Toward geomorphometry of plains - Country-level unsupervised classification of low-relief areas (Poland)

Dyba, Krzysztof; Jasiewicz, Jarosław

doi:10.1016/j.geomorph.2022.108373

Cited by 2 publications

(1 citation statement)

References 74 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Another study on supervised classification was conducted by Janowski et al 27 , in which the authors compared machine learning algorithms for classifying glacial landforms in the Lubawa Upland and Gardno–Leba Plain areas using ground truth dataset. In a previous article co-written by the present author, we clustered the landforms of the entire country using an unsupervised method 28 . This means that we made no prior assumptions about geomorphological units.…”

Section: Introductionmentioning

confidence: 99%

Explanation of the influence of geomorphometric variables on the landform classification based on selected areas in Poland

Dyba

2024

Sci Rep

Self Cite

View full text Add to dashboard Cite

In 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.

show abstract

Section: Introductionmentioning

confidence: 99%

Explanation of the influence of geomorphometric variables on the landform classification based on selected areas in Poland

Dyba

2024

Sci Rep

Self Cite

View full text Add to dashboard Cite

show abstract

LiDAR-Derived Relief Typology of Loess Patches (East Poland)

Gawrysiak

Kociuba

2023

Remote Sensing

View full text Add to dashboard Cite

The application of the automated analysis of remote sensing data processed into high-resolution digital terrain models (DTMs) using geographic information systems (GIS) tools provides a geomorphometric characterization of the diversity of the relief of loess patches over large areas. Herein, a quantitative classification of 79 loess patches with a total area of 3361 km2, distributed within the eastern part of the Polish Uplands belt, is carried out. A high-resolution 1 × 1 m DTM was generated from airborne laser scanning (ALS) data with densities ranging from 4 pts/m2 to 12 pts/m2, which was resampled to a resolution of 5 × 5 m for the study. This model was used to classify landform surfaces using the r.geomorphon (geomorphon algorithm) function in GRASS GIS software. By comparing the values in the neighborhood of each cell, a map of geomorphometric features (geomorphon) was obtained. The classification and typology of the relief of the studied loess patches was performed using GeoPAT2 (Geospatial Pattern Analysis Toolbox) software. Pattern signatures with a resolution of 100 × 100 m were extracted from the source data grid, and the similarity of geomorphological maps within the signatures was calculated and saved as a signature file and segment map using the spatial coincidence method. The distance matrix between each pair of segments was calculated, and the heterogeneity and isolation of the maps were generated. R system was used to classify the segments, which generated a dendrogram and a heat map based on the distance matrix. This made it possible to distinguish three main types and eight subtypes of relief. The morphometric approach used will contribute to a better understanding of the spatial variation in the relief of loess patches.

show abstract

Toward geomorphometry of plains - Country-level unsupervised classification of low-relief areas (Poland)

Cited by 2 publications

References 74 publications

Explanation of the influence of geomorphometric variables on the landform classification based on selected areas in Poland

Explanation of the influence of geomorphometric variables on the landform classification based on selected areas in Poland

LiDAR-Derived Relief Typology of Loess Patches (East Poland)

Contact Info

Product

Resources

About