ABSTRACT The rapid development of smart power marketing demands the building of very accurate and efficient predictive models to optimize customer segmentation, demand forecasting, and fraud detection. Traditional classification techniques are generally incapable of dealing with large‐scale, complex energy consumption data sets due to redundant attributes, class imbalance, and overfitting issues. In order to solve these problems, the paper suggests applying the Optimized C5.0 Decision Tree with Entropy‐Based Feature Selection for smart power trading to a higher level. The model proposed not only includes the traditional decision tree classification but also the entropy‐based feature selection, pruning, and advanced techniques in data preprocessing like normalization, imputation of missing data, and Synthetic Minority Over‐Sampling Technique for class imbalance and robustness of classification. The first step in the process of building a decision tree is to find out the entropy, which indicates how uncertain the dataset is. The Information Gain of each attribute is then calculated, and the one with the highest IG is taken as the root node. The data is partitioned repeatedly until a stop criterion is reached, which is where the condition to avoid the model being too complex and thus overfitting comes in. Along with the use of pruning techniques such as Reduced Error Pruning and Gini Index‐based feature scoring, the optimum classification accuracy is achieved. The results of the performance evaluation show that the Optimized C5.0 Decision Tree is better than classical classifiers.

Change detection (CD) is an essential aspect of urban planning and resource management. Deep learning (DL) has the potential to detect complex changes from massive data more automatically than traditional methods. However, current DL-based CD methods have limited abilities to efficiently extract bi-temporal features, emphasize real changes, detect weak edges, and ultimately integrate multiscale change information in detail. To address these challenges, we propose an edge and flow guided iterative convolutional neural network (EFICNN). Our network introduces an innovative and efficient iterative backbone (IB) as the feature extractor through pretrained fine-tuning. By embedding the IB into a Siamese architecture, it is possible to extract richer bi-temporal features from the input remote sensing (RS) images. On this basis, we design a 3-D difference enhancement module (3D-DEM) that utilizes a parameter-free 3-D attention mechanism to emphasize and connect bi-temporal differential and concatenated change features. Additionally, an edge-guided attention module (EGAM) is developed to enhance weak edges. This module combines reverse attention and edge attention based on the Laplacian pyramid to capture change backgrounds and high-frequency change edges, respectively. Inspired by the optical flow alignment between adjacent frames, we ultimately employ a flow-guided fusion module (FGFM) to promote the propagation of fine-grained features from deep to shallow and dynamically optimize the multiscale fusion process. Qualitative and quantitative experiments on three publicly available datasets demonstrate that our method outperforms 17 SOTA methods in terms of real change recognition and edge refinement. Our code is available athttps://github.com/LYT-work/EFICNN.

Polarimetric SAR (PolSAR) images have become a crucial data source for change detection in cloud cover and fog regions. Deep learning (DL) enables extract change features form PolSAR images automatically, overcoming the limitations of conventional methods. To address the computational complexity of global attention and inaccurate prediction of changes edges in DL, this study proposes a model integrating matrix factorization and adaptive propagation. The model adopts a Siamese structure, receiving bi-temporal PolSAR images as input. Then, a matrix factorization based attention module replaces traditional attention design, utilizing the low-rank matrix from matrix factorization as global information. Additionally, an adaptive propagation module is introduced to improve edge pixel sampling, calculating central pixel values based on fixed and variable offsets from a local neighborhood. Experimental results on Sentinel-1 dual-pol SAR dataset demonstrate the model's effectiveness in preserving edge integrity and enhancing land change detection accuracy.

The forest canopy is the most active and direct part of vegetation when it interacts with the external environment. Gap fraction (GF) is widely recognized as an accurate characterization of the spatial structure of the canopy and is decisive for the inversion of other important vegetation parameters like leaf area index (LAI). Terrestrial laser scanning (TLS) is a high-precision active remote sensing technology that can quickly acquire 3-D structure information of targets. In previous studies, the process of calculating GF using TLS data has often involved the downscaling of 3-D information, excessive computational complexity, or reduction in spatial resolution. In this work, we aimed to develop a computationally simple GF estimation model that can make full use of the vertical dimensional information of the point cloud data and accurately calculate the canopy GF. By building a quantitative relationship between the distances of point clouds at different locations from the projection surface and the laser point area, a laser point area (LPA) model connecting the point clouds to the leaves can be established and validated as effective. A comparison was made between the GF obtained by the LPA method and the results obtained by the digital hemisphere photography (DHP) and Monte Carlo (MC) methods. The results show that our method gives higher estimates than the DHP method and lower estimates than the MC method. The variation trend of the results of our method is generally in line with the other two methods ($R^{2}$= 0.86 with DHP,$R^{2}$= 0.60 with MC), indicating the validity of the LPA model developed in this letter.

Rice-crop intensity is the annual number of rice growth cycles in a field. Monitoring the intensity on a large scale is vital in evaluating grain production and its ecological impact. Synthetic Aperture Radar (SAR) has an all-weather imaging capability. However, the existing SAR-based rice-crop intensity mapping methods mostly focus on small regions due to the diversity of rice backscatter patterns, the inefficiency of the time-series feature extraction, and the unavailability of rice phenological information on a large scale. In this study, a harmonic-based method is proposed to identify the essential backscatter periodicities. It also suppresses short-term disturbance in time-series Sentinel-1 SAR data without setting filtering windows or assuming profile shapes. The method detects backscatter troughs, eliminating the requirement for point-by-point traversal mathematical operations. Annual temperature profiles are derived from time-series ERA5-Land data to identify troughs related to rice growth cycles under various agro-climatic conditions. Then, the single (135,537 km2), double (19,036 km2), and triple (259 km2) rice-crop intensities covering the entire Southern China in 2020 are mapped in a 10m resolution, without relying on region-specific prior phenological information. The method achieves an overall accuracy of 82.26%, and can potentially support the continental or global mapping task.

Canopy gaps affect the spatial distribution of radiation in the canopy. The estimation of gap fraction (GF) is important for the study of leaf area index (LAI). Terrestrial laser scanning (TLS) has been widely used for retrieving canopy structure parameters through massive high-resolution spatial samples in the form of 3-D point cloud data sets. Monte Carlo simulations can be used to obtain approximate solutions to quantitative problems through a large number of random sampling while avoiding complicated mathematical calculations. Monte Carlo simulations are suitable for analyzing TLS data and could help overcome resolution reductions inherent to current point cloud processing approaches. However, few studies have applied the Monte Carlo method to capture the canopy structure features implicit in high-resolution point clouds. This letter proposes a method for estimating the GF based on Monte Carlo simulations. A large number of randomly simulated laser beams were emitted to identify the canopy and gaps according to the discrimination distance, and the results were compared with the GFs derived from digital hemispherical photography (DHP). In general, the proposed TLS method estimated smaller GFs than the DHP method since DHP was vulnerable to exposure conditions and complex canopy structure. The GF estimated by the two methods is consistent ($R^{2} =0.7522$), indicating the effectiveness of the Monte Carlo method.

Microwave remote sensing is an important substitute for rice growth monitoring in cloudy area. Fields identification is the most basic task for tracking rice crop cultivation. A series of sophisticated classification methods such as machine learning techniques have been applied to mapping crop planting area. However, these methods are weak in interpreting the physical mechanism of identifying crop fields, and hardly illustrate the backscattering difference between rice fields and other ground objects. Besides, enormous training samples and computational expense are demanded to achieve high recognition accuracy, which limited their application to rapid and large-area agricultural monitoring. In this study, C-band quad-pol backscattering coefficients of different land cover types were extracted from RADARSAT-2 PolSAR (Polarimetric Synthetic Aperture Radar) data during the reproductive period of rice fields. The backscatter statistical characteristics of rice fields and other ground types in different polarimetric bands were visualized and analyzed using boxplots. Based on the scatter performance of the five interested ground objects, a simple decision tree scheme was proposed to identify rice paddy fields. The overall accuracy is 88.65% and Kappa coefficient is 0.77. Meanwhile, a land cover classification map of the study area was obtained though the hierarchical judgement of decision tree strategy.

Recent satellite missions have provided a variety of high spatial resolution, multi-spectral, and high-frequency revisit remote sensing datasets. The collaborative use of optical and synthetic aperture radar (SAR) imagery in remote sensing applications attracts considerable attention. The purpose of this paper is to investigate the contribution of both data to the rice planting area mapping. In particular, the red-edge band was introduced to construct a red-edge vegetation index based on Sentinel-2 data. C-band quad-pol Radarsat-2 data was also used. We finally used the random forest algorithm, collaborating with optical and radar data to map rice planting area. We found that the red-edge band and red-edge vegetation index can improve the classification accuracy compared to the classifier using NIR (near-infrared) band and NDVI (Normalized Difference Vegetation Index). The result shows that the jointly use of optical and radar data is feasible to map rice planting area. The overall accuracy, recall and F-measure are 0.9441, 0.9598 and 0.9680, respectively. Index Terms - red edge, SAR, rice mapping, random forest, Sentinel-2, Radarsat-2

Accurate mapping of paddy rice planting area is important for rice yield prediction. Polarimetric synthetic aperture radar (PolSAR) data is suitable for rice growth monitoring in cloudy and foggy weather. The potential of using quad-pol PolSAR parameters for rice area mapping needs exploration in detail. In this study, multi-temporal C-band quad-pol RADARSAT-2 datasets and random forest classification algorithm were used to identify the rice planting area in Meishan City, China in 2016. The backscatter coefficients, polarimetric variables and decomposition parameters were extracted and divided into groups as the inputs of random forest classifier. The decomposition parameters derived from single-day PolSAR data during the rice reproductive period were found the most practicable for rice area mapping, with the overall accuracy of 95.94% and kappa coefficient of 0.92.

Forest resources are of great significance in regulating climate, maintaining biodiversity, and providing ecological products. Accurate identification of tree species is the basis for research and utilization of forest resources. This study combined the characteristics of multi-source data, based on the AISA EAGLE II hyperspectral images and airborne LiDAR point clouds which were obtained in August, 2016. Point cloud characteristics, spectral and texture characteristics were extracted from both datasets. Then SVM was used to classify the main tree species of Genhe experimental area. The results showed that tree species classification accuracy can be improved by using airborne LiDAR and hyperspectral image features.

The boundary feature is of great significance in describing the object shape and constructing 3D model of object. Accurate extraction of boundary features is important to the visualization of the object. This paper presented an accurate extraction algorithm of the indoor boundary features based on point cloud data. The amount of the data was downsampled by the voxelgrid filter. The boundary features of the indoor were extracted by using the angle criterion based on the normal vector and the statistical filtering algorithm. The results show that the boundary features of the indoor can be accurately extracted by the proposed method.

The objective of this paper aims to propose a new individual tree segmentation algorithm using airborne light detection and ranging (A-LiDAR) point cloud according to the trend of tree crown shape without any priori-knowledges. First, the method of crown profile points extraction was used to remove redundant data to improve the computational efficiency and to make higher accuracy. Then a segmentation method based on trend discrimination was proposed, which used the vertical profile and spatial distribution characteristics of tree crown point cloud to segment individual trees. This algorithm was applied to six experimental plots and was appraised using recall, precision, and F-score, which reached 98%, 94% and 0.96, respectively. The results indicated that this algorithm can accurately segment individual trees with low computational cost.

Rice is one of the most important crops in the world. Rice usually grows in tropical and subtropical areas where are usually rainy and cloudy. Under this weather condition, optical remote sensing has a great limitation, but the synthetic aperture radar (SAR) runs well. This paper utilized C band multi-temporal Radarsat-2 data, which is acquired in 2016, to extract quad-polarized backscattering coefficients, Cloude-Pottier and Freeman-Durden decomposition parameters for establishing classification features. Random forest (RF) algorithm was employed as the classifier. The Overall Accuracy (OA) is 82.9% when only the data on May 15thwas used. After adding the data on July 26th, the OA increased to 88.8%. Then, the variable importance was estimated by using out-of-bag (OOB) data. The results indicate that using RF would effectively identify the ground objects in the rice planting area.

The tree skeleton is one of the important parameters for building 3D model of the tree. The accurate extraction of tree skeleton is of great significance for tree visualization. This paper proposed an effective method for extracting tree skeleton of individual trees. The complexity of canopy structure was reduced by slicing. The leaf and wood components were separated by combining classification and segmentation methods. L1-median algorithm was used to extract the tree skeleton points accurately. The results show that the method can accurately extract the tree skeleton from terrestrial LiDAR data. The extracted skeleton had a good conformance with the point cloud of the tree in morphology.

Fraction of absorbed photosynthetically active radiation (FPAR) refers to the ratio of photosynthetically active radiation (PAR) absorbed by the green part of the vegetation canopy to the total PAR. Accurate and quantitative monitoring of rice FPAR is of great significance for cultivation management and yield estimation. High spatial and temporal resolution optical images, and all-time all-weather radar images are widely used in crop monitoring. In this study, a rice FPAR synergistic inversion model was established using a combination of optical and radar remote sensing dataset. The results suggest that the estimated FPAR is more approximate to the measured values, compared with the commonly used MODIS FPAR product. The FPAR spatiotemporal distribution maps are generated based on the synergistic inversion model.

The leaf area density (LAD) plays an important role in describing the vertical canopy structure. Light Detection and Ranging (LiDAR) is an active remote-sensing technology that has already been applied to canopy measurements. In this paper, the vertical profiles of the LAD of three different size plots (15 x 15 m, 10 x 10m, and 5 x 5 m) of forest canopy were estimated and compared based on a voxel-based model using airborne LiDAR data and terrestrial LiDAR data, respectively. The LAD profiles retrieved from airborne LiDAR data were different from those obtained by terrestrial LiDAR data. The height of the maximum LAD estimated from the airborne LiDAR data was significantly higher than that estimated from the terrestrial LiDAR data. In addition, the middle and lower parts of the forest canopy LAD were underestimated by airborne LiDAR while the upper part of the forest canopy LAD were underestimated by terrestrial LiDAR compared with the actual canopy vertical structure.

The target decomposition of polarimetric synthetic aperture radar (PolSAR) data was widely used to study the electromagnetic scattering properties of rice paddies. Many studies used eigenvectors-based decomposition like Cloude-Pottier decomposition to monitor rice growth but the potential of model-based decomposition like Freeman-Durden decomposition for rice phenology monitoring was not fully explored. Therefore, a rice phenology inversion method was presented in this paper using Freeman-Durden decomposition parameters acquired from multi-temporal, quad-polarization RADARSAT-2 data over rice fields in Meishan, China. A decision tree approach was employed for the retrieval algorithm. Results show that joint use of logarithmic power scattered by the surface scattering component and double-bounce scattering component of the covariance matrix provides a good performance and obtains a total accuracy of 89.7% when four phenological stages (i.e., transplanting, vegetative, reproductive, and maturation) are considered. Most of the errors occurred in the vegetative and reproductive stages. The corresponding errors were both 16.7%.

The objective of this paper was to develop a new algorithm to segment individual trees directly by using the three-dimensional space characteristic of airborne light detection and ranging point cloud data. The local maximum method was used in the initial segmentation and the error identification tree exclusion. On the basis of the point cloud spatial distribution of individual trees and the adjacent relationship with the other trees, a point cloud clustering method was developed to decide the points belonging to the individual trees. This algorithm was tested by 6 forest plots in the Genhe forestry reserve. The results showed that this algorithm could segment individual trees quickly and accurately, and the overall accuracy of this algorithm was 96.3%.