GF-1 WFV Images Research Articles

A Methodological Approach for Gap Filling of WFV Gaofen-1 Images from Spatial Autocorrelation and Enhanced Weighting

Clouds and cloud shadow cover cause missing data in some images captured by the Gaofen-1 Wide Field of View (GF-1 WFV) cameras, limiting the extraction and analysis of the image information and further applications. Therefore, this study proposes a methodology to fill GF-1 WFV images using the spatial autocorrelation and improved weighting (SAIW) method. Specifically, the search window size is adaptively determined using Getis-Ord Gi* as a metric. The spatial and spectral weights of the pixels are computed using the Chebyshev distance and spectral angle mapper to better filter the suitable similar pixels. Each missing pixel is predicted using linear regression with similar pixels on the reference image and the corresponding similar pixel located in the non-missing region of the cloudy image. Simulation experiments showed that the average correlation coefficient of the proposed method in this study is 0.966 in heterogeneous areas, 0.983 in homogeneous farmland, and 0.948 in complex urban areas. It suggests that SAIW can reduce the spread of errors in the gap-filling process to significantly improve the accuracy of the filling results and can produce satisfactory qualitative and quantitative fill results in a wide range of typical land cover types and has extensive application potential.

A Weakly Supervised Semantic Segmentation Framework for Medium-Resolution Forest Classification With Noisy Labels and GF-1 WFV Images

A Weakly Supervised Semantic Segmentation Framework for Medium-Resolution Forest Classification With Noisy Labels and GF-1 WFV Images

Inversion of Chlorophyll-a Concentrations in Chaohu Lake Based on GF-1 WFV Images

Inversion of Chlorophyll-a Concentrations in Chaohu Lake Based on GF-1 WFV Images

Extraction method of burned area using GF-1 WFV images and FY-3D MERSI fire-point products

Extraction method of burned area using GF-1 WFV images and FY-3D MERSI fire-point products

Practical On-Orbit Geometric Recalibration of GF-1 WFV Images Based on RPC Model

Practical On-Orbit Geometric Recalibration of GF-1 WFV Images Based on RPC Model

An effective atmospheric correction method for the wide swath of Chinese GF-1 and GF-6 WFV images on lands

Accurate land surface reflectance plays an important role in the accurate inversion of surface parameters, and atmospheric correction plays a decisive role in obtaining accurate reflectance. For GF-1 WFV and GF-6 WFV images, there are two major issues to be addressed, including the spectral differences between nadir with far off-nadir pixels and the spatial variability of atmospheric components for wide imaging. Therefore, this study focuses on these two issues using the Sensor Invariant Atmospheric Correction (SIAC) method. Our results reveal that the SIAC approach improves the correlation accuracy from 0.8868 to 0.9173 for GF-1 WFV image compared with Sentinel-2 reflectance, from 0.9530 to 0.9620 for GF-6 WFV image compared with the results using FLAASH model. For alleviating wide-swathed anisotropy, the directional imaging angle is calculated with the result ranging from 5.6450° to 33.7497°. Furthermore, the atmospheric components have been inversed pixel by pixel with obvious spatial variation. And the correlation of inversed aerosol optical thickness (AOT) and total column water vapor (TCWV) with a spatial resolution of 500 m TCWV with measured results of AERONET (AErosol RObotic NETwork) observation stations are 0.9175 and 0.4442, respectively. These results reveal that the atmospheric correction method works well, which is effective for the wide swath of Chinese GF-1 WFV and GF-6 WFV images on land.

A Comparison of Random Forest Algorithm-Based Forest Extraction with GF-1 WFV, Landsat 8 and Sentinel-2 Images

Forests are an essential part of the ecosystem and play an irreplaceable role in maintaining the balance of the ecosystem and protecting biodiversity. The monitoring of forest distribution plays an important role in the conservation and management of forests. This paper analyzes and compares the performance of imagery from GF-1 WFV, Landsat 8, and Sentinel-2 satellites with respect to forest/non-forest classification tasks using the random forest algorithm (RF). The results show that in the classification task of this paper, although the differences in classification accuracy among the three satellite datasets are not remarkable, the Sentinel-2 data have the highest accuracy, GF-1 WFV the second highest, and Landsat 8 the lowest. In addition, it was found that remotely sensed data of different processing levels show little influence on the classification accuracy with respect to the forest/non-forest classification task. However, the classification accuracy of the top of the atmosphere reflectance product was the most stable, and the vegetation index has a marginal effect on the distinction between forest and non-forest areas.

Cloud and Snow Identification Based on DeepLab V3+ and CRF Combined Model for GF-1 WFV Images

Cloud and snow identification in remote sensing images is critical for snow mapping and snow hydrology research. Aimed at the problem that the semantic segmentation model is prone to producing blurred boundaries, slicing traces and isolated small patches for cloud and snow identification in high-resolution remote sensing images, the feasibility of combining DeepLab v3+ and conditional random field (CRF) models for cloud and snow identification based on GF-1 WFV images is studied. For GF-1 WFV images, the model training and testing experiments under the conditions of different sample numbers, sample sizes and loss functions are compared. The results show that, firstly, when the number of samples is 10,000, the sample size is 256 × 256, and the loss function is the Focal function, the model accuracy is the optimal and the Mean Intersection over Union (MIoU) and the Mean Pixel Accuracy (MPA) reach 0.816 and 0.918, respectively. Secondly, after post-processing with the CRF model, the MIoU and the MPA are improved to 0.836 and 0.941, respectively, compared with those without post-processing. Moreover, the misclassifications such as blurred boundaries, slicing traces and isolated small patches are significantly reduced, which indicates that the combination of the DeepLab v3+ and CRF models has high accuracy and strong feasibility for cloud and snow identification in high-resolution remote sensing images. The conclusions can provide a reference for high-resolution snow mapping and hydrology applications using deep learning models.

Identification of forest vegetation types in southern China based on spatio-temporal fusion of GF-1 WFV and MODIS data

It is difficult to obtain long time series of high spatial resolution remote sensing images in southern China because of the complex terrain and frequent cloudy and rainy weather. In contrast, the spatio-temporal fusion can sychonorously obtain remote sensing data with high spatial-temporal resolution, which is beneficial to extract forest vegetation type information. With Xingguo County of Jiangxi Province as the study area, we fused the Landsat8 OLI and GF-1 WFV images with high spatial resolution with high temporal resolution of MODIS09 A1 image on the basis of enhanced spatial and temporal adaptive reflectance fusion model (ESTARFM), reconstructed the time series data of ESTARFM_Landsat8 EVI and ESTARFM_GF-1 EVI with 8 d step of enhanced vegetation index (EVI), obtained the phenology (PH) characteristics, and identified the forest vegetation types by using random forest classification model. The results showed that the correlation coefficients between the fusion data of ESTARFM_Landsat8 EVI and ESTARFM_GF-1 EVI and the real images were all greater than 0.7, and had good consistency in spatial distribution, which could be used to supplement the missing data with high spatial resolution. The extraction accuracy of random forest classification with different combination modes was EVI+PH>EVI>PH and the classification accuracy of fusion data GF-1 was higher than that of Landsat8. A total of 43 variables were selected as the optimal feature variables for classification. The overall accuracy and Kappa coefficient were 95.6% and 94.9%, respectively, including 37 sequential EVI values and 6 phenological feature information. The sequential EVI data contributed more to the identification of forest vegetation types, while the phenological feature information was beneficial to improve the classification accuracy. The ESTARFM fusion algorithm was suitable for GF-1 and MODIS data, which could solve the problem of insufficient long-term sequence of high spatial resolution images. The GF-1 temporal fusion images had high accuracy in the identification of forest vegetation types in southern China under complex terrain and frequent cloudy and rainy weather.

SRSe-Net: Super-Resolution-Based Semantic Segmentation Network for Green Tide Extraction

Due to the phenomenon of mixed pixels in low-resolution remote sensing images, the green tide spectral features with low Enteromorpha coverage are not obvious. Super-resolution technology based on deep learning can supplement more detailed information for subsequent semantic segmentation tasks. In this paper, a novel green tide extraction method for MODIS images based on super-resolution and a deep semantic segmentation network was proposed. Inspired by the idea of transfer learning, a super-resolution model (i.e., WDSR) is first pre-trained with high spatial resolution GF1-WFV images, and then the representations learned in the GF1-WFV image domain are transferred to the MODIS image domain. The improvement of remote sensing image resolution enables us to better distinguish the green tide patches from the surrounding seawater. As a result, a deep semantic segmentation network (SRSe-Net) suitable for large-scale green tide information extraction is proposed. The SRSe-Net introduced the dense connection mechanism on the basis of U-Net and replaces the convolution operations with dense blocks, which effectively obtained the detailed green tide boundary information by strengthening the propagation and reusing features. In addition, the SRSe-Net reducs the pooling layer and adds a bridge module in the final stage of the encoder. The experimental results show that a SRSe-Net can obtain more accurate segmentation results with fewer network parameters.

A novel automatic method on pseudo-invariant features extraction for enhancing the relative radiometric normalization of high-resolution images

ABSTRACT Relative radiometric normalization (RRN) is a critical preprocessing step that is widely applied to remote sensing data. Essential to RRN are the pseudo-invariant features (PIFs) – ground objects with invariant reflectance across the acquisition dates of multiple images. To meet the requirements of RRN for change detection in multi-temporal images, this paper proposed a novel automatic method for PIF extraction: first, we calculated spectral, spatial, and visual features for the images. Next, a set of change vectors were generated to quantify the features’ differences. Then, the initial PIF was obtained by synthesizing similar pixels from each change vector. Finally, an iterative linear regression was performed to generate the final PIFs. Furthermore, we employed the random sample consensus (RANSAC) algorithm to establish the RRN model for normalizing the target image. The methods and workflow of PIF extraction and the RRN model establishment were validated with two TripleSat-2 multi-spectral images with a spatial resolution of 3.2 metres. The applicability of the proposed PIF extraction method was further validated by Landsat-8 OLI, GF-1 WFV, Sentinel-2 MSI, and WorldView-2 images. The results showed that: (1) The proposed PIF extraction method can automatically obtain the PIFs with high precision. Based on the extracted PIFs, the RANSAC algorithm is capable of building the RRN model with a significant linear relationship. (2) The PIF extraction method is applicable to multi-spectral images with different spatial resolutions. (3) Compared with the multivariate alteration detection (MAD) and the iteratively reweighted MAD (IR-MAD), the proposed PIF extraction method achieved a better performance in selecting the pixels with smaller radiometric differences, which is helpful to reach a higher accuracy of the RRN model. (4) Differences in image brightness, spectral domain, and spatial domain are considered in the PIF extraction method with fewer parameters and high portability. (5) Spectral features are dominant to obtain ideal PIFs, and feature combinations can get the advantages to enhance the effectiveness of the PIFs.

Suitability Evaluation of Typical Drought Index in Soil Moisture Retrieval and Monitoring Based on Optical Images

As a basic agricultural parameter in the formation, transformation, and consumption of surface water resources, soil moisture has a very important influence on the vegetation growth, agricultural production, and healthy operation of regional ecosystems. The Aksu river basin is a typical semi-arid agricultural area which seasonally suffers from water shortage. Due to the lack of knowledge on soil moisture change, the water management and decision-making processes have been a difficult issue for local government. Therefore, soil moisture monitoring by remote sensing became a reasonable way to schedule crop irrigation and evaluate the irrigation efficiency. Compared to in situ measurements, the use of remote sensing for the monitoring of soil water content is convenient and can be repetitively applied over a large area. To verify the applicability of the typical drought index to the rapid acquisition of soil moisture in arid and semi-arid regions, this study simulated, compared, and validated the effectiveness of soil moisture inversion. GF-1 WFV images, Landsat 8 OLI images, and the measured soil moisture data were used to determine the Perpendicular Drought Index (PDI), the Modified Perpendicular Drought Index (MPDI), and the Vegetation Adjusted Perpendicular Drought Index (VAPDI). First, the determination coefficients of the correlation analyses on the PDI, MPDI, VAPDI, and measured soil moisture in the 0–10, 10–20, and 20–30 cm depth layers based on the GF-1 WFV and Landsat 8 OLI images were good. Notably, in the 0–10 cm depth layers, the average determination coefficient was 0.68; all models met the accuracy requirements of soil moisture inversion. Both indicated that the drought indices based on the Near Infrared (NIR)-Red spectral space derived from the optical remote sensing images are more sensitive to soil moisture near the surface layer; however, the accuracy of retrieving the soil moisture in deep layers was slightly lower in the study area. Second, in areas of vegetation coverage, MPDI and VAPDI had a higher inversion accuracy than PDI. To a certain extent, they overcame the influence of mixed pixels on the soil moisture spectral information. VAPDI modified by Perpendicular Vegetation Index (PVI) was not susceptible to vegetation saturation and, thus, had a higher inversion accuracy, which makes it performs better than MPDI’s in vegetated areas. Third, the spatial heterogeneity of the soil moisture retrieved by the GF-1 WFV and Landsat 8 OLI image were similar. However, the GF-1 WFV images were more sensitive to changes in the soil moisture, which reflected the actual soil moisture level covered by different vegetation. These results provide a practical reference for the dynamic monitoring of surface soil moisture, obtaining agricultural information and agricultural condition parameters in arid and semi-arid regions.

WE-NDBI-A new index for mapping urban built-up areas from GF-1 WFV images

ABSTRACTRecently, China has witnessed rapid development of high-resolution imaging satellites, with remote sensing images from satellites being applied extensively in social and economic sectors. The normalized difference built-up index (NDBI) was first used to extract urban built-up areas from TM/ETM+ image data. After successful launch of the Landsat-8 satellite, modifications have been made to the NDBI approach (BBIOLI approach) for mapping urban built-up areas from operational land imager (OLI) data with higher accuracy. However, experiments showed that the BBIOLI approach is not very suitable for Chinese high-resolution satellite image data. This study firstly analysed the spectral curves of ground objects generated by the GF-1 satellite and then proposes an innovative WE-NDBI approach to mapping urban built-up in the study area, i.e. large counties located in Xichang. The experimental results demonstrate an overall accuracy of 91.4%. To compare the WE-NDBI and BBIOLI approaches, the latter was used to extract urban built-up areas from OLI data for the same study area, and the experimental accuracy is 77.0%. It is concluded that the WE-NDBI approach is more suitable for mapping urban built-up areas.

The method for water body information extraction in complex environment using GF-1 WFV images

Water body is one of the most active and important earth resources, and which has a profound impact on the natural system and human society. In order to acquire surface water body information quickly, accurately and efficiently, the method of water body information extraction using remote sensing imagery has attracted the attention of many searchers. On the basis of sorting out relevant research results of water body information extraction using remote sensing imagery, this paper proposed the method of water body information extraction based on the tasseled cap transformation for complex environments such as shadow and dense vegetation. First, radiometric calibration and atmospheric correction were carried out for remote sensing images. Then, the tasseled cap transformation was performed to obtain the greenness component and wetness component. Finally, the model of water body information extraction based on the tasseled cap transformation was constructed, and the water body information was extracted. In a region of Hunan province, China, the experiment using GF-1 WFV remote sensing image shows that the extracted water body information has a clear boundary and complete shape, and the Kappa coefficient, overall accuracy and user accuracy are 0.89, 92.72%, and 88.04%, respectively.

Large-Scale Crop Mapping Based on Machine Learning and Parallel Computation with Grids

Large-scale crop mapping provides important information in agricultural applications. However, it is a challenging task due to the inconsistent availability of remote sensing data caused by the irregular time series and limited coverage of the images, together with the low spatial resolution of the classification results. In this study, we proposed a new efficient method based on grids to address the inconsistent availability of the high-medium resolution images for large-scale crop classification. First, we proposed a method to block the remote sensing data into grids to solve the problem of temporal inconsistency. Then, a parallel computing technique was introduced to improve the calculation efficiency on the grid scale. Experiments were designed to evaluate the applicability of this method for different high-medium spatial resolution remote sensing images and different machine learning algorithms and to compare the results with the widely used nonparallel method. The computational experiments showed that the proposed method was successful at identifying large-scale crop distribution using common high-medium resolution remote sensing images (GF-1 WFV images and Sentinel-2) and common machine learning classifiers (the random forest algorithm and support vector machine). Finally, we mapped the croplands in Heilongjiang Province in 2015, 2016, 2017, which used a random forest classifier with the time series GF-1 WFV images spectral features, the enhanced vegetation index (EVI) and normalized difference water index (NDWI). Ultimately, the accuracy was assessed using a confusion matrix. The results showed that the classification accuracy reached 88%, 82%, and 85% in 2015, 2016, and 2017, respectively. In addition, with the help of parallel computing, the calculation speed was significantly improved by at least seven-fold. This indicates that using the grid framework to block the data for classification is feasible for crop mapping in large areas and has great application potential in the future.

Harmonizing Multi-Source Remote Sensing Images for Summer Corn Growth Monitoring

Continuous monitoring of crop growth status using time-series remote sensing image is essential for crop management and yield prediction. The growing season of summer corn in the North China Plain with the period of rain and hot, which makes the acquisition of cloud-free satellite imagery very difficult. Therefore, we focused on developing image datasets with both a high temporal resolution and medium spatial resolution by harmonizing the time-series of MOD09GA Normalized Difference Vegetation Index (NDVI) images and 30-m-resolution GF-1 WFV images using the improved Kalman filter model. The harmonized images, GF-1 images, and Landsat 8 images were then combined and used to monitor the summer corn growth from 5th June to 6th October, 2014, in three counties of Hebei Province, China, in conjunction with meteorological data and MODIS Evapotranspiration Data Set. The prediction residuals ( Δ P R K ) in NDVI between the GF-1 observations and the harmonized images was in the range of −0.2 to 0.2 with Gauss distribution. Moreover, the obtained phenological curves manifested distinctive growth features for summer corn at field scales. Changes in NDVI over time were more effectively evaluated and represented corn growth trends, when considered in conjunction with meteorological data and MODIS Evapotranspiration Data Set. We observed that the NDVI of summer corn showed a process of first decreasing and then rising in the early growing stage and discuss how the temperature and moisture of the environment changed with the growth stage. The study demonstrated that the synthesized dataset constructed using this methodology was highly accurate, with high temporal resolution and medium spatial resolution and it was possible to harmonize multi-source remote sensing imagery by the improved Kalman filter for long-term field monitoring.

Mapping regional cropping patterns by using GF-1 WFV sensor data

The successful launched Gaofen satellite no. 1 wide field-of-view (GF-1 WFV) camera is characterized by its high spatial resolution and may provide some potential for regional crop mapping. This study, taking the Bei'an City, Northeast China as the study area, aims to investigate the potential of GF-1 WFV images for crop identification and explore how to fully use its spectral, textural and temporal information to improve classification accuracy. In doing so, an object-based and Random Forest (RF) algorithm was used for crop mapping. The results showed that classification based on an optimized single temporal GF-1 image can achieve an overall accuracy of about 83%, and the addition of textural features can improve the accuracy by 8.14%. Moreover, the multi-temporal GF-1 data can produce a classification map of crops with an overall accuracy of 93.08% and the introduction of textural variables into multi-temporal GF-1 data can only increase the accuracy by about 1%, which suggests the importance of temporal information of GF-1 for crop mapping in comparison with single temporal data. By comparing classification results of GF-1 data with different feature inputs, it is concluded that GF-1 WFV data in general can meet the mapping efficiency and accuracy requirements of regional crop. But given the unique spectral characteristics of the GF-1 WFV imagery, the use of textual and temporal information is needed to yield a satisfactory accuracy.

Combined Calibration Method Based on Rational Function Model for the Chinese GF-1 Wide-Field-of-View Imagery

The internal geometric distortion of charge-coupled-device (ccd) chips degrades the internal geometric accuracy of satellite images. Even if a large number of ground control points (GCPs) are used for geometric correction, the satellite images cannot always be used for direct mapping. Because of the current lack of core parameters for building a rigorous geometric model, this paper proposes a combined interior-orientation calibration method based on the rational function model (rfm) for multiple GF-1 wide-field-of-view (wfv) images of the same general scene. A polynomial model of look angles in the image space is chosen to compensate for the systemic error of CCD chips. After calibration for sensors 1 to 4, the orientation accuracy is improved to 0.4 pixels. Moreover, the residual error figures show no obvious trend. Other GF-1 WFV images are calibrated using the obtained interior-orientation parameters, and the improvement in their orientation accuracy is demonstrated to validate the effectiveness of the proposed method.