Interferometric Coherence Data Research Articles

Sentinel-1 interferometric coherence as a vegetation index for agriculture

In this study, the use of Sentinel-1 interferometric coherence data as a tool for crop monitoring has been explored. For this purpose, time series of images acquired by Sentinel-1 and 2 spanning 2017 have been analysed. The study site is an agricultural area in Sevilla, Spain, where 16 different crop species were cultivated during that year. The time series of 6-day repeat-pass coherence measured at each polarimetric channel (VV and VH), as well as their difference, have been compared to the NDVI and to the backscattering ratio (VH/VV) and other indices based on backscatter. The contribution of different decorrelation sources and the effect of the bias from the space-averaged sample coherence magnitude estimation have been evaluated. Likewise, the usage of 12 days as temporal baseline was tested. The study has been carried for three different orbits, characterised by different incidence angles and acquisition times. All results support using coherence as a measure for monitoring the crop growing season, as it shows good correlations with the NDVI (R2>0.7), and its temporal evolution fits well the main phenological stages of the crops. Although each crop shows its own evolution, the performance of coherence as a vegetation index is high for most of them. VV is generally more correlated with the NDVI than VH. For crop types characterised by low plant density, this difference decreases, with VH even showing higher correlation values in some cases. For a few crop types, such as rice, the backscattering ratio outperforms the coherence in following the growth stages of the plants. Since both coherence and backscattering are directly computed from the radar images, they could be used as complementary sources of information for this purpose. Notably, the measured coherence performs well without the need of compensating the thermal noise decorrelation or the bias due to the finite equivalent number of looks.

Estimation of Crop Yield From Combined Optical and SAR Imagery Using Gaussian Kernel Regression

The synthetic aperture radar (SAR) interferometric coherence can complement optical data for the estimation of crop growth parameters, but it has not been yet investigated for predicting crop yield. Many studies have used machine-learning methods, such as neural networks, random forest, and Gaussian process regression, to estimate crop yield from remotely sensed data. However, their performance depends on the amount of available ground truth data. This study proposed Gaussian kernel regression for rice yield estimation from optical and SAR imagery using a limited amount of ground truth data. The main objective was to investigate the synergetic use of Sentinel-2 vegetation indices and Sentinel-1 interferometric coherence data through Gaussian kernel regression for estimating rice grain yield. The prediction accuracy was assessed using in situ measured yield data collected in 2019 and 2020 over Xinghua county in Jiangsu Province, China. In all cases, Gaussian kernel regression outperformed the probabilistic Gaussian regression and Bayesian linear inference. With the independently used optical and SAR data, a better prediction accuracy was achieved with the optical red edge difference vegetation index (RDVI1) ( r2 = 0.65, RMSE = 0.61 t/ha) than with the interferometric coherence ( r2 = 0.52 and RMSE = 0.79 t/ha).The highest prediction accuracy can be achieved by combining RDVI1 with interferometric coherence at the heading stage ( r2 = 0.81 and RMSE = 0.55 t/ha). The results suggest the value of synergy between satellite interferometric coherence and optical indices for crop yield mapping with Gaussian kernel regression.

Automated detection of rock glaciers using deep learning and object-based image analysis

Rock glaciers are an important component of the cryosphere and are one of the most visible manifestations of permafrost. While the significance of rock glacier contribution to streamflow remains uncertain, the contribution is likely to be important for certain parts of the world. High-resolution remote sensing data has permitted the creation of rock glacier inventories for large regions. However, due to the spectral similarity between rock glaciers and the surrounding material, the creation of such inventories is typically conducted based on manual interpretation, which is both time consuming and subjective. Here, we present a novel method that combines deep learning (convolutional neural networks or CNNs) and object-based image analysis (OBIA) into one workflow based on freely available Sentinel-2 optical imagery (10 m spatial resolution), Sentinel-1 interferometric coherence data, and a digital elevation model (DEM). CNNs identify recurring patterns and textures and produce a prediction raster, or heatmap where each pixel indicates the probability that it belongs to a certain class (i.e. rock glacier) or not. By using OBIA we can segment the datasets and classify objects based on their heatmap value as well as morphological and spatial characteristics. We analysed two distinct catchments, the La Laguna catchment in the Chilean semi-arid Andes and the Poiqu catchment in the central Himalaya. In total, our method mapped 108 of the 120 rock glaciers across both catchments with a mean overestimation of 28%. Individual rock glacier polygons howevercontained false positives that are texturally similar, such as debris-flows, avalanche deposits, or fluvial material causing the user's accuracy to be moderate (63.9–68.9%) even if the producer's accuracy was higher (75.0–75.4%). We repeated our method on very-high-resolution Pléiades satellite imagery and a corresponding DEM (at 2 m resolution) for a subset of the Poiqu catchment to ascertain what difference image resolution makes. We found that working at a higher spatial resolution has little influence on the producer's accuracy (an increase of 1.0%), however the rock glaciers delineated were mapped with a greater user's accuracy (increase by 9.1% to 72.0%). By running all the processing within an object-based environment it was possible to both generate the deep learning heatmap and perform post-processing through image segmentation and object reshaping. Given the difficulties in differentiating rock glaciers using image spectra, deep learning combined with OBIA offers a promising method for automating the process of mapping rock glaciers over regional scales and lead to a reduction in the workload required in creating inventories.

Classificação do uso e cobertura da terra do Distrito Federal, Brasil a partir de dado de coerência interferométrica (Use and land cover classification of the Federal District, Brazil from interferometric coherence data)

O objetivo do presente trabalho foi avaliar o potencial das imagens de radar do satélite Sentinel 1A (banda C) para discriminar classes representativas de uso e cobertura da terra do Distrito Federal, Brasil. As imagens de coerência interferométrica utilizadas nesse estudo foram obtidas a partir de pares de imagens Single Look Complex (SLC) de junho e julho de 2018. Foi gerada uma composição colorida RGB a partir de imagens de coerência, intensidade de retroespalhamento e razão de retroespalhamento. Essa imagen foi classificada pelos métodos supervisionados Support Vector Machine (SVM) e Random Forest (RF). A validação temática foi realizada por matrizes de confusão, índice Kappa e exatidão global. Nesse contexto, foram investigadas cinco classes temáticas (água, área urbana, vegetação nativa, pastagem e agricultura). O classificador RF obteve melhor desempenho classificatório (Kappa= 0,68 e exatidão global = 79,1%) que o classificador SVM (Kappa= 0,64 e exatidão global = 75,7%). A coerência mostrou-se eficiente principalmente na identificação de corpos d’água e da área urbana. Os resultados foram satisfatórios para a classificação de uso e cobertura da terra do Distrito Federal, no entanto, houve confusão entre algumas classes e erros de comissão na classe área urbana em ambas as classificações. A b s t r a c tThe aim of this study was to evaluate the potential of Sentinel 1A (C band) satellite radar images to discriminate representative classes of land use and land cover in the Federal District, Brazil. The interferometric coherence images used in this study were obtained from pairs of Single Look Complex images (SLC) of June and July of 2018. An RGB color composition was generated from images of coherence, backscattering intensity and backscattering ratio. This image was classified by the Supervised Methods Support Vector Machine (SVM) and Random Forest (RF). Thematic validation was performed by matrices of confusion, Kappa index and overall accuracy. In this context, five thematic classes (water, urban area, native vegetation, pasture and agriculture) were investigated. The RF classifier obtained a better classificatory performance (Kappa= 0.68 and overall accuracy = 79.1%) than the SVM classifier (Kappa= 0.64 and overall accuracy = 75.7%). The coherence was shown to be efficient mainly in the identification of water and the urban area. The results were satisfactory for the classification of use and land cover of the Federal District, however, there was confusion between some classes and errors of commission in the urban area class in both classifications.Keywords: use and land cover, Interferometry, interferometric coherence, Sentinel 1A, radar, Remote Sensing

Mapping Forest Height from TanDEM-X Interferometric Coherence Data in Northwest Territories, Canada

In this paper, we demonstrate the feasibility of using TanDEM-X (TX) interferometric coherence data for mapping forest height with 25-m pixels across a study area near Fort Simpson, Northwest Territories (NWT), Canada. Our simplified RVOG model locally estimates forest height by combining an optimized estimation of TX interferometric coherence amplitude with the 20-m resolution Canadian Digital Elevation Model (CDEM) accounting for local slope variations. The initial map of TX height estimates provided R2 values of 0.78 and 0.88, mean errors (ME) of 1.66 m and 1.90 m, and root-mean-square errors (RMSE) of 2.7 m and 2.9 m when compared to independent height estimates derived from field plots and airborne LiDAR, respectively. We corrected the bias of TX height estimates using two variants of a LiDAR-based linear model. An application of three cover-specific linear adjustments provided the final TX height map with absolute ME ≤0.05 m and RMSE ≤2.09 m. The approach was tailored to poorly inventoried northern boreal regions through the use of archived TX data, the CDEM, a land cover map and airborne LiDAR transects. Our encouraging results support the perspective of wall-to-wall mapping of forest height across northern boreal forests in the NWT and beyond.

Fusion of Multi-Temporal Interferometric Coherence and Optical Image Data for the 2016 Kumamoto Earthquake Damage Assessment

Earthquakes are one of the most devastating types of natural disasters, and happen with little to no warning. This study combined Landsat-8 and interferometric ALOS-2 coherence data without training area techniques by classifying the remote sensing ratios of specific features for damage assessment. Waterbodies and highly vegetated areas were extracted by the modified normalized difference water index (MNDWI) and normalized difference vegetation index (NDVI), respectively, from after-earthquake images in order to improve the accuracy of damage maps. Urban areas were classified from pre-event interferometric coherence data. The affected areas from the earthquake were detected with the normalized difference (ND) between the pre- and co-event interferometric coherence. The results presented three damage types; namely, damage to buildings caused by ground motion, liquefaction, and landslides. The overall accuracy (94%) of the confusion matrix was excellent. Results for urban areas were divided into three damage levels (e.g., none–slight, slight–heavy, heavy–destructive) at a high (90%) overall accuracy level. Moreover, data on buildings damaged by liquefaction and landslides were in good agreement with field survey information. Overall, this study illustrates an effective damage assessment mapping approach that can support post-earthquake management activities for future events, especially in areas where geographical data are sparse.

Modeling the spatial distribution of above-ground carbon in Mexican coniferous forests using remote sensing and a geostatistical approach

Forest conservation is considered an option for mitigating the effect of greenhouse gases on global climate, hence monitoring forest carbon pools at global and local levels is important. The present study explores the capability of remote-sensing variables (vegetation indices and textures derived from SPOT-5; backscattering coefficient and interferometric coherence of ALOS PALSAR images) for modeling the spatial distribution of above-ground biomass in the Environmental Conservation Zone of Mexico City. Correlation and spatial autocorrelation coefficients were used to select significant explanatory variables in fir and pine forests. The correlation for interferometric coherence in HV polarization was negative, with correlations coefficients r=−0.83 for the fir and r=−0.75 for the pine forests. Regression-kriging showed the least root mean square error among the spatial interpolation methods used, with 37.75tC/ha for fir forests and 29.15tC/ha for pine forests. The results showed that a hybrid geospatial method, based on interferometric coherence data and a regression-kriging interpolator, has good potential for estimating above-ground biomass carbon.

Monitoring a boreal wildfire using multi-temporal Radarsat-1 intensity and coherence images

Twenty-five C-band Radarsat-1 synthetic aperture radar (SAR) images acquired from the summer of 2002 to the summer of 2005 are used to map a 2003 boreal wildfire (B346) in the Yukon Flats National Wildlife Refuge, Alaska under conditions of near-persistent cloud cover. Our analysis is primarily based on the 15 SAR scenes acquired during arctic growing seasons. The Radarsat-1 intensity data are used to map the onset and progression of the fire, and interferometric coherence images are used to qualify burn severity and monitor post-fire recovery. We base our analysis of the fire on three test sites, two from within the fire and one unburned site. The B346 fire increased backscattered intensity values for the two burn study sites by approximately 5–6 dB and substantially reduced coherence from background levels of approximately 0.8 in unburned background forested areas to approximately 0.2 in the burned area. Using ancillary vegetation information from the National Land Cover Database (NLCD) and information on burn severity from Normalized Burn Ratio (NBR) data, we conclude that burn site 2 was more severely burned than burn site 1 and that C-band interferometric coherence data are useful for mapping landscape changes due to fire. Differences in burn severity and topography are determined to be the likely reasons for the observed differences in post-fire intensity and coherence trends between burn sites.

Mapping urban change in the UK using satellite radar interferometry

A multitemporal sequence of ERS interferometric coherence data acquired between 1993 and 1999 are utilised for automatically mapping urban change within South Wales, UK. Validation of the change map derived from the coherence data is performed using independent, multidate, digital survey data of the city of Cardiff, UK. All major building developments that have occurred within the study area are located. There is evidence to suggest that this approach is generalisable for a wide range of coherence data and to other regions with similar landscapes.