germany, no other published results are available for a direct comparison. To put our results in perspective, we discuss the results of Rußwurm and Körner [12], Ren et al [28] in the following. The sequential recurrent encoders approach of Rußwurm and Körner [12] achieves an overall accuracy of 90% on 17 crop type classes.…”

“…The size of the training dataset is more than 18 times larger than the presented dataset, which implies that our approach will benefit from an increased training dataset. In Ren et al [28], the presented LSTM approach achieves 90% in a pixel-based classification with three classes. The LSTM only slightly outperforms approaches, such as Random Forest and support vector machines, which implies that the value of using sequential data is marginal in this dataset.…”

“…The LSTM only slightly outperforms approaches, such as Random Forest and support vector machines, which implies that the value of using sequential data is marginal in this dataset. For our presented and the two compared studies [12,28], it would be interesting to evaluate the exact quality of the GT to estimate the maximum achievable overall accuracy.…”