The RapeseedMap10 database: annual maps of rapeseed at a spatial resolution of 10 m based on multi-source data
- State Key Laboratory of Earth Surface Processes and Resource Ecology/MoE Key Laboratory of Environmental Change and Natural Hazards, Faculty of Geographical Science, Beijing Normal University, Beijing 100875, China
Correspondence: Zhao Zhang (email@example.com)
Large-scale, high-resolution maps of rapeseed (Brassica napus L.), a major oilseed crop, are critical for predicting annual production and ensuring global energy security, but such maps are still not freely available for many areas. In this study, we developed a new pixel- and phenology-based algorithm and produced a new data product for rapeseed planting areas (2017–2019) in 33 countries at 10 m spatial resolution based on multiple data. Our product is strongly consistent at the national level with official statistics of the Food and Agricultural Organization of the United Nations. Our rapeseed maps achieved F1 spatial consistency scores of at least 0.81 when compared with the Cropland Data Layer in the United States, the Annual Crop Inventory in Canada, the Crop Map of England, and the Land Cover Map of France. Moreover, F1 scores based on independent validation samples ranged from 0.84 to 0.91, implying a good consistency with ground truth. In almost all countries covered in this study, the rapeseed crop rotation interval was at least 2 years. Our derived maps suggest, with reasonable accuracy, the robustness of the algorithm in identifying rapeseed over large regions with various climates and landscapes. Scientists and local growers can use the freely downloadable derived rapeseed planting areas to help predict rapeseed production and optimize planting structures. The product is publicly available at https://doi.org/10.17632/ydf3m7pd4j.3 (Han et al., 2021).
Although fossil fuels are currently the main source of energy (Fang et al., 2016; Shafiee and Topal, 2009), their overexploitation is increasing various threats to human survival, such as greenhouse gas emission and environmental pollution (Fang et al., 2016; Höök and Tang, 2013). Biofuel energy seems to be a promising alternative energy source (Hassan and Kalam, 2013). Rapeseed (Brassica napus L.) is an important source of biofuels, edible oil, animal feed, and plant protein powder (Firrisa et al., 2014; Malça and Freire, 2009; Sulik and Long, 2016). Data products on rapeseed planting densities, growth conditions, and productivity are dependent on precise and accurate planting area maps (Zhang et al., 2019), but such maps are still unavailable.
Global agricultural statistics on rapeseed in many regions are derived from field surveys, field sampling, and producer reports (Arata et al., 2020; Fuglie, 2010). Ground-based methods, which are time-consuming and labor-intensive, fail to provide detailed spatial information on rapeseed fields (J. Wang et al., 2020). In contrast, remote sensing technology plays an important role in agricultural monitoring and yields accurate, objective spatial–temporal crop information (Dong et al., 2016; Salmon et al., 2015).
Many current land cover products obtained by remote sensing have a publicly available cropland layer. Examples include the Fine Resolution Observation and Monitoring of Global Land Cover project (Gong et al., 2013), the Global Land Cover 2000 (GLC2000) map (Bartholomé and Belward, 2005), ChinaCropPhen1km (Luo et al., 2020), and Global Food Security-support data at 30 m (GFSAD30) (Phalke et al., 2020; Xiong et al., 2017). Nevertheless, cropland identified by these products is either undifferentiated as to crop type, has a coarse spatiotemporal resolution (Teluguntla et al., 2018), or does not include rapeseed. Few large-scale rapeseed maps, especially at 10 m resolution, are currently available. A decision tree classification method based on a large number of training samples has been used to classify various crops for the 30 m resolution Cropland Data Layer (Boryan et al., 2011) in the USA and the Annual Crop Inventory in Canada (Fisette et al., 2013), but this method is hard to apply to other developing regions because ground training samples are lacking (Xiong et al., 2017). A new method to map large-scale annual maps with a high spatial resolution that would be widely applicable to regions with few ground training samples is thus strongly needed.
Five remote-sensing-based methods for rapeseed mapping have been developed in recent decades: (1) machine learning methods (Griffiths et al., 2019; Preidl et al., 2020; She et al., 2015; Tao et al., 2020), (2) classification based on time series data (Ashourloo et al., 2019), (3) threshold segmentation based on phenology (Tian et al., 2019), (4) multi-range spectral feature fitting (Pan et al., 2013), and (5) mapping based on HSV (hue, saturation, and value) transformation and spectral features (Wang et al., 2018). Most of these methods, however, only produce rapeseed maps for a small area using very limited imageries taken on rapeseed peak flowering dates (Ashourloo et al., 2019; She et al., 2015). Rapeseed peak flowering dates vary by area and cultivar because of differences in natural conditions and cultivation habits, especially over large regions (d'Andrimont et al., 2020; Ashourloo et al., 2019; McNairn et al., 2018). Using the above methods to automatically map rapeseed areas with a finer resolution over a large region is thus still a huge challenge.
Taking into consideration the unique phenological characteristics of different crops, various researchers have developed potentially useful phenology-based methods for crop identification over large areas (Ashourloo et al., 2019; Dong et al., 2016; J. Liu et al., 2018; Zhang et al., 2020). These algorithms, which generate classification rules by analyzing the unique characteristics of the studied crop, have been successfully applied for mapping rice (Dong et al., 2016), soybean (Zhong et al., 2014), corn (Zhong et al., 2016), and sugarcane (J. Wang et al., 2020) but have rarely been applied to rapeseed. Rapeseed has unique reflectance and scattering characteristics (Ashourloo et al., 2019; McNairn et al., 2018; Sulik and Long, 2015, 2016) and has three canopy morphologies based on leaves, yellow petals, and pods/branches (Ashourloo et al., 2019; Rondanini et al., 2014). Each canopy shape strongly influences how solar radiation is intercepted (Sulik and Long, 2016). The specific features of reflectance values and scattering coefficients of rapeseed from S-1/2 data can thus provide information for the automatic mapping of rapeseed over larger areas and with a finer resolution.
Another relevant aspect of rapeseed imaging concerns crop rotation, which is beneficial for pest and disease management in crop production (Harker et al., 2015; J. Liu et al., 2018) and a major factor in rapeseed yield (Harker et al., 2015; Ren et al., 2015). The physical and chemical properties of soil are altered during crop rotation, and these changes affect rapeseed growth (Ren et al., 2015). Most current studies have been limited to field observations (Peng et al., 2015). The spatial distribution of rapeseed rotation in different regions is still unclear because high-resolution rapeseed maps are lacking. To aid cultivation and management, the characteristics of rapeseed rotation need to be explored.
Taking into account the above-mentioned issues, we integrated multi-source data to develop a new method for identifying rapeseed. We then applied our new method to generate rapeseed maps with a spatial resolution of 10 m across the main planting areas of 33 countries from 2017 to 2019 and analyzed the geographical characteristics of rapeseed cultivation and crop rotation.
2.1 Study area
We identified rapeseed planting areas in 33 countries, the world's main rapeseed producers, on three continents: North America, South America, and Europe (Fig. 1). The largest areas of rapeseed cultivation are located in Canada and the European Union (Carré and Pouzet, 2014; van Duren et al., 2015; Rondanini et al., 2012). In 2008, 79 % of biodiesel feedstock crops in Europe, which produces a large amount of biodiesel for export every year, were rapeseed (van Duren et al., 2015). In Chile, the main rapeseed producer in South America, the yield of rapeseed in 2018 was 3887.7 kg ha−1. Rapeseed cultivation in these countries is important for food and energy security (Carré and Pouzet, 2014). The climates of the three continents are different because of factors such as latitude and topography (Peel et al., 2007). The rapeseed planting season varies among countries because of these differences in climate and other natural conditions (Singha et al., 2019; Wang et al., 2018), thus posing great challenges to the mapping of rapeseed.
2.2.1 Remote sensing data
We collected imagery from Sentinel-2 (S2) and Sentinel-1 (S1) satellites (Table 1) launched by the European Space Agency (ESA) (Drusch et al., 2012; Torres et al., 2012). We used red (b4), green (b3), and blue (b2) spectral bands with 10 m spatial resolution top-of-atmosphere (TOA) reflectance observations. The S2 TOA product includes the quality assessment (QA) band, which was used to remove most of the poor-quality images (e.g., cloud-obscured information) in this study. Removal of all such information was difficult, however, because of the quality of the QA band (J. Wang et al., 2020; Zhu et al., 2015). We used the interferometric wide-swath mode of S1, which provides dual-band cross-polarization (VV) and vertical transmit–horizontal receive (VH) with a 12 or 6 d repeat cycle and 10 m spatial resolution (Torres et al., 2012). The S-1/2 images were obtained using the Google Earth Engine (GEE) (Gorelick et al., 2017). In addition, we used QA bands to remove most of the poor-quality images on GEE. (Sample code can be found at https://code.earthengine.google.com/?scriptPath=Examples:Datasets/COPERNICUS_S2, last access: 8 June 2021) Further details are provided in Table 1.
2.2.2 Digital elevation model
We used a spatial resolution of 1 arcsec (approximately 30 m) elevation data from the Space Shuttle Radar Terrain Mission (Table 1) (Farr et al., 2007). We then calculated the spatial distribution of slope using GEE (sample code at https://code.earthengine.google.com/?scriptPath=Examples:Datasets/USGS_SRTMGL1_003, last access: 8 June 2021). Finally, we extracted areas with a slope of less than 10∘ to mask hilly terrain (Jarasiunas, 2016).
2.2.3 Cropland and agricultural statistical data
In this study, cropland data from the GFSAD30 were used to identify major farming areas in different countries (Phalke et al., 2020; Xiong et al., 2017). Existing crop data products containing rapeseed information comprise four datasets: (1) the 30 m Annual Crop Inventory (ACI) in Canada (Fisette et al., 2013) and (2) the 30 m Cropland Data Layer (CDL) in the USA (Boryan et al., 2011), both of which were downloaded from GEE; (3) the Crop Map of England (CROME) generated in GBR; and (4) the 10 m Land Cover Map of France (LCMF) in France (Inglada et al., 2017). These four crop layer products were generated from satellite images and a large number of training sample collections. In this study, rapeseed maps in ACI, CDL, CROME, and LCMF were used for accuracy verification at the pixel level. For accuracy verification, we selected statistics on major crop areas in different countries and regions released annually by the Food and Agricultural Organization of the United Nations (FAO). Details are provided in Table 1.
2.2.4 Crop calendars
We used two crop phenological datasets to assist in the extraction of rapeseed phenological parameters: crop calendars in different countries (https://ipad.fas.usda.gov/ogamaps/cropcalendar.aspx, last access: 8 June 2021) and field records of crop phenology in Germany. The crop calendars originated from the United States Department of Agriculture, which only records rapeseed planting and harvest times for selected countries. The crop phenology field records in Germany were in situ observations from crop phenological records shared by the Deutscher Wetterdienst (DWD) in Germany (Kaspar et al., 2014). The DWD provides field observations of crop phenological periods throughout Germany following the Biologische Bundesanstalt, Bundessortenamt, and Chemical Industry (BBCH) scale (Table 1). DWD records include the start and end dates of rapeseed flowering (d'Andrimont et al., 2020; Kaspar et al., 2014). Neither the two crop calendars nor the DWD records contain information on rapeseed peak flowering dates. To extract rapeseed phenological parameters, we used all stations that fully recorded start and end dates of the flowering period from 2017 to 2019, namely, 281, 269, and 253 stations in 2017, 2018, and 2019, respectively.
2.3.1 Optical and SAR characteristics during the rapeseed growing period
We selected available rapeseed parcels and in situ observations of the DWD from different climate regions and different years to analyze the optical (reflectance and vegetation index) and SAR (VV and VH) characteristics of rapeseed over time. As an example, Fig. 2 shows the time series of one rapeseed parcel around a DWD station (station ID: 13126) in 2018. This rapeseed parcel exhibited unique visual characteristics during the flowering period (Fig. 2e). When rapeseed approached peak flowering, the flowers became yellow (d'Andrimont et al., 2020; Pan et al., 2013; Tao et al., 2020; Wang et al., 2018). Rapeseed was yellow-green on the true-color images of S2 and Google Earth during the flowering period (Fig. S1). The reflectance of the green and red bands separately increased – from 0.09 and 0.06, respectively, before flowering (17 April 2018) to 0.16 and 0.14 at peak flowering (7 May 2018) – and then decreased (Fig. 2a). The reflectance of the blue band was lower than that of the red and green bands during flowering. This outcome is similar to the results of previous research (Ashourloo et al., 2019; Sulik and Long, 2015). We also calculated the normalized difference yellow index (NDYI), which can capture increasing yellowness in a time series (d'Andrimont et al., 2020; Sulik and Long, 2016), as follows:
where green is the TOA reflectance of the green band (b3) of the S2 imagery, and blue is the blue band (b2) reflectance. NDYI increased from −0.03 on 17 April to 0.21 on 7 May (Fig. 2b) and reached a peak during rapeseed flowering. This unique spectral feature of rapeseed during the flowering period was due to the yellow petals.
S1 backscattering changes with rapeseed growth. We used VV and VH time series smoothed by the Savitzky–Golay (SG) filter (window size 3) (Chen et al., 2004) as inputs to identify the phenological parameters of rapeseed parcels. We used the SG filter algorithm in MATLAB 2020b, which uncovered local minima in both the VV (−11.20, 8 May) and VH (−15.60, 5 May) time series during rapeseed flowering (Fig. 2c, d). Furthermore, VH reached a maximum (−9.64, 1 June) during the pod period. Unlike other crops, rapeseed has two distinct green-up phases: the flowering period and the pod period (Ashourloo et al., 2019; Bargiel, 2017; Mercier et al., 2020; Veloso et al., 2017). The petals of rapeseed decrease the scattering of VV and VH, while the pods increase the scattering intensity of VH (d'Andrimont et al., 2020; Bargiel, 2017; McNairn et al., 2009; Mercier et al., 2020). The NDYI and backscattering (VV and VH) time series of rapeseed in different climate regions (Fig. S1) exhibited the same characteristics. Therefore, we used both optical and SAR features to identify rapeseed flowering and pod periods in this study. Because of differences in the revisit times of the S1/2 satellites, rapeseed peak flowering dates were defined as the median dates extracted using optical and SAR indicators.
2.3.2 Sample blocks collected for phenological monitoring
As a prerequisite to large-scale mapping, the phenology of rapeseed in different countries must be identified and delineated (Dong et al., 2016; Zhang et al., 2020), but not enough observational records of rapeseed phenology are available on a large scale. In accordance with the DWD method of phenological observation (Kaspar et al., 2014), we created sample blocks with a radius of 10 km over rapeseed-producing areas of different countries and randomly sampled 10 rapeseed parcels per block. The rapeseed plots were identified by their phenological characteristics, which were obtained by visual interpretation and analysis of reference data, including high-resolution images available in S2 and Google Earth as well as spectral reflectance (red and green bands), spectral index (NDYI), and scattering coefficient profiles (VV and VH) from the S1/2 time series. Google Earth images taken during rapeseed flowering were used to assist with the visual interpretation of rapeseed parcels. Rapeseed parcels with no available high-quality time-series imagery were omitted. Finally, 75 sample blocks in 2017, 84 sample blocks in 2018, and 84 sample blocks in 2019 were uniformly and randomly collected (Fig. 1).
2.3.3 Detection of flowering and pod phases in different countries
To determine the flowering dates of rapeseed in different countries, we evaluated each phenological sample block from 2017 to 2019 (Fig. 3). First, we calculated the average values of all pixels in the 10 previously selected rapeseed parcels in each block during the rapeseed growth period in conjunction with the crop calendar. VV and VH time series for each sampled rapeseed parcel were smoothed using the SG filter. Second, S1/2 peak flowering dates and pod dates were derived for all sample blocks according to the method in Sect. 2.3.1. We found that the peak flowering dates of rapeseed, especially in Europe, followed an obvious latitudinal gradient (Fig. 3j).
We also observed that the signal with the maximum VH occurred within 45 d of the peak flowering date (Fig. S2). We then calculated the difference in the peak flowering date of each sample block between different years, which revealed that the flowering peak dates of most sample parcels were advanced or delayed by only 10 d (Fig. 4d). Using the same period for different years in a given area was thus considered to be reasonable for rapeseed identification in this study. Previous studies and field observation records have indicated that the flowering period of rapeseed is approximately 30 d (d'Andrimont et al., 2020; Chen et al., 2019; Kaspar et al., 2014; She et al., 2015). Therefore, we divided each month into two time periods, with the 15th day serving as the dividing line. Two consecutive half-months were defined as a suitable period for classifying flowering dates (Fig. 4a–c). Finally, we designated the flowering period for each administrative unit based on the sample blocks.
2.3.4 Development of a phenology- and pixel-based algorithm for mapping rapeseed
Our temporal profiling of rapeseed parcels along with the results of many previous studies indicated that the spectrum at the flowering stage and the scattering signal at the pod stage are key features for identifying rapeseed (Ashourloo et al., 2019; Bargiel, 2017; Han et al., 2020; Mercier et al., 2020; Sulik and Long, 2015; Veloso et al., 2017). We developed a single phenology- and pixel-based rapeseed mapping algorithm that relies on four features: spectral bands (red and green), spectral indices (NDYI), polarization bands (VH), and terrain (slope). Four primary steps were used to map annual planting areas (Fig. 6).
In the first step, we determined the threshold of the feature indicators. We analyzed the histograms of random samples selected from different countries as suggested by previous study (Zou et al., 2018). Green and blue bands and NDYI (Fig. S3) were similar during the flowering stage in all samples from the different regions. Most rapeseed pixels (98 %) had the following values: red > 0.07, green > 0.11, and NDYI > 0.05. We observed some pixels, however, with a relatively high NDYI due to contamination by a cloud with a “rainbow” appearance, which would cause them to be misclassified as rapeseed (Fig. 5). Because of the limited quality of the QA band and the simple cloud score algorithm, such misclassifications arising from poor-quality observations from the S2 image cannot be removed (X. Wang et al., 2020b; Zhu et al., 2015). The rainbow in the cloud is the result of the push-broom design of S2 (Fig. 5a) and spectral misregistration (for more details, see ESA, 2015a, and b). Taking into account the relative displacement of each spectral channel sensor in the S2 push-broom design (Frantz et al., 2018; X. Liu et al., 2020; Zhao et al., 2018), we developed a new spectral index (NRGBI) to reduce the influence of the rainbow (Eq. 2):
where red, green, and blue are the TOA reflectance values of the red (b4), green (b3), and blue (b2) bands of the S2 imagery, respectively. A scatter plot of NDYI vs. NRGBI of rapeseed parcel samples and “rainbow” samples around clouds (visual interpretation) demonstrated that the NRGBI (threshold = −0.05) can effectively distinguish rapeseed from the rainbow (Fig. 5h). The GEE code for NRGBI index calculations can be found at https://code.earthengine.google.com/a39fc699a276d018778d59c5b085d960 (last access: 8 June 2021). In addition, NRGBI can be calculated using Eq. (2) in other GIS software programs (e.g., QGIS and ArcGIS) on a local computer.
The second step was the identification of all rapeseed pixels from different images during the flowering period and their subsequent aggregation into annual rapeseed planting areas (Fig. 6). Because peak flowering dates and the number of available images of rapeseed fields vary within a region (Fig. S4), rapeseed classifications based on a single image may fail to capture rapeseed flowering dynamics (Ashourloo et al., 2019). To avoid misclassification due to poor-quality observations during the rapeseed flowering stage, we aggregated all results classified from available S2 images during this period. The use of a larger number of images resulted in better performance (Fig. S4).
In the third step, we combined optical data with SAR images to ensure the accuracy of the rapeseed maps. High VH values during the pod stage are another distinct feature that can distinguish rapeseed from other crops (Mercier et al., 2020; Tian et al., 2019; Van Tricht et al., 2018; Veloso et al., 2017). Taking into consideration the variability of flowering in different fields and the duration of the pod stage (Sect. 2.3.2), we calculated the maximum VH between the second half of the flowering stage and the next 30 d (ca. 45 d; Fig. 6). Within this 45 d interval, at least three S1 satellite images were available in the study area. In addition, areas with a slope ≥10∘ were removed (Jarasiunas, 2016). All pixels meeting these requirements were defined as rapeseed.
In the fourth step, we removed “salt and pepper” noise by applying a threshold based on the number of connected components (objects), that is, the size of the neighborhood in pixels, and then filling the gaps inside the parcels (Hirayama et al., 2019). We used an eight-connected rule, which means that the edges or corners of the pixels were connected. If two adjacent pixels were connected, they were considered part of the same object (https://www.mathworks.com/help/images/ref/bwareaopen.html, last access: 8 June 2021). The bwareaopen function in MATLAB (version 2020b) was used to remove objects not meeting a given threshold. The thresholds of different indicators in different regions are given in Table S1.
2.4 Accuracy assessment
To test the accuracy of our proposed algorithm, we first compared rapeseed areas retrieved using the new method with FAO statistics. Our rapeseed data constituted a binary (0 or 1) map with a spatial resolution of 10 m. We calculated the total area of rapeseed maps in each country and compared these numbers with FAO national rapeseed statistics. To verify the accuracy of rapeseed mapping, we used the RMSE (Eq. 3), MAE (Eq. 4), and coefficient of determination (R2, Eq. 5), which were calculated as follows:
where n is the total number of countries, yi is the mapped rapeseed planting area for country i, is the corresponding mean value, fi is the rapeseed planting area recorded by the FAO for country i, and is the corresponding mean value.
We also compared our rapeseed maps with four open-access datasets that include rapeseed layers at the pixel level: ACI, CDL, CROME, and LCMF in Canada, the USA, GBR, and France, respectively. We used the data from 2018 and 2019 in these datasets as a reference (Boryan et al., 2011; Fisette et al., 2013). To unify the spatial resolution of our rapeseed maps, we resampled CDL, ACI, and CROME to 10 m resolution to allow comparison. To check the accuracy of our classification, we calculated UA (Eq. 6), PA (Eq. 7), and F1 (Eq. 8) based on confusion matrices (Table S2).
We also randomly selected verification samples based on previous studies (Pekel et al., 2016; X. Wang et al., 2020a) to validate our rapeseed maps. A 0.2∘ × 0.2∘ latitude–longitude grid was superimposed on our 2018 rapeseed map (Fig. S5). Two points – one rapeseed and the other non-rapeseed – were randomly generated in each grid by visually interpreting images available from S2 and Google Earth along with spectral reflectance (red and green bands), spectral index (NDYI), and scattering coefficient (VV and VH) profiles from the S1/2 time series. Confusion matrices were similarly used to assess accuracy according to Eqs. (6)–(8):
In the above equations, xij is the value of the ith row and jth column, xi is the sum of the ith row, and xj is the sum of the jth column. Although the statistical data and existing products did not completely reflect the actual areas and locations of cultivated rapeseed, these datasets were still beneficial for validating the accuracy of our rapeseed maps at national and pixel scales.
3.1 Accuracy assessment
We compared our derived rapeseed areas with those from FAO statistics. The total planting areas of rapeseed exhibited good consistency with the agricultural statistics at the national level, with a RMSE of 1459.64 km2, a MAE of 785.25 km2, and an R2 of 0.88 (Fig. 7). We found that the derived areas in 2018 were larger than those in 2017 and 2019, especially in countries with relatively small rapeseed areas. The greater availability of S2 images and higher-quality data in 2018 may have contributed to the derivation of the larger areas by our new method (X. Liu et al., 2020).
As indicated by their accuracy based on confusion matrix values, our rapeseed maps were consistent at the pixel level with maps of the American CDL in 2018 and 2019 and the Canadian ACI, British CROME, and French LCMF in 2018 (Table S3). As shown in Fig. 8a, the rapeseed areas calculated from our maps were consistently more comparable to FAO statistics than those from existing products. UA, PA, and F1, which varied by country, ranged from 0.93–0.97, 0.70–0.80, and 0.81–0.86, respectively (Fig. 8b). The rapeseed areas determined using our algorithm accounted for approximately 71 % of the 2018 CDL, 71 % of the 2018 ACI, 80 % of the 2018 CROME, 70 % of the 2018 LCMF, and 79 % of the 2019 CDL. In addition, the distributions on our rapeseed maps were consistent with those of existing products at the pixel level (Figs. S7 and S8). The differences in accuracy may have been due to the varied number of high-quality images available in different regions (Dong et al., 2016). Despite the different ground conditions, methods, images, and spatial resolutions among products, the results of our comparison further verify the accuracy of our rapeseed maps (Gong et al., 2020; Singha et al., 2019).
According to confusion matrix values (Table S4) based on random sampling points, the accuracy of our rapeseed maps varied in different regions. We obtained the highest accuracy (F1, 0.91) in zone II, followed by zone III (F1, 0.9), and zone I (F1, 0.84). These disparities may be due to differences in the availability of S1/2 images among the studied areas.
3.2 Additional features of rapeseed maps derived using our new method
To further characterize the rapeseed maps generated in this study, we selected various images in several areas of each country. The rapeseed maps showed good spatial consistency with the actual areas of rapeseed cultivation on the ground (Figs. 9 and S6). Fields with various planting densities, ranging from densely planted areas in Canada (Fig. 9a) to relatively sparse ones, such as in Chile (Fig. 9b) and European countries (e.g., Fig. 9c, d) (Lowder et al., 2016); various shapes, ranging from regular rectangles (e.g., Fig. 9a, h) to irregular parcels (Fig. 9c, d); and different climatic conditions, ranging from a temperate oceanic climate (Fig. 9c–e) to temperate sub-continental (Fig. 9a, f) or even subtropical (Fig. 9b) ones, were clearly and comprehensively indicated on our maps. The fragmented pattern of land in some European countries, especially that in eastern and central Europe due to land reform in 1989 (Hartvigsen, 2013, 2014), was clearly evident; Fig. 9f shows land in Estonia as an example (Jürgenson and Rasva, 2020; Looga et al., 2018). Although the algorithm was applied to different climates, terrains, and landscapes over a very large region, its classification accuracy across 33 countries was satisfactory. Our rapeseed maps can thus effectively identify fields in detail with high spatial resolution and clear field boundaries. More rapeseed classification details can be found in Fig. S6.
3.3 Spatial patterns of rapeseed planting areas
In our maps, the largest total area of rapeseed cultivation worldwide was in Canada. Along with GBR, Poland, and Ukraine, the two leading rapeseed growing countries in Europe – France and Germany – accounted for approximately 66.3 % of European rapeseed areas. The 3-year (2017–2019) spatial patterns were consistent at the national level (Fig. S9). We also plotted the geographic distribution of rapeseed areas along latitudinal and longitudinal gradients in the study areas (Fig. 10). With the exception of steep mountainous regions and cold northern areas, rapeseed is widely planted in European countries at latitudes of 46–53∘ N and longitudes of 2∘ W–4∘ E, 9–19∘ E, and 22–27∘ E (Fig. 10a). In Canada and the USA, areas with latitudes of 49–54 and 56–57∘ N and longitudes of 117 to 118∘ W and 98 to 114∘ W have high densities of planted rapeseed (Fig. 10b).
4.1 Investigation of rapeseed rotation systems
We obtained 3-year rapeseed maps at a spatial resolution of 10 m whose high accuracy was validated by annual national statistics books, open-access public products, and random sampling points. These rapeseed maps provided a new opportunity to investigate rapeseed rotation systems (X. Liu et al., 2018). Crop rotation information is an important factor in crop yield management (Harker et al., 2015; X. Liu et al., 2018; Ren et al., 2015; Rudiyanto et al., 2019; Zhou et al., 2015). To analyze rapeseed rotation patterns, we selected 25 representative areas (Fig. S10) that met the following three criteria: high image quality, high rapeseed classification accuracy, and large extent of planted rapeseed. Rapeseed rotation in these areas was calculated based on the frequency of each rapeseed pixel (Fig. 11).
Because only 3 years (2017–2019) of rapeseed maps were available, the longest observable rapeseed rotation break was 2 years. To more accurately discern the pattern of rapeseed rotation, we thus classified rapeseed rotation breaks in this study into three types: ≥2 years, 1 year, and 0 years. Most countries, especially European ones, were characterized by rotation breaks that were ≥2 years (mostly green areas in Fig. 12). In Canada, 70 % of fields were subjected to rotation breaks of ≥2 years, with the remainder (30 %) following a 1-year break pattern (Fig. 12a). As shown in the histogram in Fig. 12d, we identified 20 locations with ≥2-year rotation breaks, which corresponds to 90 % of planting areas. Many previous studies have found that a 2- or 3-year rotation break significantly reduces the number of fungal spores, especially those of Rhizoctonia solani and Leptosphaeria maculans, thus suggesting that a rotation system is an important component of disease control in rapeseed (Gill, 2018; Harker et al., 2015; Ren et al., 2015; Zhou et al., 2015). Rapeseed rotation also improves yield, moisture, and fertility and reduces weeds and pest insects (Bernard et al., 2012; Harker et al., 2015; Pardo et al., 2015; Peng et al., 2015; Ren et al., 2015). Additional efforts to produce longer time-series rapeseed maps and acquire detailed rotation information are thus needed.
The generation of annual high-resolution maps for a specific crop over a large region is a major challenge (Dong et al., 2016; Luo et al., 2020; Zhang et al., 2020). Pixel-and phenological-based algorithms, multisource remote sensing data, and the GEE are useful for mapping rapeseed at high resolution and over large areas. In addition to these advantages, our proposed algorithm does not require large amounts of training sample data and reduces disturbance due to agronomic differences by combining images from multiple dates. Nevertheless, uncertainty still exists due to several issues. The first of these factors is the cropland layer. We used GFSAD30 datasets to identify cropland; however, the GFSAD30 has limitations, such as classification errors (Phalke et al., 2020). A second contributory aspect is the number of satellite images available. Although our annual rapeseed maps are consistent with FAO statistics and show higher accuracy compared with existing products, they are limited by the quality of the observations during critical growth stages. For example, Fig. 13a shows an error in an area of France in 2017 that can be attributed to the lack of clear S2 images during the rapeseed flowering period (Fig. 13b). Because the rapeseed flowering period is generally characterized by high NDYI and high red and green band reflectance, rapeseed pixels are likely to be misclassified if images are missing during the flowering stage (Fig. 13c). A third issue concerns the threshold for different indicators, which is a key factor for mapping crops (Ashourloo et al., 2019; Dong et al., 2016; X. Liu et al., 2020; J. Wang et al., 2020; Zhang et al., 2015). Although reference thresholds for the three continents in this study are provided, they should be applied with caution to other regions. Finally, the complexity of the ground environment can contribute to uncertainty. For example, landscape types might impact the accuracy of rapeseed maps (J. Wang et al., 2020).
The rapeseed maps produced with 10 m resolution in this study are accessible at Mendeley Data (http://dx.doi.org/10.17632/ydf3m7pd4j.3) (Han et al., 2021). The dataset includes a set of GeoTIFF images in the ESPG: 4326 spatial reference system. The values 1 and 0 represent rapeseed and non-rapeseed, respectively. We encourage users to independently verify the rapeseed maps. In addition, Sentinel 1/2 images, CDL, ACI, and SRTM are available on GEE (https://developers.google.com/earth-engine/datasets/, Google Earth Engine Data Catalog, 2021). For more detailed information about the data collected in this work, see Table 1.
Large-scale, high-resolution rapeseed maps are the basis for crop growth monitoring and yield prediction. We developed a new method for mapping rapeseed based on spectral and polarization features and multi-source data. We used the new algorithm to produce three annual rapeseed maps (2017–2019) at 10 m spatial resolution in 33 countries. According to the results of three different verification methods, our rapeseed maps are reasonably accurate. Compared with existing products at the pixel level in Canada, USA, GBR, and France, PA, UA, and F1 were 0.70–0.80, 0.93–0.97, and 0.81–0.86, respectively. In addition, F1 ranged from 0.84 to 0.91 based on independent validation samples. Our approach reduces misclassifications due to different planting times and low-quality observations to some degree. The 10 m rapeseed maps provide more spatial details of rapeseed. Finally, we observed that the rapeseed crop rotation interval is at least 2 years in almost all countries in this study. Our proposed rapeseed mapping method can be applied to other regions. The derived rapeseed data product is useful for many purposes, including crop growth monitoring and production and rotation system planning.
The supplement related to this article is available online at: https://doi.org/10.5194/essd-13-2857-2021-supplement.
ZZ and JH designed the research. JH and YL collected datasets. JH implemented the research and wrote the paper. ZZ, JC, LZ, JZ, and ZL revised the paper.
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
We thank the editor and the two anonymous referees for their valuable comments. We thank Liwen Bianji (Edanz) (https://www.liwenbianji.cn, last access: 8 June 2021) for editing the language of a draft of this paper.
This research has been supported by the National Key Research and Development Program of China (grant no. 2019YFA0607401).
This paper was edited by David Carlson and reviewed by two anonymous referees.
Arata, L., Fabrizi, E., and Sckokai, P.: A worldwide analysis of trend in crop yields and yield variability: Evidence from FAO data, Econ. Model., 90, 190–208, https://doi.org/10.1016/j.econmod.2020.05.006, 2020.
Ashourloo, D., Shahrabi, H. S., Azadbakht, M., Aghighi, H., Nematollahi, H., Alimohammadi, A., and Matkan, A. A.: Automatic canola mapping using time series of sentinel 2 images, ISPRS J. Photogramm., 156, 63–76, https://doi.org/10.1016/j.isprsjprs.2019.08.007, 2019.
Bargiel, D.: A new method for crop classification combining time series of radar images and crop phenology information, Remote Sens. Environ., 198, 369–383, https://doi.org/10.1016/j.rse.2017.06.022, 2017.
Bartholomé, E. and Belward, A. S.: GLC2000: a new approach to global land cover mapping from Earth observation data, Int. J. Remote Sens., 26, 1959–1977, https://doi.org/10.1080/01431160412331291297, 2005.
Bernard, E., Larkin, R. P., Tavantzis, S., Erich, M. S., Alyokhin, A., Sewell, G., Lannan, A., and Gross, S. D.: Compost, rapeseed rotation, and biocontrol agents significantly impact soil microbial communities in organic and conventional potato production systems, Appl. Soil Ecol., 52, 29–41, https://doi.org/10.1016/j.apsoil.2011.10.002, 2012.
Boryan, C., Yang, Z., Mueller, R., and Craig, M.: Monitoring US agriculture: the US Department of Agriculture, National Agricultural Statistics Service, Cropland Data Layer Program, Geocarto Int., 26, 341–358, https://doi.org/10.1080/10106049.2011.562309, 2011.
Carré, P. and Pouzet, A.: Rapeseed market, worldwide and in Europe, OCL, 21, D102, https://doi.org/10.1051/ocl/2013054, 2014.
Chen, B., Jin, Y., and Brown, P.: An enhanced bloom index for quantifying floral phenology using multi-scale remote sensing observations, ISPRS J. Photogramm., 156, 108–120, https://doi.org/10.1016/j.isprsjprs.2019.08.006, 2019.
Chen, J., Jönsson, P., Tamura, M., Gu, Z., Matsushita, B., and Eklundh, L.: A simple method for reconstructing a high-quality NDVI time-series data set based on the Savitzky–Golay filter, Remote Sens. Environ., 91, 332–344, https://doi.org/10.1016/j.rse.2004.03.014, 2004.
d'Andrimont, R., Taymans, M., Lemoine, G., Ceglar, A., Yordanov, M., and van der Velde, M.: Detecting flowering phenology in oil seed rape parcels with Sentinel-1 and -2 time series, Remote Sens. Environ., 239, 111660, https://doi.org/10.1016/j.rse.2020.111660, 2020.
Dong, J., Xiao, X., Menarguez, M. A., Zhang, G., Qin, Y., Thau, D., Biradar, C., and Moore, B.: Mapping paddy rice planting area in northeastern Asia with Landsat 8 images, phenology-based algorithm and Google Earth Engine, Remote Sens. Environ., 185, 142–154, https://doi.org/10.1016/j.rse.2016.02.016, 2016.
Drusch, M., Del Bello, U., Carlier, S., Colin, O., Fernandez, V., Gascon, F., Hoersch, B., Isola, C., Laberinti, P., Martimort, P., Meygret, A., Spoto, F., Sy, O., Marchese, F., and Bargellini, P.: Sentinel-2: ESA's Optical High-Resolution Mission for GMES Operational Services, Remote Sens. Environ., 120, 25–36, https://doi.org/10.1016/j.rse.2011.11.026, 2012.
ESA: Sentinel-2 products specification document, available at: https://sentinel.esa.int/documents/247904/349490/S2_MSI_Product_Specification.pdf (last access: 24 April 2019), 2015a.
ESA: Data Product Quality Reports, available at: https://sentinels.copernicus.eu/documents/247904/3902831/Sentinel-2_L1C_Data_Quality_Report/adfff903-a337-4fc1-9439-558456bad0b1?version=1.1 (last access: 8 June 2021), 2015b.
Fang, S., Tang, W., Peng, Y., Gong, Y., Dai, C., Chai, R., and Liu, K.: Remote Estimation of Vegetation Fraction and Flower Fraction in Oilseed Rape with Unmanned Aerial Vehicle Data, Remote Sensing, 8, 416, https://doi.org/10.3390/rs8050416, 2016.
Farr, T. G., Rosen, P. A., Caro, E., Crippen, R., Duren, R., Hensley, S., Kobrick, M., Paller, M., Rodriguez, E., Roth, L., Seal, D., Shaffer, S., Shimada, J., Umland, J., Werner, M., Oskin, M., Burbank, D., and Alsdorf, D.: The Shuttle Radar Topography Mission, Rev. Geophys., 45, RG2004, https://doi.org/10.1029/2005RG000183, 2007.
Firrisa, M. T., van Duren, I., and Voinov, A.: Energy efficiency for rapeseed biodiesel production in different farming systems, Energ. Effic., 7, 79–95, https://doi.org/10.1007/s12053-013-9201-2, 2014.
Fisette, T., Rollin, P., Aly, Z., Campbell, L., Daneshfar, B., Filyer, P., Smith, A., Davidson, A., Shang, J., and Jarvis, I.: AAFC annual crop inventory, in: 2013 Second International Conference on Agro-Geoinformatics (Agro-Geoinformatics), IEEE, Fairfax, VA, USA, https://doi.org/10.1109/Argo-Geoinformatics.2013.6621920, pp. 270–274, 2013.
Frantz, D., Haß, E., Uhl, A., Stoffels, J., and Hill, J.: Improvement of the Fmask algorithm for Sentinel-2 images: Separating clouds from bright surfaces based on parallax effects, Remote Sens. Environ., 215, 471–481, https://doi.org/10.1016/j.rse.2018.04.046, 2018.
Fuglie, K. O.: Total factor productivity in the global agricultural economy: Evidence from FAO data, in: The shifting patterns of agricultural production and productivity worldwide, edited by: Alston, J. M., Babcock, B., and Pardey, P. G., Midwest Agribusiness Trade and Research Information Center, Ames, Iowa, 63–95, 2010.
Gill, K. S.: Crop rotations compared with continuous canola and wheat for crop production and fertilizer use over 6 yr, edited by: Willenborg, C., Can. J. Plant Sci., 98, 1139–1149, https://doi.org/10.1139/cjps-2017-0292, 2018.
Gong, P., Wang, J., Yu, L., Zhao, Y., Zhao, Y., Liang, L., Niu, Z., Huang, X., Fu, H., Liu, S., Li, C., Li, X., Fu, W., Liu, C., Xu, Y., Wang, X., Cheng, Q., Hu, L., Yao, W., Zhang, H., Zhu, P., Zhao, Z., Zhang, H., Zheng, Y., Ji, L., Zhang, Y., Chen, H., Yan, A., Guo, J., Yu, L., Wang, L., Liu, X., Shi, T., Zhu, M., Chen, Y., Yang, G., Tang, P., Xu, B., Giri, C., Clinton, N., Zhu, Z., Chen, J., and Chen, J.: Finer resolution observation and monitoring of global land cover: first mapping results with Landsat TM and ETM+ data, Int. J. Remote Sens., 34, 2607–2654, https://doi.org/10.1080/01431161.2012.748992, 2013.
Gong, P., Li, X., Wang, J., Bai, Y., Chen, B., Hu, T., Liu, X., Xu, B., Yang, J., Zhang, W., and Zhou, Y.: Annual maps of global artificial impervious area (GAIA) between 1985 and 2018, Remote Sens. Environ., 236, 111510, https://doi.org/10.1016/j.rse.2019.111510, 2020.
Google Earth Engine Data Catalog: A planetary-scale platform for Earth science data & analysis, available at: https://developers.google.com/earth-engine/datasets/, last access: 8 June 2021
Gorelick, N., Hancher, M., Dixon, M., Ilyushchenko, S., Thau, D., and Moore, R.: Google Earth Engine: Planetary-scale geospatial analysis for everyone, Remote Sensing of Environment, 202, 18–27, https://doi.org/10.1016/j.rse.2017.06.031, 2017.
Griffiths, P., Nendel, C., and Hostert, P.: Intra-annual reflectance composites from Sentinel-2 and Landsat for national-scale crop and land cover mapping, Remote Sens. Environ., 220, 135–151, https://doi.org/10.1016/j.rse.2018.10.031, 2019.
Han, J., Zhang, Z., and Cao, J.: Developing a New Method to Identify Flowering Dynamics of Rapeseed Using Landsat 8 and Sentinel-1/2, Remote Sensing, 13, 105, https://doi.org/10.3390/rs13010105, 2020.
Han, J., Zhang, Z., Luo, Y., Cao, J., Zhang, L., Zhang, J., and Li, Z.: Data for: Developing a phenology- and pixel-based algorithm for mapping rapeseed at 10m spatial resolution using multi-source data, Mendeley Data, V3, https://doi.org/10.17632/ydf3m7pd4j.3, 2021.
Harker, K. N., O'Donovan, J. T., Turkington, T. K., Blackshaw, R. E., Lupwayi, N. Z., Smith, E. G., Johnson, E. N., Gan, Y., Kutcher, H. R., Dosdall, L. M., and Peng, G.: Canola rotation frequency impacts canola yield and associated pest species, Can. J. Plant Sci., 95, 9–20, https://doi.org/10.4141/cjps-2014-289, 2015.
Hartvigsen, M.: Land reform in Central and Eastern Europe after 1989 and its outcome in the form of farm structures and land fragmentation, FAO Land Tenure Working Paper 24, Food and Agriculture Organization of the United Nations, Rome, Italy, 2013.
Hartvigsen, M.: Land reform and land fragmentation in Central and Eastern Europe, Land Use Policy, 36, 330–341, https://doi.org/10.1016/j.landusepol.2013.08.016, 2014.
Hassan, M. Hj. and Kalam, Md. A.: An Overview of Biofuel as a Renewable Energy Source: Development and Challenges, Procedia Engineer., 56, 39–53, https://doi.org/10.1016/j.proeng.2013.03.087, 2013.
Hirayama, H., Sharma, R. C., Tomita, M., and Hara, K.: Evaluating multiple classifier system for the reduction of salt-and-pepper noise in the classification of very-high-resolution satellite images, Int. J. Remote Sens., 40, 2542–2557, https://doi.org/10.1080/01431161.2018.1528400, 2019.
Höök, M. and Tang, X.: Depletion of fossil fuels and anthropogenic climate change – A review, Energ. Policy, 52, 797–809, https://doi.org/10.1016/j.enpol.2012.10.046, 2013.
Inglada, J., Vincent, A., Arias, M., Tardy, B., Morin, D., and Rodes, I.: Operational High Resolution Land Cover Map Production at the Country Scale Using Satellite Image Time Series, Remote Sensing, 9, 95, https://doi.org/10.3390/rs9010095, 2017.
Jarasiunas, G.: Assessment of the agricultural land under steep slope in Lithuania, J. Cent. Eur. Agric., 17, 176–187, https://doi.org/10.5513/JCEA01/17.1.1688, 2016.
Jürgenson, E. and Rasva, M.: The Changing Structure and Concentration of Agricultural Land Holdings in Estonia and Possible Threat for Rural Areas, Land, 9, 41, https://doi.org/10.3390/land9020041, 2020.
Kaspar, F., Zimmermann, K., and Polte-Rudolf, C.: An overview of the phenological observation network and the phenological database of Germany's national meteorological service (Deutscher Wetterdienst), Adv. Sci. Res., 11, 93–99, https://doi.org/10.5194/asr-11-93-2014, 2014.
Liu, J., Zhu, W., Atzberger, C., Zhao, A., Pan, Y., and Huang, X.: A Phenology-Based Method to Map Cropping Patterns under a Wheat-Maize Rotation Using Remotely Sensed Time-Series Data, Remote Sensing, 10, 1203, https://doi.org/10.3390/rs10081203, 2018.
Liu, X., Hu, G., Chen, Y., Li, X., Xu, X., Li, S., Pei, F., and Wang, S.: High-resolution multi-temporal mapping of global urban land using Landsat images based on the Google Earth Engine Platform, Remote Sens. Environ., 209, 227–239, https://doi.org/10.1016/j.rse.2018.02.055, 2018.
Liu, X., Huang, Y., Xu, X., Li, X., Li, X., Ciais, P., Lin, P., Gong, K., Ziegler, A. D., Chen, A., Gong, P., Chen, J., Hu, G., Chen, Y., Wang, S., Wu, Q., Huang, K., Estes, L., and Zeng, Z.: High-spatiotemporal-resolution mapping of global urban change from 1985 to 2015, Nat. Sustain., 3, 564–570, https://doi.org/10.1038/s41893-020-0521-x, 2020.
Liu, Y., Xu, B., Zhi, W., Hu, C., Dong, Y., Jin, S., Lu, Y., Chen, T., Xu, W., Liu, Y., Zhao, B., and Lu, W.: Space eye on flying aircraft: From Sentinel-2 MSI parallax to hybrid computing, Remote Sens. Environ., 246, 111867, https://doi.org/10.1016/j.rse.2020.111867, 2020.
Looga, J., Jürgenson, E., Sikk, K., Matveev, E., and Maasikamäe, S.: Land fragmentation and other determinants of agricultural farm productivity: The case of Estonia, Land Use Policy, 79, 285–292, https://doi.org/10.1016/j.landusepol.2018.08.021, 2018.
Lowder, S. K., Skoet, J., and Raney, T.: The Number, Size, and Distribution of Farms, Smallholder Farms, and Family Farms Worldwide, World Dev., 87, 16–29, https://doi.org/10.1016/j.worlddev.2015.10.041, 2016.
Luo, Y., Zhang, Z., Chen, Y., Li, Z., and Tao, F.: ChinaCropPhen1km: a high-resolution crop phenological dataset for three staple crops in China during 2000–2015 based on leaf area index (LAI) products, Earth Syst. Sci. Data, 12, 197–214, https://doi.org/10.5194/essd-12-197-2020, 2020.
Malça, J. and Freire, F.: Energy and Environmental Benefits of Rapeseed Oil Replacing Diesel, Int. J. Green Energy, 6, 287–301, https://doi.org/10.1080/15435070902886551, 2009.
McNairn, H., Champagne, C., Shang, J., Holmstrom, D., and Reichert, G.: Integration of optical and Synthetic Aperture Radar (SAR) imagery for delivering operational annual crop inventories, ISPRS J. Photogramm., 64, 434–449, https://doi.org/10.1016/j.isprsjprs.2008.07.006, 2009.
McNairn, H., Jiao, X., Pacheco, A., Sinha, A., Tan, W., and Li, Y.: Estimating canola phenology using synthetic aperture radar, Remote Sens. Environ. 219, 196–205, https://doi.org/10.1016/j.rse.2018.10.012, 2018.
Mercier, A., Betbeder, J., Baudry, J., Le Roux, V., Spicher, F., Lacoux, J., Roger, D., and Hubert-Moy, L.: Evaluation of Sentinel-1 & 2 time series for predicting wheat and rapeseed phenological stages, ISPRS J. Photogramm., 163, 231–256, https://doi.org/10.1016/j.isprsjprs.2020.03.009, 2020.
Pan, Z., Huang, J., and Wang, F.: Multi range spectral feature fitting for hyperspectral imagery in extracting oilseed rape planting area, Int. J. Appl. Earth Obs., 25, 21–29, https://doi.org/10.1016/j.jag.2013.03.002, 2013.
Pardo, N., Sánchez, M. L., Pérez, I. A., and García, M. A.: Energy balance and partitioning over a rotating rapeseed crop, Agr. Water Manage., 161, 31–40, https://doi.org/10.1016/j.agwat.2015.07.015, 2015.
Peel, M. C., Finlayson, B. L., and McMahon, T. A.: Updated world map of the Köppen-Geiger climate classification, Hydrol. Earth Syst. Sci., 11, 1633–1644, https://doi.org/10.5194/hess-11-1633-2007, 2007.
Pekel, J.-F., Cottam, A., Gorelick, N., and Belward, A. S.: High-resolution mapping of global surface water and its long-term changes, Nature, 540, 418–422, https://doi.org/10.1038/nature20584, 2016.
Peng, G., Pageau, D., Strelkov, S. E., Gossen, B. D., Hwang, S.-F., and Lahlali, R.: A >2-year crop rotation reduces resting spores of Plasmodiophora brassicae in soil and the impact of clubroot on canola, Eur. J. Agron., 70, 78–84, https://doi.org/10.1016/j.eja.2015.07.007, 2015.
Phalke, A. R., Özdoğan, M., Thenkabail, P. S., Erickson, T., Gorelick, N., Yadav, K., and Congalton, R. G.: Mapping croplands of Europe, Middle East, Russia, and Central Asia using Landsat, Random Forest, and Google Earth Engine, ISPRS J. Photogramm., 167, 104–122, https://doi.org/10.1016/j.isprsjprs.2020.06.022, 2020.
Preidl, S., Lange, M., and Doktor, D.: Introducing APiC for regionalised land cover mapping on the national scale using Sentinel-2A imagery, Remote Sens. Environ., 240, 111673, https://doi.org/10.1016/j.rse.2020.111673, 2020.
Ren, T., Li, H., Lu, J., Bu, R., Li, X., Cong, R., and Lu, M.: Crop rotation-dependent yield responses to fertilization in winter oilseed rape (Brassica napus L.), The Crop Journal, 3, 396–404, https://doi.org/10.1016/j.cj.2015.04.007, 2015.
Rondanini, D. P., Gomez, N. V., Agosti, M. B., and Miralles, D. J.: Global trends of rapeseed grain yield stability and rapeseed-to-wheat yield ratio in the last four decades, Eur. J. Agron., 37, 56–65, https://doi.org/10.1016/j.eja.2011.10.005, 2012.
Rondanini, D. P., del Pilar Vilariño, M., Roberts, M. E., Polosa, M. A., and Botto, J. F.: Physiological responses of spring rapeseed (Brassica napus) to red/far-red ratios and irradiance during pre- and post-flowering stages, Physiol. Plantarum, 152, 784–794, https://doi.org/10.1111/ppl.12227, 2014.
Rudiyanto, R., Minasny, B., Shah, R. M., Soh, N. C., Arif, C., and Setiawan, B. I.: Automated Near-Real-Time Mapping and Monitoring of Rice Extent, Cropping Patterns, and Growth Stages in Southeast Asia Using Sentinel-1 Time Series on a Google Earth Engine Platform, Remote Sensing, 11, 1666, https://doi.org/10.3390/rs11141666, 2019.
Salmon, J. M., Friedl, M. A., Frolking, S., Wisser, D., and Douglas, E. M.: Global rain-fed, irrigated, and paddy croplands: A new high resolution map derived from remote sensing, crop inventories and climate data, Int. J. Appl. Earth Obs., 38, 321–334, https://doi.org/10.1016/j.jag.2015.01.014, 2015.
Shafiee, S. and Topal, E.: When will fossil fuel reserves be diminished?, Energ. Policy, 37, 181–189, https://doi.org/10.1016/j.enpol.2008.08.016, 2009.
She, B., Huang, J., Guo, R., Wang, H., and Wang, J.: Assessing winter oilseed rape freeze injury based on Chinese HJ remote sensing data, J. Zhejiang Univ. Sci. B, 16, 131–144, https://doi.org/10.1631/jzus.B1400150, 2015.
Singha, M., Dong, J., Zhang, G., and Xiao, X.: High resolution paddy rice maps in cloud-prone Bangladesh and Northeast India using Sentinel-1 data, Sci. Data, 6, 26, https://doi.org/10.1038/s41597-019-0036-3, 2019.
Sulik, J. J. and Long, D. S.: Spectral indices for yellow canola flowers, Int. J. Remote Sens., 36, 2751–2765, https://doi.org/10.1080/01431161.2015.1047994, 2015.
Sulik, J. J. and Long, D. S.: Spectral considerations for modeling yield of canola, Remote Sens. Environ., 184, 161–174, https://doi.org/10.1016/j.rse.2016.06.016, 2016.
Tao, J., Wu, W., Liu, W., and Xu, M.: Exploring the Spatio-Temporal Dynamics of Winter Rape on the Middle Reaches of Yangtze River Valley Using Time-Series MODIS Data, Sustainability, 12, 466, https://doi.org/10.3390/su12020466, 2020.
Teluguntla, P., Thenkabail, P. S., Oliphant, A., Xiong, J., Gumma, M. K., Congalton, R. G., Yadav, K., and Huete, A.: A 30-m landsat-derived cropland extent product of Australia and China using random forest machine learning algorithm on Google Earth Engine cloud computing platform, ISPRS J. Photogramm., 144, 325–340, https://doi.org/10.1016/j.isprsjprs.2018.07.017, 2018.
Tian, H., Meng, M., Wu, M., and Niu, Z.: Mapping spring canola and spring wheat using Radarsat-2 and Landsat-8 images with Google Earth Engine, Curr. Sci., 116, 291–298, 2019.
Torres, R., Snoeij, P., Geudtner, D., Bibby, D., Davidson, M., Attema, E., Potin, P., Rommen, B., Floury, N., Brown, M., Traver, I. N., Deghaye, P., Duesmann, B., Rosich, B., Miranda, N., Bruno, C., L'Abbate, M., Croci, R., Pietropaolo, A., Huchler, M., and Rostan, F.: GMES Sentinel-1 mission, Remote Sensing of Environment, 120, 9–24, https://doi.org/10.1016/j.rse.2011.05.028, 2012.
van Duren, I., Voinov, A., Arodudu, O., and Firrisa, M. T.: Where to produce rapeseed biodiesel and why? Mapping European rapeseed energy efficiency, Renew. Energ., 74, 49–59, https://doi.org/10.1016/j.renene.2014.07.016, 2015.
Van Tricht, K., Gobin, A., Gilliams, S., and Piccard, I.: Synergistic Use of Radar Sentinel-1 and Optical Sentinel-2 Imagery for Crop Mapping: A Case Study for Belgium, Remote Sensing, 10, 1642, https://doi.org/10.3390/rs10101642, 2018.
Veloso, A., Mermoz, S., Bouvet, A., Le Toan, T., Planells, M., Dejoux, J.-F., and Ceschia, E.: Understanding the temporal behavior of crops using Sentinel-1 and Sentinel-2-like data for agricultural applications, Remote Sens. Environ., 199, 415–426, https://doi.org/10.1016/j.rse.2017.07.015, 2017.
Wang, D., Fang, S., Yang, Z., Wang, L., Tang, W., Li, Y., and Tong, C.: A Regional Mapping Method for Oilseed Rape Based on HSV Transformation and Spectral Features, ISPRS Int. Geo-Inf., 7, 224, https://doi.org/10.3390/ijgi7060224, 2018.
Wang, J., Xiao, X., Liu, L., Wu, X., Qin, Y., Steiner, J. L., and Dong, J.: Mapping sugarcane plantation dynamics in Guangxi, China, by time series Sentinel-1, Sentinel-2 and Landsat images, Remote Sens. Environ., 247, 111951, https://doi.org/10.1016/j.rse.2020.111951, 2020.
Wang, X., Xiao, X., Zou, Z., Dong, J., Qin, Y., Doughty, R. B., Menarguez, M. A., Chen, B., Wang, J., Ye, H., Ma, J., Zhong, Q., Zhao, B., and Li, B.: Gainers and losers of surface and terrestrial water resources in China during 1989–2016, Nat. Commun., 11, 3471, https://doi.org/10.1038/s41467-020-17103-w, 2020a.
Wang, X., Xiao, X., Zou, Z., Hou, L., Qin, Y., Dong, J., Doughty, R. B., Chen, B., Zhang, X., Chen, Y., Ma, J., Zhao, B., and Li, B.: Mapping coastal wetlands of China using time series Landsat images in 2018 and Google Earth Engine, ISPRS J. Photogramm., 163, 312–326, https://doi.org/10.1016/j.isprsjprs.2020.03.014, 2020b.
Xiong, J., Thenkabail, P., Tilton, J., Gumma, M., Teluguntla, P., Oliphant, A., Congalton, R., Yadav, K., and Gorelick, N.: Nominal 30-m Cropland Extent Map of Continental Africa by Integrating Pixel-Based and Object-Based Algorithms Using Sentinel-2 and Landsat-8 Data on Google Earth Engine, Remote Sensing, 9, 1065, https://doi.org/10.3390/rs9101065, 2017.
Zhang, G., Xiao, X., Dong, J., Kou, W., Jin, C., Qin, Y., Zhou, Y., Wang, J., Menarguez, M. A., and Biradar, C.: Mapping paddy rice planting areas through time series analysis of MODIS land surface temperature and vegetation index data, ISPRS J. Photogramm., 106, 157–171, https://doi.org/10.1016/j.isprsjprs.2015.05.011, 2015.
Zhang, G., Xiao, X., Dong, J., Xin, F., Zhang, Y., Qin, Y., Doughty, R. B., and Moore, B.: Fingerprint of rice paddies in spatial–temporal dynamics of atmospheric methane concentration in monsoon Asia, Nat. Commun., 11, 554, https://doi.org/10.1038/s41467-019-14155-5, 2020.
Zhang, Y., Chipanshi, A., Daneshfar, B., Koiter, L., Champagne, C., Davidson, A., Reichert, G., and Bédard, F.: Effect of using crop specific masks on earth observation based crop yield forecasting across Canada, Remote Sensing Applications: Society and Environment, 13, 121–137, https://doi.org/10.1016/j.rsase.2018.10.002, 2019.
Zhao, F., Xia, L., Kylling, A., Li, R. Q., Shang, H., and Xu, M.: Detection flying aircraft from Landsat 8 OLI data, ISPRS J. Photogramm., 141, 176–184, https://doi.org/10.1016/j.isprsjprs.2018.05.001, 2018.
Zhong, L., Gong, P., and Biging, G. S.: Efficient corn and soybean mapping with temporal extendability: A multi-year experiment using Landsat imagery, Remote Sens. Environ., 140, 1–13, https://doi.org/10.1016/j.rse.2013.08.023, 2014.
Zhong, L., Hu, L., Yu, L., Gong, P., and Biging, G. S.: Automated mapping of soybean and corn using phenology, ISPRS J. Photogramm., 119, 151–164, https://doi.org/10.1016/j.isprsjprs.2016.05.014, 2016.
Zhou, M., Zhu, B., Brüggemann, N., Wang, X., Zheng, X., and Butterbach-Bahl, K.: Nitrous oxide and methane emissions from a subtropical rice–rapeseed rotation system in China: A 3-year field case study, Agriculture, Ecosystems & Environment, 212, 297–309, https://doi.org/10.1016/j.agee.2015.07.010, 2015.
Zhu, Z., Wang, S., and Woodcock, C. E.: Improvement and expansion of the Fmask algorithm: cloud, cloud shadow, and snow detection for Landsats 4–7, 8, and Sentinel 2 images, Remote Sens. Environ., 159, 269–277, https://doi.org/10.1016/j.rse.2014.12.014, 2015.
Zou, Z., Xiao, X., Dong, J., Qin, Y., Doughty, R. B., Menarguez, M. A., Zhang, G., and Wang, J.: Divergent trends of open-surface water body area in the contiguous United States from 1984 to 2016, P. Natl. Acad. Sci. USA, 115, 3810–3815, https://doi.org/10.1073/pnas.1719275115, 2018.