Generating a series of fine spatial and temporal resolution land cover maps by fusing coarse spatial resolution remotely sensed images and fine spatial resolution land cover maps

Li, Xiaodong; Ling, Feng; Foody, Giles M.; Ge, Yong; Zhang, Yihang; Du, Youwei

doi:10.1016/j.rse.2017.05.011

Cited by 106 publications

(60 citation statements)

References 50 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, this requires abundant Landsat images (>4) (Xu et al, 2018a) that are not available in the humid tropical regions and may cause "false changes" and "inter-annual inconsistency" (Broich et al, 2011). Recently, a superresolution mapping method (Li et al, 2017;Qin et al, 2017;Xu et al, 2017) was used to reconstruct missing forest cover change 80 during 2011-2014 (Zhang et al, 2019) by fusing the PALSAR/PALSAR-2 and the MODIS normalized difference vegetation index (NDVI) with dense temporal resolution and phenological information. However, it is difficult to separate oil palm and natural forest with similar NDVI variation using such classification-based fusion.…”

Section: Introductionmentioning

confidence: 99%

Annual oil palm plantation maps in Malaysia and Indonesia from 2001 to 2016

Xu¹,

Yu²,

Li³

et al. 2019

Preprint

View full text Add to dashboard Cite

Increasing global demand of vegetable oils and biofuels results in significant oil palm expansion in Southeast Asia, predominately in Malaysia and Indonesia. The land conversion to oil palm plantations leads to deforestation, loss of biodiversity, and greenhouse gas emission over the past decades. Quantifying the consequences of oil palm expansion requires fine scale and frequently updated datasets of land cover dynamics. Previous studies focused on total changes for a multi-year interval without identifying the exact time of conversion, causing uncertainty in the timing of carbon emission estimates from 15 land cover change. Using Advanced Land Observing Satellite (ALOS) Phased Array Type L-band Synthetic Aperture Radar (PALSAR), ALOS-2 PALSAR-2 and Moderate Resolution Imaging Spectroradiometer (MODIS) datasets, we produced an Annual Oil Palm Area Dataset (AOPD) at 100-meter resolution in Malaysia and Indonesia from 2001 to 2016. We first mapped the oil palm extent using PALSAR/PALSAR-2 data for 2007-2010 and 2015-2016 and then applied a disturbance and recovery algorithm (BFAST) to detect land cover change time-points using MODIS data during the years without PALSAR data (2011-20 2014 and 2001-2006). The new oil palm land cover maps are assessed to have an accuracy of 86.61% in the mapping step (2007-2010 and 2015-2016). During the intervening years when MODIS data are used, 75.74% of the change detected time matched the timing of actual conversion using Google Earth and Landsat images. The AOPD dataset revealed spatiotemporal oil palm dynamics every year and shows that plantations expanded from 2.59 to 6.39 M ha and from 3.00 to 12.66 M ha in Malaysia and Indonesia, respectively (i.e., a net increase of 146.60% and 322.46%) between 2001 and 2016. The increasing 25 trends from our dataset are consistent with those from the national inventories, but slightly greater because of inclusion of smallholder oil palm plantations in our dataset. We highlight the capability of combining multiple resolution radar and optical satellite datasets in annual plantation mapping at large extent using image classification and statistical boundary-based change detection to achieve long time-series. The consistent characterization of oil palm dynamics can be further used in downstream applications. The annual oil palm plantation maps from 2001 to 2016 at 100 m resolution is published in the Tagged Image 30File Format with georeferencing information (GeoTIFF) at https://doi.org/10.5281/zenodo.3467071 .

show abstract

Section: Introductionmentioning

confidence: 99%

Annual oil palm plantation maps in Malaysia and Indonesia from 2001 to 2016

Xu¹,

Yu²,

Li³

et al. 2019

Preprint

View full text Add to dashboard Cite

show abstract

“…Recently, dictionary-based fusion methods dominate hyperspectral and multispectral image fusion, including spectral dictionary [4][5][6] and spatial dictionary based methods, but they cannot effectively utilize both spatial and spectral information equally. The Spatial-Temporal remotely sensed Images and land cover Maps Fusion Model (STIMFM) was proposed to produce land cover maps at both fine spatial and temporal resolutions using a series of coarse spatial resolution images together with a few fine spatial resolution land cover maps that pre-and post-date the series of coarse spatial resolution images [7]. The major limitation in these fusion techniques for HSI spatial-resolution enhancement is that an auxiliary co-registered image with a higher spatial resolution is required, which may be unavailable in practice.…”

Section: Introductionmentioning

confidence: 99%

Hyperspectral Image Spatial Super-Resolution via 3D Full Convolutional Neural Network

et al. 2017

View full text Add to dashboard Cite

Hyperspectral images are well-known for their fine spectral resolution to discriminate different materials. However, their spatial resolution is relatively low due to the trade-off in imaging sensor technologies, resulting in limitations in their applications. Inspired by recent achievements in convolutional neural network (CNN) based super-resolution (SR) for natural images, a novel three-dimensional full CNN (3D-FCNN) is constructed for spatial SR of hyperspectral images in this paper. Specifically, 3D convolution is used to exploit both the spatial context of neighboring pixels and spectral correlation of neighboring bands, such that spectral distortion when directly applying traditional CNN based SR algorithms to hyperspectral images in band-wise manners is alleviated. Furthermore, a sensor-specific mode is designed for the proposed 3D-FCNN such that none of the samples from the target scene are required for training. Fine-tuning by a small number of training samples from the target scene can further improve the performance of such a sensor-specific method. Extensive experimental results on four benchmark datasets from two well-known hyperspectral sensors, namely hyperspectral digital imagery collection experiment (HYDICE) and reflective optics system imaging spectrometer (ROSIS) sensors, demonstrate that our proposed 3D-FCNN outperforms several existing SR methods by ensuring higher quality both in reconstruction and spectral fidelity.

show abstract

“…In most cases, one data source has high temporal but low spatial resolution (HTLS), while another has low temporal but high spatial resolution (LTHS). After years of development, the research field of spatiotemporal data fusion has established certain theories and methods, and some of these methods have been applied in practical geoscience analysis with respectable accuracy [13][14][15]. As far as we have considered, the existing spatiotemporal fusion algorithms can be classified into three categories: (1) transformation-based; (2) reconstruction-based; and (3) learning-based [16].…”

Section: Introductionmentioning

confidence: 99%

Deriving High Spatiotemporal Remote Sensing Images Using Deep Convolutional Network

et al. 2018

View full text Add to dashboard Cite

Due to technical and budget limitations, there are inevitably some trade-offs in the design of remote sensing instruments, making it difficult to acquire high spatiotemporal resolution remote sensing images simultaneously. To address this problem, this paper proposes a new data fusion model named the deep convolutional spatiotemporal fusion network (DCSTFN), which makes full use of a convolutional neural network (CNN) to derive high spatiotemporal resolution images from remotely sensed images with high temporal but low spatial resolution (HTLS) and low temporal but high spatial resolution (LTHS). The DCSTFN model is composed of three major parts: the expansion of the HTLS images, the extraction of high frequency components from LTHS images, and the fusion of extracted features. The inputs of the proposed network include a pair of HTLS and LTHS reference images from a single day and another HTLS image on the prediction date. Convolution is used to extract key features from inputs, and deconvolution is employed to expand the size of HTLS images. The features extracted from HTLS and LTHS images are then fused with the aid of an equation that accounts for temporal ground coverage changes. The output image on the prediction day has the spatial resolution of LTHS and temporal resolution of HTLS. Overall, the DCSTFN model establishes a complex but direct non-linear mapping between the inputs and the output. Experiments with MODerate Resolution Imaging Spectroradiometer (MODIS) and Landsat Operational Land Imager (OLI) images show that the proposed CNN-based approach not only achieves state-of-the-art accuracy, but is also more robust than conventional spatiotemporal fusion algorithms. In addition, DCSTFN is a faster and less time-consuming method to perform the data fusion with the trained network, and can potentially be applied to the bulk processing of archived data.

show abstract

Generating a series of fine spatial and temporal resolution land cover maps by fusing coarse spatial resolution remotely sensed images and fine spatial resolution land cover maps

Cited by 106 publications

References 50 publications

Annual oil palm plantation maps in Malaysia and Indonesia from 2001 to 2016

Annual oil palm plantation maps in Malaysia and Indonesia from 2001 to 2016

Hyperspectral Image Spatial Super-Resolution via 3D Full Convolutional Neural Network

Deriving High Spatiotemporal Remote Sensing Images Using Deep Convolutional Network

Contact Info

Product

Resources

About