Symptoms of nutrient deficiencies in rice plants often appear on the leaves. The leaf color and shape, therefore, can be used to diagnose nutrient deficiencies in rice. Image classification is an efficient and fast approach for this diagnosis task. Deep convolutional neural networks (DCNNs) have been proven to be effective in image classification, but their use to identify nutrient deficiencies in rice has received little attention. In the present study, we explore the accuracy of different DCNNs for diagnosis of nutrient deficiencies in rice. A total of 1818 photographs of plant leaves were obtained via hydroponic experiments to cover full nutrition and 10 classes of nutrient deficiencies. The photographs were divided into training, validation, and test sets in a 3 : 1 : 1 ratio. Fine-tuning was performed to evaluate four state-of-the-art DCNNs: Inception-v3, ResNet with 50 layers, NasNet-Large, and DenseNet with 121 layers. All the DCNNs obtained validation and test accuracies of over 90%, with DenseNet121 performing best (validation accuracy = 98.62 ± 0.57%; test accuracy = 97.44 ± 0.57%). The performance of the DCNNs was validated by comparison to color feature with support vector machine and histogram of oriented gradient with support vector machine. This study demonstrates that DCNNs provide an effective approach to diagnose nutrient deficiencies in rice.

Deep learning is characterized by its strong ability of data feature extraction. This method can provide unique advantages when applying it to visible and near-infrared spectroscopy for predicting soil organic matter (SOM) content in those cases where the SOM content is negatively correlated with the spectral reflectance of soil. This study relied on the SOM content data of 248 red soil samples and their spectral reflectance data of 400–2450 nm in Fengxin County, Jiangxi Province (China) to meet three objectives. First, a multilayer perceptron and two convolutional neural networks (LeNet5 and DenseNet10) were used to predict the SOM content based on spectral variation and variable selection, and the outcomes were compared with that from the traditional back-propagation neural network (BPN). Second, the four methods were applied to full-spectrum modeling to test the difference to selected feature variables. Finally, the potential of direct modeling was evaluated using spectral reflectance data without any spectral variation. The results of prediction accuracy showed that deep learning performed better at predicting the SOM content than did the traditional BPN. Based on full-spectrum data, deep learning was able to obtain more feature information, thus achieving better and more stable results (i.e., similar average accuracy and far lower standard deviation) than those obtained through variable selection. DenseNet achieved the best prediction result, with a coefficient of determination (R²) = 0.892 ± 0.004 and a ratio of performance to deviation (RPD) = 3.053 ± 0.056 in validation. Based on DenseNet, the application of spectral reflectance data (without spectral variation) produced robust results for application-level purposes (validation R² = 0.853 ± 0.007 and validation RPD = 2.639 ± 0.056). In conclusion, deep learning provides an effective approach to predict the SOM content by visible and near-infrared spectroscopy and DenseNet is a promising method for reducing the amount of data preprocessing.

Reasonable delineation of the urban growth boundary (UGB) plays a vital role in guiding orderly urban space growth and ensuring urban environmental health. Existing methodologies for UGB delineation have failed to address the significance of ecological security. Therefore, this study presents a framework that couples ecological security pattern (ESP) establishment and construction land expansion (CLE) simulation to delineate the UGB. The proposed framework is applied to the Nanchang Metropolitan Area (NCMA) in southeastern China. First, we established the regional ESP of the NCMA in 2018 based on an improved minimum cumulative resistance model. The areas of low-, medium-, and high-level ESP were 1050.75, 736.42, and 720.59 km², respectively. Second, we implemented a multi-scenario simulation of CLE in the NCMA in 2025 based on a cellular automata–Markov model. A natural development scenario was superior to urban growth and ecological protection scenarios for social, economic, and ecological development at the regional scale. Accordingly, we delineated the UGB of the NCMA in 2025 with a scale of 687.87 km², based on dynamic adjustment using the results of ESP establishment and CLE simulation in the natural development scenario. The rationality and scientificity of the proposed framework were verified by comparing the scale and layout of the delineated UGB with the regional planning of Nanchang City. The framework incorporating dynamic adjustment with ESP establishment and multi-scenario CLE simulation provides a useful tool for the delineation of the UGB in similar urbanized cities. Its application is conducive to achieving a win–win outcome of regional ecological security and urban development.

Aims: Intensive soybean/maize intercropping, a specific form of intercropping, holds promise in addressing the challenges posed by increasing food demands, diminishing cropland areas, deteriorating soil quality, and escalating environmental pollution. Methods: To evaluate the potential of this system, we conducted a national meta-analysis, quantifying its absolute yield gain (net effect, NE) and land use efficacy (land equivalent ratio, LER). We further investigated the underlying mechanisms by examining local climate, soil properties, and field management practices and then developed random forest (RF) models to assess the system's potential, incorporating current information on natural resources. Results: In China, an average NE of 3.2 ± 0.1 Mg ha⁻¹ and LER of 1.4 ± 0.02 were achieved by intensive soybean/maize intercropping. The variance of NE was significantly influenced by air temperature (10%), soybean delay days (8%), and maize plant density (9%). Similarly, the LER was strongly driven by soybean delay days (14%), sunshine hours (11%), and maize density (10%). Notably, this intensive intercropping system efficiently utilizes available resources, such as light, temperature (heat), accumulated temperature, and soil nutrients, particularly in regions characterized by low soil fertility and limited agricultural resources. Ultimately, the RF model estimated substantial overyielding of 2 800 kg per hectare, representing approximately 1.4 times the current soybean and maize production under China's monoculture. Conclusions: The implementation of intensive soybean/maize intercropping is highly beneficial throughout China, especially in areas with limited agricultural resources. The Yangtze River Basin, in potentially, emerges as the most suitable region for adopting this intensive intercropping practice.

Sloping farmland is prevalent in hilly red soil areas of South China. Improper tillage patterns induce decreased soil organic matter, soil aggregate breakdown, and nutrient imbalance, thereby restricting crop production. However, the stoichiometric characteristics could reflect the nutrient availability which was mostly studied on bulk soil. The stoichiometric characteristics of soil aggregates with multiple functions in farmlands has rarely been studied. The study was to reveal the impact of tillage patterns on the size distribution, nutrient levels, and stoichiometric ratios of soil aggregates after 20 years’ cultivation. Soil samples of 0–20 cm and 20–40 cm from five tillage patterns, bare-land control (BL), longitudinal-ridge tillage (LR), conventional tillage + straw mulching (CS), cross-ridge tillage (CR), and longitudinal-ridge tillage + hedgerows (LH) were collected. The elemental content (C, N and P) and soil aggregate size distribution were determined, and the stoichiometric ratios were subsequently calculated. Through our analysis and study, it was found that the nutrient content of >2 mm soil aggregates in all plots was the highest. In the hedgerow plots, >2 mm water-stable soil aggregate content was increased. Therefore, LH plots have the highest content of organic matter and nutrients. After 20 years of cultivation, stoichiometric ratio of each plot showed different changes on soil aggregates at different levels. the C:N, C:P, and N:P ratios are lower than the national average of cultivated land. Among of them, the stoichiometric ratio in the LH plot is closer to the mean and showed better water-stable aggregate enhancement. Therefore, longitudinal-ridge tillage + hedgerows can be recommended as a cultivation measure. This study provides a reference for determining appropriate tillage measures, balancing nutrient ratios, and implementing rational fertilization.