Search

Landslides are a common geodynamic phenomenon that cause substantial life and property damage worldwide. In the present study, we developed models to evaluate landslide susceptibility in forest-covered areas in Lin'an, southeastern China using logistic regression (LR), decision tree (DT), and random forest (RF) techniques. In addition to conventional landslide-related natural and human disturbance factors, factors describing forest cover, including forest type (two plantations (hickory and bamboo) and four natural forests (conifer, hardwood, shrub, and moso bamboo) and understory vegetation conditions, were included as predictors. Model performance was evaluated based on true-positive rate, Kappa value, and area under the ROC curve using a 10-fold cross-validation method. All models exhibited good performance with measures of ≥0.70, although the LR model was relatively inferior. The key predictors were forest type, understory vegetation height (UVH), normalized differential vegetation index (NDVI) in summer, distance to road (DTRD), and maximum daily rainfall (MDR). Hickory plantations yielded the highest landslide probability, while conifer and hardwood forests had the lowest values. Bamboo plantations had probability results comparable to those of natural forests. Using the RF model, areas with a shorter UVH (<1.2 m), a lower NDVI (<0.70), a heavier MDR (>115 mm), or a shorter DTRD (<500 m) were predicted to be landslide-prone. Information on forest cover is essential for predicting landslides in areas with rich forest cover, and conversion from natural forests to plantations could increase landslide risk. Across the study areas, the northwestern part was the most landslide-prone. In terms of landslide prevention, the RF model-based map produced the most accurate predictions for the "very high" category of landslide. These results will help us better understand landslide occurrences in forest-covered areas and provide valuable information for governments in designing disaster mitigation. [ABSTRACT FROM AUTHOR]

Large-scale crop mapping is of fundamental importance to tackle food security problems. SAR remote sensing has lately received great attention for crop type mapping due to its stability in the revisit cycle and is not hindered by cloud cover. However, most SAR image-classification studies focused on the application of backscattering characteristics with machine learning models, while few investigated the potential of the polarization decomposition and deep-learning models. This study investigated whether the radar polarization information mined by polarization decomposition, the patch strategy and the approaches for combining recurrent and convolutional neural networks (Conv2d + LSTM and ConvLSTM2d) could effectively improve the accuracy of crop type mapping. Sentinel-1 SLC and GRD products in 2020 were collected as data sources to extract VH, VV, VH/VV, VV + VH, Entropy, Anisotropy, and Alpha 7-dimensional features for classification. The results showed that the three-dimensional Convolutional Neural Network (Conv3d) was the best classifier with an accuracy and kappa up to 88.9% and 0.875, respectively, and the ConvLSTM2d and Conv2d + LSTM achieved the second and third position. Compared to backscatter coefficients, the polarization decomposition features could provide additional phase information for classification in the time dimension. The optimal patch size was 17, and the patch-based Conv3d outperformed the pixel-based Conv1d by 11.3% in accuracy and 0.128 in kappa. This study demonstrated the value of applying polarization decomposition features to deep-learning models and provided a strong technical support to efficient large-scale crop mapping. [ABSTRACT FROM AUTHOR]

Machine vision technology enables the continuous and nondestructive monitoring of leaf responses to different nutrient supplies and thereby contributes to the improvement of diagnostic effects. In this study, we analysed the temporal dynamics of rice leaf morphology and colour under different nitrogen (N), phosphorus (P), and potassium (K) treatments by continuous imaging and further evaluated the effectiveness of dynamic characteristics for identification. The top four leaves (the 1st incomplete leaf and the top three fully expanded leaves) were scanned every three days, and all images were processed in MATLAB to extract the morphological and colour characteristics for dynamic analysis. Subsequently, the mean impact value was applied to evaluate the effectiveness of dynamic indices for identification. According to the results, higher nutrient supply resulted in a faster leaf extension rate and a lower developing rate of chlorosis, and the influence of N deficiency on leaf growth was the greatest, followed by P deficiency and then K deficiency. Furthermore, the optimal indices for identification were mainly calculated from morphological characteristics of the 1st incomplete leaf and colour characteristics of the 3rd fully expanded leaf. Overall, dynamic analysis contributes not only to the exploration of the plant growth mechanism but also to the improvement of diagnostics. [ABSTRACT FROM AUTHOR]

Digital image processing is widely used in the non-destructive diagnosis of plant nutrition. Previous plant nitrogen diagnostic studies have mostly focused on characteristics of the rice canopy or leaves at some specific points in time, with the long sampling intervals unable to provide detailed and specific “dynamic features.” According to plant growth mechanisms, the dynamic changing rate in leaf shape and color differ between different nitrogen supplements. Therefore, the objective of this study was to diagnose nitrogen stress levels by analyzing the dynamic characteristics of rice leaves. Scanning technology was implemented to collect rice leaf images every 3 days, with the characteristics of the leaves from different leaf positions extracted utilizing MATLAB. Newly developed shape characteristics such as etiolation area (EA) and etiolation degree (ED), in addition to shape (area, perimeter) and color characteristics (green, normalized red index, etc.), were used to quantify the process of leaf change. These characteristics allowed sensitive indices to be established for further model validation. Our results indicate that the changing rates in dynamic characteristics, in particular the shape characteristics of the first incomplete leaf (FIL) and the characteristics of the 3rd leaf (leaf color and etiolation indices), expressed obvious distinctions among different nitrogen treatments. Consequently, we achieved acceptable diagnostic accuracy (training accuracy 77.3%, validation accuracy 64.4%) by using the FIL at six days after leaf emergence, and the new shape characteristics developed in this article (ED and EA) also showed good performance in nitrogen diagnosis. Based on the aforementioned results, dynamic analysis is valuable not only in further studies but also in practice. [ABSTRACT FROM AUTHOR]

Shadow is one of the major problems in remotely sensed imagery which hampers the accuracy of information extraction and change detection. In these images, shadow is generally produced by different objects, namely, cloud, mountain and urban materials. The shadow correction process consists of two steps: detection and de-shadowing. This paper reviews a range of techniques for both steps, focusing on urban regions (urban shadows), mountainous areas (topographic shadow), cloud shadows and composite shadows. Several issues including the problems and the advantages of those algorithms are discussed. In recent years, thresholding and recovery techniques have become important for shadow detection and de-shadowing, respectively. Research on shadow correction is still an important topic, particularly for urban regions (in high spatial resolution data) and mountainous forest (in high and medium spatial resolution data). Moreover, new algorithms are needed for shadow correction, especially given the advent of new satellite images. [ABSTRACT FROM AUTHOR]

In general, topographic shadow may reduce performance of forest mapping over mountainous regions in remotely sensed images. In this paper, information in shadow was synthesized by using two filling techniques, namely, roifill and imfill, in order to improve the accuracy of forest mapping over mountainous regions. These two methods were applied to Landsat Enhanced Thematic Mapper (ETM +) multispectral image from Dong Yang County, Zhejiang Province, China. The performance of these methods was compared with two conventional techniques, including cosine correction and multisource classification. The results showed that by applying filling approaches, average overall accuracy of classification was improved by 14 percent. However, through conventional methods this value increased only by 9 percent. The results also revealed that estimated forest area on the basis of shadow- corrected images by 'roifill' technique was much closer to the survey data compared to traditional algorithms. Apart from this finding, our finding indicated that topographic shadow was an accentuated problem in medium resolution images such as Landsat ETM+ over mountainous regions. [ABSTRACT FROM AUTHOR]

Impervious surfaces play an important role in urban planning and sustainable environmental management. High-spatial-resolution (HSR) images containing pure pixels have significant potential for the detailed delineation of land surfaces. However, due to high intraclass variability and low interclass distance, the mapping and monitoring of impervious surfaces in complex town–rural areas using HSR images remains a challenge. The fully convolutional network (FCN) model, a variant of convolution neural networks (CNNs), recently achieved state-of-the-art performance in HSR image classification applications. However, due to the inherent nature of FCN processing, it is challenging for an FCN to precisely capture the detailed information of classification targets. To solve this problem, we propose an object-based deep CNN framework that integrates object-based image analysis (OBIA) with deep CNNs to accurately extract and estimate impervious surfaces. Specifically, we also adopted two widely used transfer learning technologies to expedite the training of deep CNNs. Finally, we compare our approach with conventional OBIA classification and state-of-the-art FCN-based methods, such as FCN-8s and the U-Net methods. Both of these FCN-based methods are well designed for pixel-wise classification applications and have achieved great success. Our results show that the proposed approach effectively identified impervious surfaces, with 93.9% overall accuracy. Compared with the existing methods, i.e., OBIA, FCN-8s and U-Net methods, it shows that our method achieves obviously improvement in accuracy. Our findings also suggest that the classification performance of our proposed method is related to training strategy, indicating that significantly higher accuracy can be achieved through transfer learning by fine-tuning rather than feature extraction. Our approach for the automatic extraction and mapping of impervious surfaces also lays a solid foundation for intelligent monitoring and the management of land use and land cover. [ABSTRACT FROM AUTHOR]