Genomic prediction modeling of soybean biomass using UAV‐based remote sensing and longitudinal model parameters

AbstractIn order to monitor the high-level landslides frequently occurring in Jinsha River area of Southwest China, and protect the lives and property safety of people in mountainous areas, the data of satellite remote sensing images are combined with various factors inducing landslides and transformed into landslide influence factors, which provides data basis for the establishment of landslide detection model. Then, based on the deep belief networks (DBN) and convolutional neural network (CNN) algorithm, two landslide detection models DBN and convolutional neural-deep belief network (CDN) are established to monitor the high-level landslide in Jinsha River. The influence of the model parameters on the landslide detection results is analyzed, and the accuracy of DBN and CDN models in dealing with actual landslide problems is compared. The results show that when the number of neurons in the DBN is 100, the overall error is the minimum, and when the number of learning layers is 3, the classification error is the minimum. The detection accuracy of DBN and CDN is 97.56% and 97.63%, respectively, which indicates that both DBN and CDN models are feasible in dealing with landslides from remote sensing images. This exploration provides a reference for the study of high-level landslide disasters in Jinsha River.

Download Full-text

Boundary-Assisted Learning for Building Extraction from Optical Remote Sensing Imagery

Remote Sensing ◽

10.3390/rs13040760 ◽

2021 ◽

Vol 13 (4) ◽

pp. 760

Author(s):

Sheng He ◽

Wanshou Jiang

Keyword(s):

Remote Sensing ◽

Receptive Fields ◽

Morphological Characteristics ◽

Learning Task ◽

Aerial Image ◽

Model Parameters ◽

Optical Remote Sensing ◽

Building Extraction ◽

Convolutional Network ◽

Remote Sensing Imagery

Deep learning methods have been shown to significantly improve the performance of building extraction from optical remote sensing imagery. However, keeping the morphological characteristics, especially the boundaries, is still a challenge that requires further study. In this paper, we propose a novel fully convolutional network (FCN) for accurately extracting buildings, in which a boundary learning task is embedded to help maintain the boundaries of buildings. Specifically, in the training phase, our framework simultaneously learns the extraction of buildings and boundary detection and only outputs extraction results while testing. In addition, we introduce spatial variation fusion (SVF) to establish an association between the two tasks, thus coupling them and making them share the latent semantics and interact with each other. On the other hand, we utilize separable convolution with a larger kernel to enlarge the receptive fields while reducing the number of model parameters and adopt the convolutional block attention module (CBAM) to boost the network. The proposed framework was extensively evaluated on the WHU Building Dataset and the Inria Aerial Image Labeling Dataset. The experiments demonstrate that our method achieves state-of-the-art performance on building extraction. With the assistance of boundary learning, the boundary maintenance of buildings is ameliorated.

Download Full-text

Unsupervised labelling of remote sensing images based on force field clustering

Journal of Intelligent & Fuzzy Systems ◽

10.3233/jifs-210802 ◽

2021 ◽

pp. 1-14

Author(s):

Zhenggang Wang ◽

Jin Jin

Keyword(s):

Remote Sensing ◽

Force Field ◽

Image Data ◽

Model Parameters ◽

Remote Sensing Images ◽

Initial Cluster ◽

Data Points ◽

Global Optimal ◽

Density Force

Remote sensing image segmentation provides technical support for decision making in many areas of environmental resource management. But, the quality of the remote sensing images obtained from different channels can vary considerably, and manually labeling a mass amount of image data is too expensive and Inefficiently. In this paper, we propose a point density force field clustering (PDFC) process. According to the spectral information from different ground objects, remote sensing superpixel points are divided into core and edge data points. The differences in the densities of core data points are used to form the local peak. The center of the initial cluster can be determined by the weighted density and position of the local peak. An iterative nebular clustering process is used to obtain the result, and a proposed new objective function is used to optimize the model parameters automatically to obtain the global optimal clustering solution. The proposed algorithm can cluster the area of different ground objects in remote sensing images automatically, and these categories are then labeled by humans simply.

Download Full-text

Evaluating and improving the Community Land Model's sensitivity to land cover

Biogeosciences ◽

10.5194/bg-15-4731-2018 ◽

2018 ◽

Vol 15 (15) ◽

pp. 4731-4757 ◽

Cited By ~ 13

Author(s):

Ronny Meier ◽

Edouard L. Davin ◽

Quentin Lejeune ◽

Mathias Hauser ◽

Yan Li ◽

...

Keyword(s):

Remote Sensing ◽

Land Cover ◽

Land Surface ◽

Soil Column ◽

Plant Functional Type ◽

Model Parameters ◽

Functional Type ◽

Land Surface Models ◽

Surface Models ◽

Open Land

Abstract. Modeling studies have shown the importance of biogeophysical effects of deforestation on local climate conditions but have also highlighted the lack of agreement across different models. Recently, remote-sensing observations have been used to assess the contrast in albedo, evapotranspiration (ET), and land surface temperature (LST) between forest and nearby open land on a global scale. These observations provide an unprecedented opportunity to evaluate the ability of land surface models to simulate the biogeophysical effects of forests. Here, we evaluate the representation of the difference of forest minus open land (i.e., grassland and cropland) in albedo, ET, and LST in the Community Land Model version 4.5 (CLM4.5) using various remote-sensing and in situ data sources. To extract the local sensitivity to land cover, we analyze plant functional type level output from global CLM4.5 simulations, using a model configuration that attributes a separate soil column to each plant functional type. Using the separated soil column configuration, CLM4.5 is able to realistically reproduce the biogeophysical contrast between forest and open land in terms of albedo, daily mean LST, and daily maximum LST, while the effect on daily minimum LST is not well captured by the model. Furthermore, we identify that the ET contrast between forests and open land is underestimated in CLM4.5 compared to observation-based products and even reversed in sign for some regions, even when considering uncertainties in these products. We then show that these biases can be partly alleviated by modifying several model parameters, such as the root distribution, the formulation of plant water uptake, the light limitation of photosynthesis, and the maximum rate of carboxylation. Furthermore, the ET contrast between forest and open land needs to be better constrained by observations to foster convergence amongst different land surface models on the biogeophysical effects of forests. Overall, this study demonstrates the potential of comparing subgrid model output to local observations to improve current land surface models' ability to simulate land cover change effects, which is a promising approach to reduce uncertainties in future assessments of land use impacts on climate.

Download Full-text

Texture Image Segmentation Based on MRMRF in Contourlet Domain

Advanced Materials Research ◽

10.4028/www.scientific.net/amr.532-533.732 ◽

2012 ◽

Vol 532-533 ◽

pp. 732-737

Author(s):

Xi Jie Wang ◽

Xiao Fan Zhao

Keyword(s):

Remote Sensing ◽

Image Segmentation ◽

Field Model ◽

Model Parameters ◽

Texture Image ◽

Remote Sensing Images ◽

Unsupervised Segmentation ◽

Markov Random Field Model ◽

Markov Random ◽

Texture Image Segmentation

This paper presents a new multi-resolution Markov random field model in Contourlet domain for unsupervised texture image segmentation. In order to make full use of the merits of Contourlet transformation, we introduce the taditional MRMRF model into Contourlet domain, in a manner of variable interation between two components in the tradtional MRMRF model. Using this method, the new model can automatically estimate model parameters and produce accurate unsupervised segmentation results. The results obtained on synthetic texture images and remote sensing images demonstrate that a better segmentation is achieved by our model than the traditional MRMRF model.

Download Full-text

Estimation of the HEC-HMS model parameters in data-scarce regions. Application to the Ouergha watershed (Sebou, Morocco)

E3S Web of Conferences ◽

10.1051/e3sconf/202131405002 ◽

2021 ◽

Vol 314 ◽

pp. 05002

Author(s):

Hasna Moumni ◽

Karima Sebari ◽

Laila Stour ◽

Abdellatif Ahbari

Keyword(s):

Remote Sensing ◽

Hydrological Model ◽

Model Parameters ◽

Flow Data ◽

Random Parameters ◽

Quality Of Data ◽

Initial Values ◽

Digital Elevation ◽

Elevation Model

The availability, accessibility and quality of data are significant obstacles to hydrological modelling. Estimating the initial values of the hydrological model´’ ’s parameters is a laborious and determining task requiring much attention. Geographic information systems (GIS) and spatial remote sensing are prometting tools for processing and collecting data. In this work, we use an innovative approach to estimate the HEC-HMS hydrological model parameters from the soil map of Africa (250m), the land use map GLC30, the depth to bedrock map, the digital elevation model and observed flow data. The estimation approach is applied to the Ouergha basin (Sebou, Morocco). The proposed approach’s interest is to feed the HEC-HMS hydrological model with initial values of parameters close to the study area reality instead of using random parameters.

Download Full-text

Determine the Stumpf 2003 Model Parameters for Multispectral Remote Sensing Shallow Water Bathymetry

Journal of Coastal Research ◽

10.2112/si102-007.1 ◽

2020 ◽

Vol 102 (sp1) ◽

Author(s):

Jinshan Zhu ◽

Peng Hu ◽

Lulu Zhao ◽

Lei Gao ◽

Jiawei Qi ◽

...

Keyword(s):

Remote Sensing ◽

Shallow Water ◽

Model Parameters ◽

Multispectral Remote Sensing

Download Full-text

Training Convolutional Neural Networks with Multi-Size Images and Triplet Loss for Remote Sensing Scene Classification

Sensors ◽

10.3390/s20041188 ◽

2020 ◽

Vol 20 (4) ◽

pp. 1188 ◽

Cited By ~ 10

Author(s):

Jianming Zhang ◽

Chaoquan Lu ◽

Jin Wang ◽

Xiao-Guang Yue ◽

Se-Jung Lim ◽

...

Keyword(s):

Remote Sensing ◽

Classification Accuracy ◽

Model Parameters ◽

Classification Algorithms ◽

Scene Classification ◽

Training Strategy ◽

Network Training ◽

Training Stage ◽

Triplet Loss ◽

And Training

Many remote sensing scene classification algorithms improve their classification accuracy by additional modules, which increases the parameters and computing overhead of the model at the inference stage. In this paper, we explore how to improve the classification accuracy of the model without adding modules at the inference stage. First, we propose a network training strategy of training with multi-size images. Then, we introduce more supervision information by triplet loss and design a branch for the triplet loss. In addition, dropout is introduced between the feature extractor and the classifier to avoid over-fitting. These modules only work at the training stage and will not bring about the increase in model parameters at the inference stage. We use Resnet18 as the baseline and add the three modules to the baseline. We perform experiments on three datasets: AID, NWPU-RESISC45, and OPTIMAL. Experimental results show that our model combined with the three modules is more competitive than many existing classification algorithms. In addition, ablation experiments on OPTIMAL show that dropout, triplet loss, and training with multi-size images improve the overall accuracy of the model on the test set by 0.53%, 0.38%, and 0.7%, respectively. The combination of the three modules improves the overall accuracy of the model by 1.61%. It can be seen that the three modules can improve the classification accuracy of the model without increasing model parameters at the inference stage, and training with multi-size images brings a greater gain in accuracy than the other two modules, but the combination of the three modules will be better.

Download Full-text

Improvement of Remote Sensing-Based Assessment of Defoliation of Pinus spp. Caused by Thaumetopoea Pityocampa Denis and Schiffermüller and Related Environmental Drivers in Southeastern Spain

Remote Sensing ◽

10.3390/rs11141736 ◽

2019 ◽

Vol 11 (14) ◽

pp. 1736 ◽

Cited By ~ 2

Author(s):

Pérez-Romero ◽

Navarro-Cerrillo ◽

Palacios-Rodriguez ◽

Acosta ◽

Mesas-Carrascosa

Keyword(s):

Remote Sensing ◽

Google Earth ◽

Environmental Data ◽

Environmental Drivers ◽

Series Data ◽

Model Parameters ◽

Summer Drought ◽

Southeastern Spain ◽

Pinus Species ◽

Pinus Spp

This study used Landsat temporal series to describe defoliation levels due to the Pine Processionary Moth (PPM) in Pinus forests of southeastern Andalusia (Spain), utilizing Google Earth Engine. A combination of remotely sensed data and field survey data was used to detect the defoliation levels of different Pinus spp. and the main environmental drivers of the defoliation due to the PPM. Four vegetation indexes were also calculated for remote sensing defoliation assessment, both inside the stand and in a 60-m buffer area. In the area of study, all Pinus species are affected by defoliation due to the PPM, with a cyclic behavior that has been increasing in frequency in recent years. Defoliation levels were practically equal for all species, with a high increase in defoliation levels 2 and 3 since 2014. The Moisture Stress Index (MSI) and Normalized Difference Infrared Index (NDII) exhibited similar overall (P < 0.001) accuracy in the assessment of defoliation due to the PPM. The synchronization of NDII-defoliation data had a similar pattern for all together and individual Pinus species, showing the ability of this index to adjust the model parameters based on the characteristics of specific defoliation levels. Using Landsat-based NDII-defoliation maps and interpolated environmental data, we have shown that the PPM defoliation in southeastern Spain is driven by the minimum temperature in February and the precipitation in June, March, September, and October. Therefore, the NDII-defoliation assessment seems to be a general index that can be applied to forests in other areas. The trends of NDII-defoliation related to environmental variables showed the importance of summer drought stress in the expansion of the PPM on Mediterranean Pinus species. Our results confirm the potential of Landsat time-series data in the assessment of PPM defoliation and the spatiotemporal patterns of the PPM; hence, these data are a powerful tool that can be used to develop a fully operational system for the monitoring of insect damage.

Download Full-text