scholarly journals Deep Learning for Land Cover Change Detection

2020 ◽  
Vol 13 (1) ◽  
pp. 78
Author(s):  
Oliver Sefrin ◽  
Felix M. Riese ◽  
Sina Keller
Keyword(s):  
Deep Learning ◽  
Land Cover ◽  
Change Detection ◽  
Land Cover Change ◽  
Short Term Memory ◽  
Image Sequences ◽  
Land Cover Changes ◽  
Water Classification ◽  
Land Cover Change Detection ◽  
Sentinel 2

Land cover and its change are crucial for many environmental applications. This study focuses on the land cover classification and change detection with multitemporal and multispectral Sentinel-2 satellite data. To address the challenging land cover change detection task, we rely on two different deep learning architectures and selected pre-processing steps. For example, we define an excluded class and deal with temporal water shoreline changes in the pre-processing. We employ a fully convolutional neural network (FCN), and we combine the FCN with long short-term memory (LSTM) networks. The FCN can only handle monotemporal input data, while the FCN combined with LSTM can use sequential information (multitemporal). Besides, we provided fixed and variable sequences as training sequences for the combined FCN and LSTM approach. The former refers to using six defined satellite images, while the latter consists of image sequences from an extended training pool of ten images. Further, we propose measures for the robustness concerning the selection of Sentinel-2 image data as evaluation metrics. We can distinguish between actual land cover changes and misclassifications of the deep learning approaches with these metrics. According to the provided metrics, both multitemporal LSTM approaches outperform the monotemporal FCN approach, about 3 to 5 percentage points (p.p.). The LSTM approach trained on the variable sequences detects 3 p.p. more land cover changes than the LSTM approach trained on the fixed sequences. Besides, applying our selected pre-processing improves the water classification and avoids reducing the dataset effectively by 17.6%. The presented LSTM approaches can be modified to provide applicability for a variable number of image sequences since we published the code of the deep learning models. The Sentinel-2 data and the ground truth are also freely available.

scholarly journals Monitoring Land Cover Changes in the Tropics using Satellite Remote Sensing Data

2019 ◽  
Vol 8 (12) ◽  
pp. 3631-3636
Keyword(s):  
Remote Sensing ◽  
Land Cover ◽  
Change Detection ◽  
Land Cover Change ◽  
Satellite Data ◽  
Land Cover Changes ◽  
Classification Methods ◽  
Land Covers ◽  
The Tropics ◽  
Land Cover Change Detection

Changes in land cover are inevitable phenomena that occur in all parts of the world. Land cover changes can occur due to natural phenomena that include runoff, soil erosion and sedimentation besides man-made phenomena that include deforestation, urbanization and conversion of land covers to suit human needs. Several works on change detection have been carried out elsewhere, however there were lack of effort in analyzing the issues that affect the performance of existing change detection techniques. The study presented in this paper aims to detect changes of land covers by using remote sensing satellite data. The study involves detection of land cover changes using remote sensing techniques. This makes use satellite data taken at different times over a particular area of interest. The data has resolution of 30 m and records surface reflectance at approximately 0.4 to 0.7 micrometers wavelengths. The study area is located in Selangor, Malaysia and occupied with tropical land covers including coastal swamp water, sediment plumes, urban, industry, water, bare land, cleared land, oil palm, rubber and coconut. Initially, region of interests (ROI) were drawn on each of the land covers in order to extract the training pixels. Landsat satellite bands 1, 2, 3, 4, 5 and 7 were then used as the input for the three supervised classification methods namely Support Vector Machine (SVM), Maximum Likelihood (ML) and Neural Network (NN). Different sizes of training pixels were used as the input for the classification methods so that the performance can be better understood. The accuracy of the classifications was then assessed by analyzing the classifications with a set of reference pixels using a confusion matrix. The classification methods were then used to identify the conversion of land cover from year 2000 to 2005 within the study area. The outcomes of the land cover change detection were reported in terms quantitative and qualitative analyses. The study shows that SVM gives a more accurate and realistic land cover change detection compared to ML and NN mainly due to not being much influenced by the size of the training pixels. The findings of the study serve as important input for decision makers in managing natural resources and environment in the tropics systematically and efficiently.

Land cover change detection in Crete Island, Greece, using different combinations of biophysical indices in change vector analysis

2020 ◽  
Author(s):  
Christos Polykretis ◽  
Manolis G. Grillakis ◽  
Dimitrios D. Alexakis
Keyword(s):  
Land Cover ◽  
Change Detection ◽  
Land Cover Change ◽  
Vector Analysis ◽  
Land Cover Changes ◽  
Change Vector Analysis ◽  
Tasseled Cap ◽  
Crete Island ◽  
Change Vector ◽  
Land Cover Change Detection

<p>Land cover describes the general biophysical state of the surface providing also information about other aspects of the land, such as soils and water. Changes in land cover may have noticeable impact on the ecosystem biodiversity, water resources, climate system and socio-economic sectors. Therefore, the need for detecting these changes is more and more imperative, especially given the emergence of unbalances caused by natural and anthropogenic driving forces like climate change, intensive agriculture and wrong land management decisions. Land cover changes are mainly represented by changes in the biophysical properties of land surface. These properties can be measured by remote sensing-derived indices representing both the vegetation and soil conditions of a given region. In this research effort, by applying a change detection technique like change vector analysis (CVA), the relationship between the dynamic changes in such indices and land cover changes in Crete Island, Greece, was assessed and mapped for the time periods of 1999–2009 and 2009–2019. Vegetation indices such as normalized difference vegetation index (NDVI) and tasseled cap greenness (TCG), and soil indices such as albedo and tasseled cap brightness (TCB), were estimated by Landsat satellite images captured in 1999, 2009 and 2019. Based on two different index combinations (NDVI–albedo and TCG–TCB), CVA produced change results for each of the periods indicating the magnitude and type (direction) of changes, respectively. The most appropriate combination for land cover change detection in the study area was determined by an evaluation process resulting to the estimation of accuracy statistics (kappa index and overall accuracy). Although promising accuracy results were provided for both examined combinations, the change maps produced by the combination of NDVI–albedo were found to be more accurate.</p><p><em>Acknowledgments: This research has received funding from the Hellenic Foundation for Research and Innovation (HFRI) and the General Secretariat for Research and Technology Hellas (GSRT), under Agreement No 651.</em></p>

scholarly journals An Automatic and Operational Method for Land Cover Change Detection Using Spatiotemporal Analysis of MODIS Data: A Northern Ontario (Canada) Case Study

2021 ◽  
Vol 10 (5) ◽  
pp. 325
Author(s):  
Ima Ituen ◽  
Baoxin Hu
Keyword(s):  
Land Use ◽  
Land Cover ◽  
Change Detection ◽  
Land Cover Change ◽  
Basic Unit ◽  
Operational Method ◽  
Northern Ontario ◽  
Modis Data ◽  
Forest Fire Management ◽  
Land Cover Change Detection

Mapping and understanding the differences in land cover and land use over time is an essential component of decision-making in sectors such as resource management, urban planning, and forest fire management, as well as in tracking of the impacts of climate change. Existing methods sometimes pose a barrier to the effective monitoring of changes in land cover and land use, since a threshold parameter is often needed and determined based on trial and error. This study aimed to develop an automatic and operational method for change detection on a large scale from Moderate Resolution Imaging Spectroradiometer (MODIS) data. Super pixels were the basic unit of analysis instead of traditional individual pixels. T2 tests based on the feature vectors of temporal Normalized Difference Vegetation Index (NDVI) and land surface temperature were used for change detection. The developed method was applied to data over a predominantly vegetated area in northern Ontario, Canada spanning 120,000 sq. km from 2001–2016. The accuracies ranged between 78% and 88% for the NDVI-based test, from 74% to 86% for the LST-based test, and from 70% to 86% for the joint method compared with manual interpretation. Our proposed method for detecting land cover change provides a functional and viable alternative to existing methods of land cover change detection as it is reliable, repeatable, and free from uncertainty in establishing a threshold for change.

Sign in / Sign up

Export Citation Format

Share Document