Search

Winter wheat is a significant global food crop, and it is crucial to monitor its distribution for better agricultural management, land planning, and environmental sustainability. However, the distribution style of winter wheat planting fields is not consistent due to different terrain conditions. In mountainous areas, winter wheat planting units are smaller in size and fragmented in distribution compared to plain areas. Unfortunately, most crop-mapping research based on deep learning ignores the impact of topographic relief on crop distribution and struggles to handle hilly areas effectively. In this paper, we propose a cross-modal segmentation network for winter wheat mapping in complex terrain using remote-sensing multi-temporal images and DEM data. First, we propose a diverse receptive fusion (DRF) module, which applies a deformable receptive field to optical images during the feature fusion process, allowing it to match winter wheat plots of varying scales and a fixed receptive field to the DEM to extract evaluation features at a consistent scale. Second, we developed a distributed weight attention (DWA) module, which can enhance the feature intensity of winter wheat, thereby reducing the omission rate of planting areas, especially for the small-sized regions in hilly terrain. Furthermore, to demonstrate the performance of our model, we conducted extensive experiments and ablation studies on a large-scale dataset in Lanling county, Shandong province, China. Our results show that our proposed CM-Net is effective in mapping winter wheat in complex terrain. [ABSTRACT FROM AUTHOR]

Accurate yield estimation at the field scale is essential for the development of precision agriculture management, whereas at the district level it can provide valuable information for supply chain management. In this paper, Huan Jing (HJ) satellite HJ1A/B and Landsat 8 Operational Land Imager (OLI) images were employed to retrieve leaf area index (LAI) and canopy cover (CC) in the Yangling area (Central China). These variables were then assimilated into two crop models, Aquacrop and simple algorithm for yield (SAFY), in order to compare their performances and practicalities. Due to the models’ specificities and computational constraints, different assimilation methods were used. For SAFY, the ensemble Kalman filter (EnKF) was applied using LAI as the observed variable, while for Aquacrop, particle swarm optimization (PSO) was used, using canopy cover (CC). These techniques were applied and validated both at the field and at the district scale. In the field application, the lowest relative root-mean-square error (RRMSE) value of 18% was obtained using EnKF with SAFY. On a district scale, both methods were able to provide production estimates in agreement with data provided by the official statistical offices. From an operational point of view, SAFY with the EnKF method was more suitable than Aquacrop with PSO, in a data assimilation context. [ABSTRACT FROM AUTHOR]

PROBA-V is a new global vegetation monitoring satellite launched in the second quarter of 2013 that provides data with a 100 m to 1 km spatial resolution and a daily to 10-day temporal resolution in the visible and near infrared (VNIR) bands. A major mission of the PROBA-V satellite is global agriculture monitoring, in which the accuracy of crop mapping plays a key role. In countries such as China, crop fields are typically small, in assorted shapes and with various management approaches, which deem traditional methods of crop identification ineffective, and accuracy is highly dependent on image resolution and acquisition time. The five-day temporal and 100 m spatial resolution PROBA-V data make it possible to automatically identify crops using time series phenological information. This paper takes advantage of the improved spatial and temporal resolution of the PROBA-V data, to map crops at the Yucheng site in Shandong Province and the Hongxing farm in Heilongjiang province of China. First, the Swets filter algorithm was employed to eliminate noisy pixels and fill in data gaps on time series data during the growing season. Then, the crops are classified based on the Iterative Self-Organizing Data Analysis Technique (ISODATA) clustering, the maximum likelihood method (MLC) and similarity analysis. The mapping results were validated using field-collected crop type polygons and high resolution crop maps based on GaoFen-1 satellite (GF-1) data in 16 m resolution. Our study showed that, for the Yucheng site, the cropping system is simple, mainly dominated by winter wheat-maize rotation. The overall accuracy of crop identification was 73.39% which was slightly better than the result derived from MODIS data. For the Hongxing farm, the cropping system is more complex (i.e., more than three types of crops were planted). The overall accuracy of the crop mapping by PROBA-V was 73.29% which was significantly higher than the MODIS product (46.81%). This study demonstrates that time series PROBA-V data can serve as a useful source for reliable crop identification and area estimation. The high revisiting frequency and global coverage of the PROBA-V data show good potential for future global crop mapping and agricultural monitoring. [ABSTRACT FROM AUTHOR]

Detailed information from global remote sensing has greatly advanced our understanding of Earth as a system in general and of agricultural processes in particular. Vegetation monitoring with global remote sensing systems over long time periods is critical to gain a better understanding of processes related to agricultural change over long time periods. This specifically relates to sub-humid to semi-arid ecosystems, where agricultural change in grazing lands can only be detected based on long time series. By integrating data from different sensors it is theoretically possible to construct NDVI time series back to the early 1980s. However, such integration is hampered by uncertainties in the comparability between different sensor products. To be able to rely on vegetation trends derived from integrated time series it is therefore crucial to investigate whether vegetation trends derived from NDVI and phenological parameters are consistent across products. In this paper we analyzed several indicators of vegetation change for a range of agricultural systems in Inner Mongolia, China, and compared the results across different satellite archives. Specifically, we compared two of the prime NDVI archives-AVHRR Global Inventory Modeling and Mapping Studies (GIMMS) and SPOT Vegetation (VGT) NDVI. Because a true accuracy assessment of long time series is not possible, we further compared SPOT VGT NDVI with NDVI from MODIS Terra as a benchmark. We found high similarities in interannual trends, and also in trends of the seasonal amplitude and integral between SPOT VGT and MODIS Terra (r > 0.9). However, we observed considerable disagreements in NDVI-derived trends between AVHRR GIMMS and SPOT VGT. We detected similar discrepancies for trends based on phenological parameters, such as amplitude and integral of NDVI curves corresponding to seasonal vegetation cycles. Inconsistencies were partially related to land cover and vegetation density. Different pre-processing schemes and the coarser spatial resolution of AVHRR GIMMS introduced further uncertainties. Our results corroborate findings from other studies that vegetation trends derived from AVHRR GIMMS data not always reflect true vegetation changes. A more thorough understanding of the factors introducing uncertainties in AVHRR GIMMS time series is needed, and we caution against using AVHRR GIMMS data in regional studies without applying regional sensitivity analyses. [ABSTRACT FROM AUTHOR]