Alfalfa is a valuable and intensively produced forage crop in the United States, and the timely estimation of its yield can inform precision management decisions. However, traditional yield assessment approaches are laborious and time-consuming, and thus hinder the acquisition of timely information at the field scale. Recently, unmanned aerial vehicles (UAVs) have gained significant attention in precision agriculture due to their efficiency in data acquisition. In addition, compared with other imaging modalities, hyperspectral data can offer higher spectral fidelity for constructing narrow-band vegetation indices which are of great importance in yield modeling. In this study, we performed an in-season alfalfa yield prediction using UAV-based hyperspectral images. Specifically, we firstly extracted a large number of hyperspectral indices from the original data and performed a feature selection to reduce the data dimensionality. Then, an ensemble machine learning model was developed by combining three widely used base learners including random forest (RF), support vector regression (SVR) and K-nearest neighbors (KNN). The model performance was evaluated on experimental fields in Wisconsin. Our results showed that the ensemble model outperformed all the base learners and a coefficient of determination (R²) of 0.874 was achieved when using the selected features. In addition, we also evaluated the model adaptability on different machinery compaction treatments, and the results further demonstrate the efficacy of the proposed ensemble model.

Winter wheat (Triticum aestivum L.) is one of the most important cereal crops, supplying essential food for the world population. Because the United States is a major producer and exporter of wheat to the world market, accurate and timely forecasting of wheat yield in the United States (U.S.) is fundamental to national crop management as well as global food security. Previous studies mainly have focused on developing empirical models using only satellite remote sensing images, while other yield determinants have not yet been adequately explored. In addition, these models are based on traditional statistical regression algorithms, while more advanced machine learning approaches have not been explored. This study used advanced machine learning algorithms to establish within-season yield prediction models for winter wheat using multi-source data to address these issues. Specifically, yield driving factors were extracted from four different data sources, including satellite images, climate data, soil maps, and historical yield records. Subsequently, two linear regression methods, including ordinary least square (OLS) and least absolute shrinkage and selection operator (LASSO), and four well-known machine learning methods, including support vector machine (SVM), random forest (RF), Adaptive Boosting (AdaBoost), and deep neural network (DNN), were applied and compared for estimating the county-level winter wheat yield in the Conterminous United States (CONUS) within the growing season. Our models were trained on data from 2008 to 2016 and evaluated on data from 2017 and 2018, with the results demonstrating that the machine learning approaches performed better than the linear regression models, with the best performance being achieved using the AdaBoost model (R² = 0.86, RMSE = 0.51 t/ha, MAE = 0.39 t/ha). Additionally, the results showed that combining data from multiple sources outperformed single source satellite data, with the highest accuracy being obtained when the four data sources were all considered in the model development. Finally, the prediction accuracy was also evaluated against timeliness within the growing season, with reliable predictions (R² > 0.84) being able to be achieved 2.5 months before the harvest when the multi-source data were combined.

The main goal of this study was to use the synthetic minority oversampling technique (SMOTE) to expand the quantity of landslide samples for machine learning methods (i.e., support vector machine (SVM), logistic regression (LR), artificial neural network (ANN), and random forest (RF)) to produce high-quality landslide susceptibility maps for Lishui City in Zhejiang Province, China. Landslide-related factors were extracted from topographic maps, geological maps, and satellite images. Twelve factors were selected as independent variables using correlation coefficient analysis and the neighborhood rough set (NRS) method. In total, 288 soil landslides were mapped using field surveys, historical records, and satellite images. The landslides were randomly divided into two datasets: 70% of all landslides were selected as the original training dataset and 30% were used for validation. Then, SMOTE was employed to generate datasets with sizes ranging from two to thirty times that of the training dataset to establish and compare the four machine learning methods for landslide susceptibility mapping. In addition, we used slope units to subdivide the terrain to determine the landslide susceptibility. Finally, the landslide susceptibility maps were validated using statistical indexes and the area under the curve (AUC). The results indicated that the performances of the four machine learning methods showed different levels of improvement as the sample sizes increased. The RF model exhibited a more substantial improvement (AUC improved by 24.12%) than did the ANN (18.94%), SVM (17.77%), and LR (3.00%) models. Furthermore, the ANN model achieved the highest predictive ability (AUC = 0.98), followed by the RF (AUC = 0.96), SVM (AUC = 0.94), and LR (AUC = 0.79) models. This approach significantly improves the performance of machine learning techniques for landslide susceptibility mapping, thereby providing a better tool for reducing the impacts of landslide disasters.

In recent years, research on the spatiotemporal distribution and health effects of fine particulate matter (PM_2.5) has been conducted in China. However, the limitations of different research scopes and methods have led to low comparability between regions regarding the mortality burden of PM_2.5. A kriging model was used to simulate the distribution of PM_2.5 in 2015 and 2016. Relative risk (RR) at a specified PM_2.5 exposure concentration was estimated with an integrated exposure–response (IER) model for different causes of mortality: lung cancer (LC), ischaemic heart disease (IHD), cerebrovascular disease (stroke) and chronic obstructive pulmonary disease (COPD). The population attributable fraction (PAF) was adopted to estimate deaths attributed to PM_2.5. 72.02% of cities experienced decreases in PM_2.5 from 2015 to 2016. Due to the overall decrease in the PM_2.5 concentration, the total number of deaths decreased by approximately 10,658 per million in 336 cities, including a decrease of 1400, 1836, 6312 and 1110 caused by LC, IHD, stroke and COPD, respectively. Our results suggest that the overall PM_2.5 concentration and PM_2.5-related deaths exhibited decreasing trends in China, although air quality in local areas has deteriorated. To improve air pollution control strategies, regional PM_2.5 concentrations and trends should be fully considered.

Determining the flow accumulation threshold (FAT) is a key task in the extraction of river networks from digital elevation models (DEMs). Several methods have been developed to extract river networks from Digital Elevation Models. However, few studies have considered the geomorphologic complexity in the FAT estimation and river network extraction. Recent studies estimated influencing factors’ impacts on the river length or drainage density without considering anthropogenic impacts and landscape patterns. This study contributes two FAT estimation methods. The first method explores the statistical association between FAT and 47 tentative explanatory factors. Specifically, multi-source data, including meteorologic, vegetation, anthropogenic, landscape, lithology, and topologic characteristics are incorporated into a drainage density-FAT model in basins with complex topographic and environmental characteristics. Non-negative matrix factorization (NMF) was employed to evaluate the factors’ predictive performance. The second method exploits fractal geometry theory to estimate the FAT at the regional scale, that is, in basins whose large areal extent precludes the use of basin-wide representative regression predictors. This paper’s methodology is applied to data acquired for Hubei and Qinghai Provinces, China, from 2001 through 2018 and systematically tested with visual and statistical criteria. Our results reveal key local features useful for river network extraction within the context of complex geomorphologic characteristics at relatively small spatial scales and establish the importance of properly choosing explanatory geomorphologic characteristics in river network extraction. The multifractal method exhibits more accurate extracting results than the box-counting method at the regional scale.

Accurate crop mapping is crucial for ensuring food security. Recently, many studies have developed diverse crop mapping models based on deep learning. However, these models generally rely on a large amount of labeled crop samples to investigate the intricate relationship between the crop types of the samples and the corresponding remote sensing features. Moreover, their efficacy is often compromised when applied to other areas owing to the disparities between source and target data. To address this issue, a new multi-modal deep adaptation crop classification network (MDACCN) was proposed in this study. Specifically, MDACCN synergistically exploits time series optical and SAR images using a middle fusion strategy to achieve good classification capacity. Additionally, local maximum mean discrepancy (LMMD) is embedded into the model to measure and decrease domain discrepancies between source and target domains. As a result, a well-trained model in a source domain can still maintain satisfactory accuracy when applied to a target domain. In the training process, MDACCN incorporates the labeled samples from a source domain and unlabeled samples from a target domain. When it comes to the inference process, only unlabeled samples of the target domain are required. To assess the validity of the proposed model, Arkansas State in the United States was chosen as the source domain, and Heilongjiang Province in China was selected as the target domain. Supervised deep learning and traditional machine learning models were chosen as comparison models. The results indicated that the MDACCN achieved inspiring performance in the target domain, surpassing other models with overall accuracy, Kappa, and a macro-averaged F1 score of 0.878, 0.810, and 0.746, respectively. In addition, the crop-type maps produced by the MDACCN exhibited greater consistency with the reference maps. Moreover, the integration of optical and SAR features exhibited a substantial improvement of the model in the target domain compared with using single-modal features. This study indicated the considerable potential of combining multi-modal remote sensing data and an unsupervised domain adaptive approach to provide reliable crop distribution information in areas where labeled samples are missing.