|
|
Assessment of vegetation phenology using Sentinel-2 time series data
Danilchyk, Tatsiana ; Štych, Přemysl (advisor) ; Bohovic, Roman (referee)
This work aims to evaluate the detection of phenological phases of vegetation based on phenometric parameters according to archival Sentinel-2 data in the selected areas over the period 2018-2020. The first part of the work describes literature review of the relevant publications, which is followed by the description of the suggested methodology. Then, there are the results with the graphic material and description for each monitored site. In the final part of the work, advantages and disadvantages of the developed algorithm are discussed followed up by suggestions for future research and improvement. The developed algorithm consists of two parts. Masking out cloudy and cloud shadow pixels and generation on the vegetation indices time series is done in the GEE platform. The time series analysis and detection of SOS and EOS as well as statistical analysis are done in the R environment. The study areas of size 20 x 20 m represent different species of perennial vegetation across the Czech Republic. For the assessment of the phenophases detection are selected NDVI, RENDVI, NDRE, NDMI and MCARI. The Asymmetric Gaussian function and Double Logistic function are fitted to the time series of each vegetation season in each tested site, the phenology metrics are derived based on threshold or derivatives...
|
|
|
Multitemporal segmentation of landscape based on dynamic time warping
Suske, Daniel ; Štych, Přemysl (advisor) ; Kolář, Jan (referee)
The thesis provides new insight into the segmentation processing of freely available satellite time- series data. It is based on open source technologies and offers a new method of unsupervised segmentation, using time series data along with the method originally designed for speech recognition, Dynamic Time Warping (DTW). The thesis deals with the selection of suitable data with sufficient temporal, spatial, and spectral resolution. The most suitable solution in all respects is the Sentinel 2 constellation. The thesis is based on the assumption that the use of time series improves segmentation results compared to segmentation based on one image. The main objective of the work was to find or to create a segmentation method that will take into account not only the time- series data but also the method of calculating the closest relationships using the DTW method. Another goal was to create an algorithm that preprocesses the time series data of the chosen constellation with minimal user input, allows the user to gradually visualize intermediate results and offers a comparison to alternative, currently available segmentation methods used in the scientific community. The main part is the developed segmentation algorithm, whose results can be influenced by tuning the parameters. The results of the work...
|
|
|
Landcover classification of selected parts of Ethiopia based on machine learning method
Valchářová, Daniela ; Štych, Přemysl (advisor) ; Nedbal, Václav (referee)
Diploma thesis deals with the land cover classification in Sidama region of Ethiopia and 2 kebeles, Chancho and Dangora Morocho. High resolution Sentinel-2 and very high resolution PlanetScope satellite images are used. The development of the classification algorithm is done in the Google Earth Engine cloud based environment. Ten combinations of the 4 most important parameters of the Random Forest classification method are tested. The defined legend contains 8 land cover classes, namely built-up, crops, grassland/pasture, forest, scrubland, bareland, wetland and water body. The training dataset is collected in the field during the fall 2020. The classification results of the two data types at two scales are compared. The highest overall accuracy for land cover classification of Sidama region came out to be 84.1% and kappa index of 0.797, with Random Forest method parameters of 100 trees, 4 spectral bands entering each tree, value of 1 for leaf population and 40% of training data used for each tree. For the land cover classification of Chancho and Dangora Morocho kebele with the same method settings, the overall accuracy came out to be 66.00 and 73.73% and kappa index of 0.545 and 0.601. For the classification of Chancho kebele, a different combination of parameters (80, 3, 1, 0.4) worked out better...
|
|
|
Classification of land cover change in Ethiopia using Landsat and Sentinel-2 data
Zadražil, Filip ; Laštovička, Josef (advisor) ; Svoboda, Jan (referee)
This bachelor thesis is focused on the comparison of Random Forest (RF) and CART classifiers on the example of the Ethiopian region of Sidama. An analysis of land cover change between 2014 and 2020 was performed for this region. The cloud-based platform Google Earth Engine (GEE) was used for classifications. Supervised classifications were performed on images from Landsat 8 and Sentinel-2 missions, which were retrieved from the Earth Engine data catalogue. Data from in-situ measurement was used for training polygons, variability of input data over time was verified with Google Earth Pro. In the research part, the work deals with the methods and results of research that were conducted in a topic close to this work. In the empirical part, the work deals with the analysis of Landsat 8 and Sentinel-2 data. The temporal, spatial and spectral resolution were compared. In terms of temporal resolution, it has been shown that Sentinel-2 data allows up to three times more images for the same area thanks to the two satellites scanning in parallel. Spectral and spatial resolution of Sentinel-2 allows better observation of smaller and less distinguishable elements. The data was then used for land cover classifications using RF and CART classifiers in the cloud-based GEE environment. The RF classifier made it...
|
|
|
Evaluation of forest vegetation based on time series of remote sensing data
Laštovička, Josef
Příloha k disertační práci: Abstrakt v AJ (Mgr. Josef Laštovička) Abstract This dissertation thesis deals with the study of forest ecosystems in the central Europe with the time series of multispectral optical satellite data. These forest ecosystems have been influenced by biotic and abiotic disturbances for the last decade. The time series of the satellite data with high spatial resolution allow the detection and analysis of forest disturbances. This thesis is mainly focused primally on free available Landsat and Sentinel-2 data, these two data types were compared. From methods, the difference time series analyses / algorithms were used. The whole thesis can be divided into two main parts. The first one analyses usability of classifiers for detection of forest ecosystems with per-pixel and sub-pixel methods. Specifically, the Neural Network, the Support Vector Machine and the Maximum Likelihood per-pixel classifiers were used and compared for different types of data (for data with high spatial resolution - Landsat or Sentinel-2; very high spatial resolution - WorldView-2) and for classification of protected forest areas. The Support Vector Machine were selected as the most suitable method for forest classifications (with most accurate outputs) from the list of selected per-pixel classifiers. Also, Spectral...
|
|
|
Evaluation of forest vegetation based on time series of remote sensing data
Laštovička, Josef ; Štych, Přemysl (advisor) ; Brom, Jakub (referee) ; Bucha, Tomáš (referee)
Příloha k disertační práci: Abstrakt v AJ (Mgr. Josef Laštovička) Abstract This dissertation thesis deals with the study of forest ecosystems in the central Europe with the time series of multispectral optical satellite data. These forest ecosystems have been influenced by biotic and abiotic disturbances for the last decade. The time series of the satellite data with high spatial resolution allow the detection and analysis of forest disturbances. This thesis is mainly focused primally on free available Landsat and Sentinel-2 data, these two data types were compared. From methods, the difference time series analyses / algorithms were used. The whole thesis can be divided into two main parts. The first one analyses usability of classifiers for detection of forest ecosystems with per-pixel and sub-pixel methods. Specifically, the Neural Network, the Support Vector Machine and the Maximum Likelihood per-pixel classifiers were used and compared for different types of data (for data with high spatial resolution - Landsat or Sentinel-2; very high spatial resolution - WorldView-2) and for classification of protected forest areas. The Support Vector Machine were selected as the most suitable method for forest classifications (with most accurate outputs) from the list of selected per-pixel classifiers. Also, Spectral...
|
|
|
Archiv družicových dat CENIA
Kvapil, Jiří
Laboratoř dálkového průzkumu CENIA, české informační agentury životního prostředí v rámci své výzkumné činnosti vyvinula a pro širokou veřejnost zpřístupnila informační systém Archiv družicových dat. Aplikace je dostupná na webu Laboratoře dálkového průzkumu na https://dpz.cenia.cz/archiv. Data jsou využitelná pro nejrůznější aplikace v zemědělství, lesnictví, monitoringu životního prostředí, sledování vývoje území apod.
Fulltext:  PDF
|
|
| |
|
| |
|
|
Classification of selected agricultural crops from time series of Sentinel-2 and PlanetScope imagery in Kutnohorsko model area
Kuthan, Tomáš ; Kupková, Lucie (advisor) ; Potůčková, Markéta (referee)
Classification of selected agricultural crops from time series of Sentinel-2 and PlanetScope imagery in Kutnohorsko model area Abstract The thesis is focused on the analysis of spectral characteristics of selected agricultural crops druring agriculutural season from time series of Sentinel -2 (A and B) and PlanetScope sensors in the model area situated around the settlements of Kolín and Kutná Hora. It is based on the assumption that the use of multiple dates of image data acquired crops in different phenological phases of the crops allows better identification of crop species (Lu et al., 2004). The aim of the thesis was to analyse the characteristics of the seasonal course of spectral features of selected agricultural crops (sugar beet, spring barley, winter barley, maize, spring wheat, winter wheat, winter rape) and to determine the period of the year suitable for the differentiation of individual crops. Another aim of the thesis was to classify these crops in the model area from time series of two above-mentioned sensors and to compare the accuracy of the pixel and object-oriented classification approach for multitemporal composites and the accuracy for monotemporal image from the term when the individual crops are clearly distinguishable. The training and validation datasets and the classification mask...
|
guest ::
