scholarly journals TRADITIONAL AND MODERN METHODS OF SATELLITE IMAGES PROCESSING FOR OPERATIONAL MAPPING OF FOREST COVER DISTUBANCES

2020 ◽  
Vol 25 (3) ◽  
pp. 201-213
Author(s):  
Andrey V. Tarasov ◽  
Keyword(s):  
Machine Learning ◽  
Forest Management ◽  
Forest Cover ◽  
Forest Disturbance ◽  
Remote Sensing Data ◽  
Machine Learning Algorithms ◽  
Forest Disturbances ◽  
Detection Algorithms ◽  
Vegetation Indexes ◽  
Images Processing

Real-time mapping of forest disturbances is important for forest management. Detection of forest stands damaged by natural or human-induced factors allows making immediate necessary management decisions. To implement such a management strategy, it is necessary to use the methods of operational mapping. With the advent of the Earth remote sensing data (RSD), which have high spatial and temporal resolution (Planet Scope and Sentinel-2), it becomes possible to implement modern operational mapping methods for forest management operations (particularly, forest disturbance detection). Since the monitoring area and the number of images sharply increases, the need for automated image processing methods also rises. This paper provides an overview of “traditional methods” for identifying forest cover disturbances (vegetation indexes, Tasseled Cap, multiband and single band change detection etc), their basis, limitations, and experience of their application in Russia and in the world. Instead, algorithm based on machine learning methods and their classification are presented. Benefits and limitations of both groups of forest disturbances detection algorithms are noted. In addition, it was found out that there is limited experience of application of machine learning algorithms for RSD processing and such kind of research is relevant.

scholarly journals Mapping Allochemical Limestone Formations in Hazara, Pakistan Using Google Cloud Architecture: Application of Machine-Learning Algorithms on Multispectral Data

2021 ◽  
Vol 10 (2) ◽  
pp. 58
Author(s):  
Muhammad Fawad Akbar Khan ◽  
Khan Muhammad ◽  
Shahid Bashir ◽  
Shahab Ud Din ◽  
Muhammad Hanif
Keyword(s):  
Machine Learning ◽  
Remote Sensing ◽  
Learning Algorithms ◽  
Remote Sensing Data ◽  
Kappa Coefficient ◽  
Machine Learning Algorithms ◽  
Landsat 8 ◽  
Sensing Data ◽  
Fossiliferous Limestone

Low-resolution Geological Survey of Pakistan (GSP) maps surrounding the region of interest show oolitic and fossiliferous limestone occurrences correspondingly in Samanasuk, Lockhart, and Margalla hill formations in the Hazara division, Pakistan. Machine-learning algorithms (MLAs) have been rarely applied to multispectral remote sensing data for differentiating between limestone formations formed due to different depositional environments, such as oolitic or fossiliferous. Unlike the previous studies that mostly report lithological classification of rock types having different chemical compositions by the MLAs, this paper aimed to investigate MLAs’ potential for mapping subclasses within the same lithology, i.e., limestone. Additionally, selecting appropriate data labels, training algorithms, hyperparameters, and remote sensing data sources were also investigated while applying these MLAs. In this paper, first, oolitic (Samanasuk), fossiliferous (Lockhart and Margalla) limestone-bearing formations along with the adjoining Hazara formation were mapped using random forest (RF), support vector machine (SVM), classification and regression tree (CART), and naïve Bayes (NB) MLAs. The RF algorithm reported the best accuracy of 83.28% and a Kappa coefficient of 0.78. To further improve the targeted allochemical limestone formation map, annotation labels were generated by the fusion of maps obtained from principal component analysis (PCA), decorrelation stretching (DS), X-means clustering applied to ASTER-L1T, Landsat-8, and Sentinel-2 datasets. These labels were used to train and validate SVM, CART, NB, and RF MLAs to obtain a binary classification map of limestone occurrences in the Hazara division, Pakistan using the Google Earth Engine (GEE) platform. The classification of Landsat-8 data by CART reported 99.63% accuracy, with a Kappa coefficient of 0.99, and was in good agreement with the field validation. This binary limestone map was further classified into oolitic (Samanasuk) and fossiliferous (Lockhart and Margalla) formations by all the four MLAs; in this case, RF surpassed all the other algorithms with an improved accuracy of 96.36%. This improvement can be attributed to better annotation, resulting in a binary limestone classification map, which formed a mask for improved classification of oolitic and fossiliferous limestone in the area.

scholarly journals Comparison of Machine Learning Algorithms for Wildland-Urban Interface Fuelbreak Planning Integrating ALS and UAV-Borne LiDAR Data and Multispectral Images

Drones ◽  
2020 ◽  
Vol 4 (2) ◽  
pp. 21 ◽  
Author(s):  
Francisco Rodríguez-Puerta ◽  
Rafael Alonso Ponce ◽  
Fernando Pérez-Rodríguez ◽  
Beatriz Águeda ◽  
Saray Martín-García ◽  
...  
Keyword(s):  
Machine Learning ◽  
Remote Sensing ◽  
Random Forest ◽  
Learning Algorithms ◽  
Remote Sensing Data ◽  
Machine Learning Algorithms ◽  
Data Sources ◽  
Lidar Data ◽  
Sensing Data ◽  
Sentinel 2

Controlling vegetation fuels around human settlements is a crucial strategy for reducing fire severity in forests, buildings and infrastructure, as well as protecting human lives. Each country has its own regulations in this respect, but they all have in common that by reducing fuel load, we in turn reduce the intensity and severity of the fire. The use of Unmanned Aerial Vehicles (UAV)-acquired data combined with other passive and active remote sensing data has the greatest performance to planning Wildland-Urban Interface (WUI) fuelbreak through machine learning algorithms. Nine remote sensing data sources (active and passive) and four supervised classification algorithms (Random Forest, Linear and Radial Support Vector Machine and Artificial Neural Networks) were tested to classify five fuel-area types. We used very high-density Light Detection and Ranging (LiDAR) data acquired by UAV (154 returns·m−2 and ortho-mosaic of 5-cm pixel), multispectral data from the satellites Pleiades-1B and Sentinel-2, and low-density LiDAR data acquired by Airborne Laser Scanning (ALS) (0.5 returns·m−2, ortho-mosaic of 25 cm pixels). Through the Variable Selection Using Random Forest (VSURF) procedure, a pre-selection of final variables was carried out to train the model. The four algorithms were compared, and it was concluded that the differences among them in overall accuracy (OA) on training datasets were negligible. Although the highest accuracy in the training step was obtained in SVML (OA=94.46%) and in testing in ANN (OA=91.91%), Random Forest was considered to be the most reliable algorithm, since it produced more consistent predictions due to the smaller differences between training and testing performance. Using a combination of Sentinel-2 and the two LiDAR data (UAV and ALS), Random Forest obtained an OA of 90.66% in training and of 91.80% in testing datasets. The differences in accuracy between the data sources used are much greater than between algorithms. LiDAR growth metrics calculated using point clouds in different dates and multispectral information from different seasons of the year are the most important variables in the classification. Our results support the essential role of UAVs in fuelbreak planning and management and thus, in the prevention of forest fires.

Forest Cover Types Classification Based on Online Machine Learning on Distributed Cloud Computing Platforms of Storm and SAMOA

2014 ◽  
Vol 955-959 ◽  
pp. 3803-3812
Author(s):  
Guang Di Li ◽  
Guo Yin Wang ◽  
Xue Rui Zhang ◽  
Wei Hui Deng ◽  
Fan Zhang
Keyword(s):  
Machine Learning ◽  
Forest Cover ◽  
Stream Processing ◽  
Processing Technique ◽  
Machine Learning Algorithms ◽  
Learning Tasks ◽  
Hoeffding Tree ◽  
Computing Platforms ◽  
Processing Platform ◽  
Forest Cover Types

Storm is the most popular realtime stream processing platform, which can be used to deal with online machine learning. Similar to how Hadoop provides a set of general primitives for doing batch processing, Storm provides a set of general primitives for doing realtime computation. SAMOA includes distributed algorithms for the most common machine learning tasks like Mahout for Hadoop. SAMOA is both a platform and a library. In this paper, Forest cover types, a large benchmaking dataset available at the UCI KDD Archive is used as the data stream source. Vertical Hoeffding Tree, a parallelizing streaming decision tree induction for distributed enviroment, which is incorporated in SAMOA API is applied on Storm platform. This study compared stream prcessing technique for predicting forest cover types from cartographic variables with traditional classic machine learning algorithms applied on this dataset. The test then train method used in this system is totally different from the traditional train then test. The results of the stream processing technique indicated that it’s output is aymptotically nearly identical to that of a conventional learner, but the model derived from this system is totally scalable, real-time, capable of dealing with evolving streams and insensitive to stream ordering.

scholarly journals Predictive performance of linear regression models in estimation of Artemia salina abundance using field and remote sensing data

2021 ◽  
pp. 88-95
Author(s):  
D. Krivoguz ◽  
◽  
R. Borovskaya ◽  
Keyword(s):  
Machine Learning ◽  
Linear Regression ◽  
Spatial Data ◽  
Regression Models ◽  
Remote Sensing Data ◽  
Artemia Salina ◽  
Machine Learning Algorithms ◽  
Linear Regression Models ◽  
Light Spectrum ◽  
Fourth Degree

This research has been aimed at finding the possibilities for application of the linear regression models, as a part of the machine learning methods, in visual representation of the spatial patterns of Artemia salina distribution in the Southern Sivash. Development of such models allows for estimation of A. salina biomass in water bodies with high accuracy. For investigation of maximum absorption levels in different parts of the light spectrum, spectral signatures at all the monitoring stations have been compared with the satellite data, and the analysis of the absorption spectra for astaxanthin and hemoglobin has been conducted with a spectrophotometer. As a result, Sentinel-2 satellite looks very promising as a key spatial data provider that can be of major help in increasing the frequency of A. salina monitoring in the Southern Sivash. The linear regression models, fitted by the third and the fourth degree polynomials, have shown satisfactory results, suitable for their subsequent use in fisheries. On the other hand, it should be noted that these models are slightly prone to overfitting, which to some extent can distort further forecasts feeding upon the new data. In turn, linear regression models fitted by a polynomial of the first degree show less accurate results, but their advantages include the lack of tendency to overfit. It is also worth noting that small-sized datasets within the scope of this investigation do not appear to be problematic, and simple machine learning algorithms can provide good accuracy results, which are suitable for further application in this field.

Forest Disturbance Mapping with Sentinel-2 Time Series in Austria

2021 ◽  
Author(s):  
Markus Löw ◽  
Tatjana Koukal
Keyword(s):  
Time Series ◽  
Forest Management ◽  
Time Series Analysis ◽  
Bark Beetle ◽  
Management Practices ◽  
Forest Ecosystems ◽  
Forest Disturbance ◽  
Spatial Scales ◽  
Forest Disturbances ◽  
Sentinel 2

<p>Worldwide, forests provide natural resources and ecosystem services. However, forest ecosystems are threatened by increasing forest disturbance dynamics, caused by direct human activities or by altering environmental conditions. It is decisive to reconstruct and trace the intra- to transannual dynamics of forest ecosystems. Therefore, the monitoring of large and small scale vegetation changes such as those caused by natural events (e.g., pest infestation, higher mortality due to altering site conditions) or forest management practices (e.g., thinning or selective timber extraction) becomes more and more crucial. National to local forest authorities and other stakeholders request detailed area-wide maps that delineate forest disturbance dynamics at various spatial scales.</p><p>We developed a time series analysis (TSA) framework that comprises data download, data management, image preprocessing and an advanced but flexible TSA. We use dense Sentinel-2 time series and a dynamic Savitzky–Golay-filtering approach to model robust but sensitive phenology courses. Deviations from the phenology models are used to derive detailed spatiotemporal information on forest disturbances. In a first case study, we apply the TSA to map forest disturbances directly or indirectly linked to recurring bark beetle infestation in Northern Austria.</p><p>In addition to spatiotemporal disturbance maps, we produce zonal statistics on different spatial scales that provide aggregated information on the extent of forest disturbances between 2018 and 2019. The outcomes are (a) area-wide consistent data of individual phenology models and deduced phenology metrics for Austrian forests and (b) operational forest disturbance maps, useful to investigate and monitor forest disturbances, for example to facilitate sustainable forest management.</p><p>At a forest stand level, we reconstruct the origin date of forest disturbances (FDD – Forest Disturbance Date). Theses FDD outputs show the spatiotemporal patterns and the development of damages and indicate that most dynamics are caused by recurring and spreading bark beetle infestation. The validation results based on field data confirm a high detection rate and show that the derived temporal information is reliable. In total, 23400 hectares, i.e., on average 2.8% of the forest area in the study area, are found to be affected by forest disturbance. The zonal statistic maps point out hotspots of significant forest disturbances, where adequate forest management measures are highly needed. Furthermore, this study highlights the TSA’s potential to also depict and monitor minor human impacts on forests, such as thinning, selective timber extraction or other moderate forest management practices.</p><p><strong>Keywords:  </strong><em>forest disturbance; forest monitoring; bark beetle infestation; forest management; time series analysis; phenology modelling; remote sensing; satellite imagery; Sentinel-2</em></p>

scholarly journals Landslide Susceptibility Assessment Using an AutoML Framework

2021 ◽  
Vol 18 (20) ◽  
pp. 10971
Author(s):  
Adrián G. Bruzón ◽  
Patricia Arrogante-Funes ◽  
Fátima Arrogante-Funes ◽  
Fidel Martín-González ◽  
Carlos J. Novillo ◽  
...  
Keyword(s):  
Machine Learning ◽  
Landslide Susceptibility ◽  
Spatial Databases ◽  
Learning Algorithms ◽  
Area Under The Curve ◽  
Remote Sensing Data ◽  
Machine Learning Algorithms ◽  
Sources Of Information ◽  
Natural Catastrophes ◽  
Input Variables

The risks associated with landslides are increasing the personal losses and material damages in more and more areas of the world. These natural disasters are related to geological and extreme meteorological phenomena (e.g., earthquakes, hurricanes) occurring in regions that have already suffered similar previous natural catastrophes. Therefore, to effectively mitigate the landslide risks, new methodologies must better identify and understand all these landslide hazards through proper management. Within these methodologies, those based on assessing the landslide susceptibility increase the predictability of the areas where one of these disasters is most likely to occur. In the last years, much research has used machine learning algorithms to assess susceptibility using different sources of information, such as remote sensing data, spatial databases, or geological catalogues. This study presents the first attempt to develop a methodology based on an automatic machine learning (AutoML) framework. These frameworks are intended to facilitate the development of machine learning models, with the aim to enable researchers focus on data analysis. The area to test/validate this study is the center and southern region of Guerrero (Mexico), where we compare the performance of 16 machine learning algorithms. The best result achieved is the extra trees with an area under the curve (AUC) of 0.983. This methodology yields better results than other similar methods because using an AutoML framework allows to focus on the treatment of the data, to better understand input variables and to acquire greater knowledge about the processes involved in the landslides.

Sign in / Sign up

Export Citation Format

Share Document