Browsing by Author "Dyba, Krzysztof"
Now showing 1 - 6 of 6
Results Per Page
Sort Options
Item Evaluation of Methods for Estimating Lake Surface Water Temperature Using Landsat 8(2022-08-08) Dyba, Krzysztof; Ermida, Sofia; Ptak, Mariusz; Piekarczyk, Jan; Sojka, MariuszChanges in lake water temperature, observed with the greatest intensity during the last two decades, may significantly affect the functioning of these unique ecosystems. Currently, in situ studies in Poland are conducted only for 38 lakes using the single-point method. The aim of this study was to develop a method for remote sensing monitoring of lake water temperature in a spatio-temporal context based on Landsat 8 imagery. For this purpose, using data obtained for 28 lakes from the period 2013–2020, linear regression (LM) and random forest (RF) models were developed to estimate surface water temperature. In addition, analysis of Landsat Level-2 Surface Temperature Science Product (LST-L2) data provided by United States Geological Survey (USGS) and the National Aeronautics and Space Administration (NASA) was performed. The remaining 10 lakes not previously used in the model development stage were used to validate model performance. The results showed that the most accurate estimation is possible using the RF method for which RMSE = 1.83 °C and R^2 = 0.89, while RMSE = 3.68 °C and R^2 = 0.8 for the LST-L2 method. We found that LST-L2 contains a systematic error in the coastal zone, which can be corrected and eventually improve the quality of estimation. The satellite-based method makes it possible to determine water temperature for all lakes in Poland at different times and to understand the influence of climatic factors affecting temperature at the regional scale. On the other hand, spatial presentation of thermics within individual lakes enables understanding the influence of local factors and morphometric conditions.Item Explanation of the influence of geomorphometric variables on the landform classification based on selected areas in Poland(Nature Publishing Group, 2024-03-05) Dyba, KrzysztofIn recent years, automatic image classification methods have significantly progressed, notably black box algorithms such as machine learning and deep learning. Unfortunately, such efforts only focused on improving performance, rather than attempting to explain and interpret how classification models actually operate. This article compares three state-of-the-art algorithms incorporating random forests, gradient boosting and convolutional neural networks for geomorphological mapping. It also attempts to explain how the most effective classifier makes decisions by evaluating which of the geomorphometric variables are most important for automatic mapping and how they affect the classification results using one of the explainable artificial intelligence techniques, namely accumulated local effects (ALE). This method allows us to understand the relationship between predictors and the model’s outcome. For these purposes, eight sheets of the digital geomorphological map of Poland on the scale of 1:100,000 were used as the reference material. The classification results were validated using the holdout method and cross-validation for individual sheets representing different morphogenetic zones. The terrain elevation entropy, absolute elevation, aggregated median elevation and standard deviation of elevation had the greatest impact on the classification results among the 15 geomorphometric variables considered. The ALE analysis was conducted for the XGBoost classifier, which achieved the highest accuracy of 92.8%, ahead of Random Forests at 84% and LightGBM at 73.7% and U-Net at 59.8%. We conclude that automatic classification can support geomorphological mapping only if the geomorphological characteristics in the predicted area are similar to those in the training dataset. The ALE plots allow us to analyze the relationship between geomorphometric variables and landform membership, which helps clarify their role in the classification process.Item Interpretacja danych geoprzestrzennych przy użyciu wyjaśnialnych metod uczenia maszynowego(2024) Dyba, Krzysztof; Jasiewicz, Jarosław. Promotor; Interpretation of geospatial data using explainable machine learning methodsUczenie maszynowe niewątpliwie stało się wszechobecne w różnych dyscyplinach naukowych uwzględniając nauki o Ziemi i środowisku, rewolucjonizując przy tym sposób analizy i interpretacji danych geoprzestrzennych. Rozwój złożonych modeli typu czarnej skrzynki spowodował nowe wyzwania związane z przejrzystością ich działania oraz interpretowalnością wyników. Przedstawiona dysertacja analizuje przydatność wyjaśnialnych metod uczenia maszynowego do interpretacji danych geoprzestrzennych, odpowiadając na pytanie czy niniejsze metody mogą wspomagać proces interpretacji czynników prowadzących do uzyskania wyniku. Cel dysertacji został zrealizowany przez trzy eksperymenty badawcze związane z zastosowaniem: 1) analizy regresji do estymacji temperatury jezior; 2) klasyfikacji nadzorowanej do automatycznego kartowania form geomorfologicznych; 3) klasyfikacji nienadzorowanej do wyznaczenia i interpretacji typów powierzchni terenu. Na podstawie uzyskanych wyników stwierdzono, że zastosowane metody wyjaśniające są przydatne do interpretacji działania modeli typu czarnej skrzynki. Pozwoliły one na lepsze zrozumienie decyzji podejmowanych przez modele i ujawnienie relacji pomiędzy zmiennymi wyjaśniającymi a wynikiem modelu. Wykorzystane metody wyjaśniające wskazały, nie tylko, które cechy są istotne, ale przede wszystkim, w jaki sposób wpłynęły na wynik predykcji. Machine learning has undoubtedly become widespread in various scientific disciplines, including Earth and environmental sciences, revolutionizing the way of analyzing and interpreting geospatial data. The development of complex black-box models has brought new challenges related to the transparency of their operation and the interpretability of the results. The presented dissertation examines the usefulness of explainable machine learning methods for interpreting geospatial data, answering the question of whether these methods can support the process of interpreting factors leading to a result. The objective of the thesis was achieved through three research experiments involving the application of: 1) regression analysis for estimating the surface temperature of lakes; 2) supervised classification for automatic mapping of landforms; 3) unsupervised classification for determining and interpreting land surface types. Based on the obtained results, the applied explainability methods were found to be useful for interpreting the operation of black-box models. They allowed for a better understanding of the decisions made by the models and revealing the relationships between the explanatory variables and the model output. These methods indicated not only which features were significant, but more importantly, how they affected the prediction result.Item Reflectance spectroscopy and machine learning as a tool for the categorization of twin species based on the example of the Diachrysia genus(2022-02-17) Dyba, Krzysztof; Wąsala, Roman; Piekarczyk, Jan; Gabała, Elżbieta; Gawlak, Magdalena; Jasiewicz, Jarosław; Ratajkiewicz, HenrykIn our work we used noninvasive point reflectance spectroscopy in the range from 400 to 2100 nm coupled with machine learning to study scales on the brown and golden iridescent areas on the dorsal side of the forewing of Diachrysia chrysitis and D. stenochrysis. We used our approach to distinguish between these species of moths. The basis for the study was a statistically significant collection of 95 specimens identified based on morphological feature and gathered during 23 years in Poland. The numerical part of an experiment included two independent discriminant analyses: stochastic and deterministic. The more sensitive stochastic approach achieved average compliance with the species identification made by entomologists at the level of 99–100%. It demonstrated high stability against the different configurations of training and validation sets, hence strong predictors of Diachrysia siblings distinctiveness. Both methods resulted in the same small set of relevant features, where minimal fully discriminating subsets of wavelengths were three for glass scales on the golden area and four for the brown. The differences between species in scales primarily concern their major components and ultrastructure. In melanin-absent glass scales, this is mainly chitin configuration, while in melanin-present brown scales, melanin reveals as an additional factor.Item rgugik: Search and Retrieve Spatial Data from the Polish Head Office of Geodesy and Cartography in R(2021-03-16) Dyba, Krzysztof; Nowosad, JakubItem Toward geomorphometry of plains - Country-level unsupervised classification of low-relief areas (Poland)(2022) Dyba, Krzysztof; Jasiewicz, JarosławLow-relief areas are not fully the main subject of geomorphometric analyses. The development of the automatic classification of landforms mainly focuses on landforms related to the fluvial morphogenetic cycle. Thus, the morphogenetic diversity of the plains is not reflected in the existing classification systems. The area of Poland where the low relief area exceeds 80 % of the country's territory and results in various morphogenetic processes was selected for the analysis. The purpose of the analysis was recognition of the differentiation of surface types. The first step includes selecting appropriate morphogenetic variables, the second unsupervised classification using the Gaussian Mixture Model, and the third one encompassing the interpretation, namely the labeling process. Twenty Land Surface Types were distinguished, five belonging to uplands, and the remaining 15 types of plains were divided into four subgroups: rolling plains, dissection plains, smooth plains, and near-flat plains. Compared with other classification systems, terrain forms, morphogenetic strides, and physiographic division. The comparison showed a strong correspondence between the morphogenesis of the area and the inventory of surface types, and the high consistency of the Land Surface Types patterns within physiographic units.