Detecting Land Abandonment in Łódź Voivodeship Using Convolutional Neural Networks

Stanisław Krysiak,Elżbieta Papińska,Mikołaj Koziarkiewicz,Maciej Adamiak,Anna Majchrowska

doi:10.3390/land9030082

Abstract

The wide availability of multispectral satellite imagery through projects such as Landsat and Sentinel, combined with the introduction of deep learning in general and Convolutional Neural Networks (CNNs) in particular, has allowed for the rapid and effective analysis of multiple classes of problems pertaining to land coverage. Taking advantage of the two phenomena, we propose a machine learning model for the classification of land abandonment. We designed a Convolutional Neural Network architecture that outputs a classification probability for the presence of land abandonment in a given 15–25 ha grid element by using multispectral imaging data obtained through Sentinel Hub. For both the training and validation of the model, we used imagery of the Łódź Voivodeship in central Poland. The main source of truth was a 2009 orthophoto study available from the WMS (Web Map Service) of the Geoportal site. The model achieved 0.855 auc (area under curve), 0.47 loss, and 0.78 accuracy for the test dataset. Using the classification results and the Getis–Ord Gi* statistic, we prepared a map of cold- and hotspots with individual areas that exceed 50 km2. This thresholded heatmap allowed for an analysis of contributing factors for both low and intense land abandonment, demonstrating that common trends are identifiable through the interpretation of the classification results of the chosen model. We additionally performed a comparative field study on two selected cold- and hotspots. The study, along with the high-accuracy results of the model’s validation, confirms that CNN-type models are an effective tool for the automatic detection of land abandonment.

Highlights

The rural landscape of the Łódź Voivodeship (Figure 1), as everywhere else in Poland, underwent massive structural and functional transformations because of the political upheaval of 1989–2000, followed by the country’s integration with the European Union [1,2]
Deep learning in general and Convolutional Neural Network (CNN) architecture in particular have proven revolutionary in essentially all fields that can be assisted by computer vision, and the subfield of land use classification proves to be no different
The final heatmap is the result of further postprocessing, thresholding, in order to solely render cold- and hotspots (Figure 8), displaying clusters of low and high probability of land abandonment according to the Getis–Ord Gi* statistic

Summary

Introduction

The rural landscape of the Łódź Voivodeship (Figure 1), as everywhere else in Poland, underwent massive structural and functional transformations because of the political upheaval of 1989–2000, followed by the country’s integration with the European Union [1,2]. This restructuring process was beset with contradictory phenomena. High-quality land was subjected to more intensive agriculture, resulting in more modern farming methods and the expansion of cultivated areas. Mid- and poor-quality land experienced the emergence of substantial areas subject to agricultural abandonment [3]. Reasons for the reduction in agricultural activity include the progressive urbanization of rural areas (most prominently in suburban zones), declining returns on agricultural production

Objectives

Methods

Results

Discussion

Conclusion