Search

China is currently experiencing a severe issue of excessive fertilization. Variable rate fertilization (VRF) technology is key to solving this issue in precision agriculture, and one way to implement VRF is through management zone (MZ) delineation. This study is aimed at evaluating the feasibility and potential benefits of VRF based on site-specific MZs in smallholder farm fields. This study determined the amounts of basal and top-dressing fertilizers in different spatial units, based on soil nutrient MZs and crop growth MZs, respectively. The potential agronomic, economic, and environmental advantages of spatial variable rate fertilization were further assessed by comparing the farmer's treatment, the expert's treatment, and the variable rate fertilization treatment based on management zones (VR-MZ). The results showed that VR-MZ reduced the use of nitrogen (N), phosphorus (P), and potassium (K) fertilizers by 22.90–43.95%, 59.11–100%, and 8.21–100%, respectively, and it also increased the use efficiency of N, P, and K by 12.27–28.71, 89.64–176.85, and 5.48–266.89 kg/kg, respectively, without yield loss. The net incomes of VR-MZ were 15.5–449.61 USD ha−1 higher than that of traditional spatially uniform rate fertilization. Meanwhile, less nitrous oxide emission (23.50–45.81%), ammonia volatilization (19.38–51.60%), and nitrate ion leaching amounts (28.77–53.98%) were found in VR-MZ compared to those in uniform fertilization. The results suggest that the VR-MZ has great potential for saving fertilizers, significantly increasing farmers' net income, reducing environmental pollution, and promoting the sustainable use of resources. This study provides a theoretical basis and technical support for exploring a VRF suitable for village-scale farming. [ABSTRACT FROM AUTHOR]

Abstract Farm size plays a critical role in agricultural sustainability. This may have far-reaching consequences for the economic and environmental performance of agricultural production, resulting, for instance, in an excessive use of mineral fertilizers. However, the magnitude of such effects and their main causes are not well understood, while being essential for effective policy development, especially for countries like China where the agricultural sector is still largely dominated by smallholder farms. In this paper, we review the current understanding of how farm size affects agricultural sustainability using China's crop farming as an example from economic, environment and social aspects. We analyze impacts from both a Chinese and a global perspective to identify intervention points to improve agricultural performance. We found that increasing farm size has a positive impact on farmer's net profit, as well as economic, technical and labor efficiency with mean coefficients of 0.005, 0.02 and 2.25, respectively. Nevertheless, the relationships between farm size and overall productivity, total factor productivity and allocative efficiency are still not well understood and therefore require more research. Meanwhile, increase in farm size is associated with statistically significant decrease in fertilizer and pesticide use per hectare, showing clear benefits for environmental protection. In line with the experiences documented for the evolution of agricultural practices in developed countries, the expansion of large-scale farming is a critical path for modernizing agricultural production and ensuring sustainable food production from the social perspective. Measures concerning farm size should be implemented in an interaction between farmers and the government to promote the green development of agriculture. Highlights • Farm size plays an important role in the performance of agriculture. • Increasing farm size shows clear benefits for environmental protection. • Large-scale farming is a critical path for modernizing and sustaining agriculture. • Smallholders prefer to use more non-fixed inputs to increase yields. [ABSTRACT FROM AUTHOR]

• A large number of field-surveyed data are used to predict smallholder maize yield. • GCVI outperforms traditional visible and NIR-based VIs. • Satellite-derived CIs perform comparably to ground-observed ones. • Combining the optimal VI and CI captures 78% of the yield variation. • LSTM can reasonably predict final yield two/one-month before harvesting. Timely and accurately estimating smallholder crop yield is essential for optimizing agronomic management, guiding investment and policy-making to reduce poverty and improve food security. However, the productivity of most smallholder farms around the world is still poorly estimated due to absence of technologies and field-scale data, especially for the largest smallholders across China. Here, we integrated 11,857 of field-surveyed yields across maize cultivation areas in China and heterogeneous geospatial data to predict field-level maize yield using three data-driven approaches, i.e., Least Absolute Shrinkage and Selection Operator (LASSO), Light Gradient Boosting Machine (LightGBM) and Long Short-Term Memory (LSTM). We determined the most suitable vegetation index (VI), compared the performances of satellite-derived and ground-observed climate data, and identified the optimal combination of input variables and the best method for maize yield estimation. We found that the green chlorophyll vegetation index (GCVI) outperformed the traditional visible and near-infrared-based VIs. Combining Land Surface Temperature (LST), cumulative precipitation (Pgs) and standardized precipitation index (SPI) explained about 70% of the yield variation, which was comparable to ground-observed indices. Integrating GCVI improved R2 by 0.01–0.20 depending on the methods, suggesting valuable information on biotic or abiotic stress contained by GCVI. Maize yield could be predicted 1–2-month ahead before harvesting in all agro-ecological zones. The data collected during silking period contributed more information. The LSTM model did better than LightGBM and LASSO, because of its neural network characterizing the cumulative effects of environmental factors on yield. Our study demonstrates an effective mean for large-scale crop yield estimation using publicly available data, particularly for smallholder systems where ground observations are currently limited and sparse. [Display omitted] [ABSTRACT FROM AUTHOR]

Explicit spatial information about crop types on smallholder farms is important for the development of local precision agriculture. However, due to highly fragmented and heterogeneous cropland landscapes, fine-scale mapping of smallholder crops, based on low- and medium-resolution satellite images and relying on a single machine learning (ML) classifier, generally fails to achieve satisfactory performance. This paper develops an ensemble ML-based framework to improve the accuracy of parcel-level smallholder crop mapping from very high spatial resolution (VHSR) images. A typical smallholder agricultural area in central China covered by WorldView-2 images is selected to demonstrate our approach. This approach involves the task of distinguishing eight crop-level agricultural land use types. To this end, six widely used individual ML classifiers are evaluated. We further improved their performance by independently implementing bagging and stacking ensemble learning (EL) techniques. The results show that the bagging models improved the performance of unstable classifiers, but these improvements are limited. In contrast, the stacking models perform better, and the Stacking #2 model (overall accuracy = 83.91%, kappa = 0.812), which integrates the three best-performing individual classifiers, performs the best of all of the built models and improves the classwise accuracy of almost all of the land use types. Since classification performance can be significantly improved without adding costly data collection, stacking-ensemble mapping approaches are valuable for the spatial management of complex agricultural areas. We also demonstrate that using geometric and textural features extracted from VHSR images can improve the accuracy of parcel-level smallholder crop mapping. The proposed framework shows the great potential of combining EL technology with VHSR imagery for accurate mapping of smallholder crops, which could facilitate the development of parcel-level crop identification systems in countries dominated by smallholder agriculture. [ABSTRACT FROM AUTHOR]