Crop Mapping Based on Historical Samples and New Training Samples Generation in Heilongjiang Province, China

Abstract

Accurate, year-by-year crop distribution information is a key element in agricultural production regulation and global change governance. However, due to the high sampling costs and insufficient use of historical samples, a supervised classifying method for sampling every year is unsustainable for mapping crop types over time. Therefore, this paper proposes a method for the generation and screening of new samples for 2018 based on historical crop samples, and then it builds a crop mapping model for that current season. Pixels with the same crop type in the historical year (2013-2017) were extracted as potential samples, and their spectral features and spatial information in the current year (2018) were used to generate new samples based on clustering screening. The research result shows that when the clustering number is different, the number and structure of new generated sample also changes. The sample structure generated in Luobei County was not balanced, with the 'other crop' representing less than 3.97%, but the structure of southwest Hulin City was more balanced. Based on the newly generated samples and the ground reference data of classified year, the classification models were constructed. The average classification accuracies of Luobei County in 2018 based on new generated samples and field samples were 69.35% and 77.59%, respectively, while those of southwest Hulin City were 80.44% and 82.94%, respectively. Combined with historical samples and the spectral information of the current year, this study proposes a method to generate new samples. It can overcome the problem of crop samples only being collected in the field due to the difficulty of visual interpretation, effectively improve the use of historical data, and also provide a new idea for sustainable crop mapping in many regions lacking seasonal field samples.Sustainability 2019, 11, 5052 2 of 17 samples. In addition, in existing crop classification studies, seasonal samples are often only used for crop classification in the current season, and they are rarely used for the next season or even the next few seasons, resulting in a low utilization rate of historical data [1,6,8]. The difficulty of sampling and the low utilization of historical data have led to the high cost of year-by-year mapping. At the same time, after collecting data every year, it is necessary to carry out complex operations such as data preprocessing, and the final crop map is often obtained after the harvest, which leads to a lag of crop mapping and its guiding role for farming being greatly weakened. Therefore, it is difficult to meet the demand for crop mapping by sampling every year. The question of how to sustainably carry out crop classification based on historical data has become a research hotspot.There are three main methods for classifying crops based on historical data: One is to reuse the spectral curve, the second is to reuse the training model, and the third is to generate "training samples."Reusing spectral curves is a method for classification based on the fact t...

A Cloud Detection Approach Based on Hybrid Multispectral Features with Dynamic Thresholds for GF-1 Remote Sensing Images

“…The main contributions of this work as follows: 1. we explored an effective strategy based on previous researchers' works without any other auxiliary data to detect clouds of the GF-1 images which are not suitable for some popular algorithms. 2. the proposed method has been applied into scientific production and supports other researches in the preprocessing of remote sensing images [33] and crop mapping [38,39].…”

Identification of Seed Maize Fields With High Spatial Resolution and Multiple Spectral Remote Sensing Using Random Forest Classifier

“…The VI quantifies the vegetation properties by helping transform the reflectance of two or more spectral bands [50]. Considering the differences in phenology, seasonal differences, as well as the significance and anti-saturation degree of different VIs, the commonly used VIs can be divided into four categories: 1 [8]. The formula is in Table 2 as follows: In the above formula, B, G, R, NIR are the reflectance of blue, green, red and near-infrared bands, respectively.…”

“…Random forest (RF) is an integrated algorithm, which belongs to the Bagging type. By combining multiple weak classifiers, the final result is obtained by voting, which gives the overall model result a higher precision and generalization ability [8].…”

A High-Temperature Risk Assessment Model for Maize Based on MODIS LST

“…Maize is a thermophilic crop, but high temperatures have an adverse impact on its growth and development [1][2][3][4][5]. Climate change has aggravated the risk of high temperatures in temperate regions.…”

“…Finally, due to the large spatial and temporal scale of this study, there is a data imbalance problem. In order to minimize this imbalance, we propose a pixel-by-pixel calculation of the effective value in the region, as shown in Eq (5)…”