The International Soil Moisture Network: serving Earth system science for over a decade

Hydrology and earth system sciences, Journal Year: 2021, Volume and Issue: 25(11), P. 5749 - 5804

Published: Nov. 9, 2021

Abstract. In 2009, the International Soil Moisture Network (ISMN) was initiated as a community effort, funded by European Space Agency, to serve centralised data hosting facility for globally available in situ soil moisture measurements (Dorigo et al., 2011b, a). The ISMN brings together collected and freely shared multitude of organisations, harmonises them terms units sampling rates, applies advanced quality control, stores database. Users can retrieve from this database through an online web portal (https://ismn.earth/en/, last access: 28 October 2021). Meanwhile, has evolved into primary reference worldwide, evidenced more than 3000 active users over 1000 scientific publications referencing sets provided network. As July 2021, now contains 71 networks 2842 stations located all globe, with time period spanning 1952 present. number covered is still growing, approximately 70 % contained continue be updated on regular or irregular basis. main scope paper inform readers about evolution past decade, including description network set updates control procedures. A comprehensive review existing literature making use also order identify current limitations functionality usage shape priorities next decade operations unique community-based repository.

Language: Английский

---

A 1 km daily soil moisture dataset over China using in situ measurement and machine learning

Earth system science data, Journal Year: 2022, Volume and Issue: 14(12), P. 5267 - 5286

Published: Nov. 30, 2022

Abstract. High-quality gridded soil moisture products are essential for many Earth system science applications, while the recent reanalysis and remote sensing data often available at coarse resolution only surface soil. Here, we present a 1 km long-term dataset of derived through machine learning trained by in situ measurements 1789 stations over China, named SMCI1.0 (Soil Moisture China data, version 1.0). Random forest is used as robust approach to predict using ERA5-Land time series, leaf area index, land cover type, topography properties predictors. provides 10-layer with 10 cm intervals up 100 deep daily period 2000–2020. Using benchmark, two independent experiments were conducted evaluate estimation accuracy SMCI1.0: year-to-year (ubRMSE ranges from 0.041 0.052 R 0.883 0.919) station-to-station 0.045 0.051 0.866 0.893). generally has advantages other products, including ERA5-Land, SMAP-L4, SoMo.ml. However, high errors located North Monsoon Region. Overall, highly accurate estimations both ensure applicability study spatial–temporal patterns. As based on it can be useful complement existing model-based satellite-based datasets various hydrological, meteorological, ecological analyses models. The DOI link http://dx.doi.org/10.11888/Terre.tpdc.272415 (Shangguan et al., 2022).

Language: Английский

---

Space-time super-resolution for satellite video: A joint framework based on multi-scale spatial-temporal transformer

International Journal of Applied Earth Observation and Geoinformation, Journal Year: 2022, Volume and Issue: 108, P. 102731 - 102731

Published: Feb. 25, 2022

Satellite video is an emerging type of earth observation tool, which has attracted increasing attention because its application in dynamic analysis. However, most studies only focus on improving the spatial resolution satellite imagery. In contrast, few works are committed to enhancing temporal resolution, and joint spatial-temporal improvement even less. The enhancement can not produce high-resolution imagery for subsequent applications, but also provide potentials clear motion dynamics extreme events observation. this paper, we propose a framework enhance simultaneously. Firstly, alleviate problem scale variation scarce video, design feature interpolation module that deeply couples optical flow multi-scale deformable convolution predict unknown frames. Deformable adaptively learn information profoundly complement information. Secondly, transformer proposed aggregate contextual long-time series frames effectively. Since patches embedded multiple heads self-attention calculation, comprehensively exploit details all Extensive experiments Jilin-1 demonstrate our model superior existing methods. source code available at https://github.com/XY-boy.

Language: Английский

---

Hyperspectral Image Denoising: From Model-Driven, Data-Driven, to Model-Data-Driven

IEEE Transactions on Neural Networks and Learning Systems, Journal Year: 2023, Volume and Issue: 35(10), P. 13143 - 13163

Published: June 7, 2023

Mixed noise pollution in HSI severely disturbs subsequent interpretations and applications. In this technical review, we first give the analysis different noisy HSIs conclude crucial points for programming denoising algorithms. Then, a general restoration model is formulated optimization. Later, comprehensively review existing methods, from model-driven strategy (nonlocal mean, total variation, sparse representation, low-rank matrix approximation, tensor factorization), data-driven 2-D convolutional neural network (CNN), 3-D CNN, hybrid, unsupervised networks, to model-data-driven strategy. The advantages disadvantages of each are summarized contrasted. Behind this, present an evaluation methods various simulated real experiments. classification results denoised execution efficiency depicted through these methods. Finally, prospects future listed guide ongoing road denoising. dataset could be found at https://qzhang95.github.io.

Language: Английский

---

Combined deep prior with low-rank tensor SVD for thick cloud removal in multitemporal images

ISPRS Journal of Photogrammetry and Remote Sensing, Journal Year: 2021, Volume and Issue: 177, P. 161 - 173

Published: May 23, 2021

Language: Английский

---

InSAR-derived seasonal subsidence reflects spatial soil moisture patterns in Arctic lowland permafrost regions

˜The œcryosphere, Journal Year: 2025, Volume and Issue: 19(3), P. 1103 - 1133

Published: March 11, 2025

Abstract. The identification of spatial soil moisture patterns is high importance for various applications in high-latitude permafrost regions but challenging with common remote sensing approaches due to landscape heterogeneity. Seasonal thawing and freezing near-surface lead subsidence–heave cycles the presence ground ice, which exhibit magnitudes typically less than 10 cm. Our investigations document higher Sentinel-1 InSAR (interferometric synthetic aperture radar) seasonal subsidence rates (calculated per degree days – a measure heating) locations compared drier ones. Based on this, we demonstrate that relationship signals can be interpreted assess variations moisture. A range challenges, however, need addressed. We discuss implications using different sources temperature data deriving results. Atmospheric effects must considered, as simple filtering suppress large-scale permafrost-related underestimation displacement values, making Generic Correction Online Service (GACOS)-corrected results preferable tested sites. rate retrieval considers these aspects provides valuable tool distinguishing between wet dry features, relevant degradation monitoring Arctic lowland regions. Spatial resolution constraints, remain smaller typical features drive versus conditions such high- low-centred polygons.

Language: Английский

---

Remote sensing image gap filling based on spatial-spectral random forests

Science of Remote Sensing, Journal Year: 2022, Volume and Issue: 5, P. 100048 - 100048

Published: April 8, 2022

Remote sensing images play a significant role in global land cover monitoring. However, due to the influence of cloud contamination, optical remote inevitably contain large number missing data, which severely limits their applicability. Existing removal methods generally use only effective information from single band temporally close known images, is insufficient predict accurately changes between and cloudy images. In this paper, we proposed spatial-spectral random forest (SSRF) method for thick by gap filling, uses spatially adjacent multispectral simultaneously based on forests. With its capability fit nonlinear relations adaptively assign variable contributions, SSRF can handle potentially complex relationship thus, producing more accurate predictions. Based Landsat 8 OLI Sentinel-2 MSI data 13 regions, effectiveness was demonstrated through experiments both simulated real The results show that urban areas with strong heterogeneity agricultural temporal changes, yield satisfactory predictions visually quantitatively. Moreover, than two popular benchmark methods. addition, less affected size time interval still produce reliable when used are also contaminated clouds reduce amount available neighborhood information. omission error detection caused thin clouds. simple implement has great potential widespread application.

Language: Английский

---

DsTer: A dense spectral transformer for remote sensing spectral super-resolution

International Journal of Applied Earth Observation and Geoinformation, Journal Year: 2022, Volume and Issue: 109, P. 102773 - 102773

Published: April 30, 2022

To obtain high-resolution hyperspectral data, spectral super-resolution is a popular computational imaging technique directly from multispectral images. Besides sparse recovery, deep learning-based methods perform well in the past years for their powerful nonlinear mapping to domains. However, convolutions learning only focus on local information and have been blamed neglect of long-range relationships. Nowadays, transformer has attracting great interest its global attention interaction. In this study, we propose dense with ResNet achieve remote sensing Combining meets need 3D data handling images as Dense connection helps model exploit features multi-level transformers. Moreover, recovery results natural three sets all prove advantage proposed model. Furthermore, also carry out classification experiments real verify dependability reconstructed spectra.

Language: Английский

---

A Hybrid Triple Collocation-Deep Learning Approach for Improving Soil Moisture Estimation from Satellite and Model-Based Data

Remote Sensing, Journal Year: 2022, Volume and Issue: 14(7), P. 1744 - 1744

Published: April 5, 2022

Satellite retrieval and land surface models have become the mainstream methods for monitoring soil moisture (SM) over large regions; however, uncertainty coarse spatial resolution of these products limit their applications at regional local scales. We proposed a hybrid approach combining triple collocation (TC) long short-term memory (LSTM) network, which was designed to generate high-quality SM dataset from satellite modeled data. applied merge data Soil Moisture Active Passive (SMAP), Global Land Data Assimilation System-Noah (GLDAS-Noah), component fifth generation European Reanalysis (ERA5-Land), we then downscaled merged 0.36° 0.01° based on relationship between auxiliary environmental variables (elevation, temperature, vegetation index, albedo, texture). The results were validated against in situ observations. showed that: (1) TC-based validation consistent with situ-based validation, indicating that TC method reasonable comparison evaluation (2) merging superior simple arithmetic average when parent had differences. (3) Downscaled product better performance than terms ubRMSE bias values, implying fusion model-based would result downscaling accuracy. (4) not only improved representation variability but also satisfactory accuracy median R (0.7244), (0.0459 m3/m3), (−0.0126 m3/m3). effective generating fine reliable wide hydrometeorological applications.

Language: Английский

---

A robust gap-filling approach for European Space Agency Climate Change Initiative (ESA CCI) soil moisture integrating satellite observations, model-driven knowledge, and spatiotemporal machine learning

Hydrology and earth system sciences, Journal Year: 2023, Volume and Issue: 27(2), P. 577 - 598

Published: Jan. 30, 2023

Abstract. Spatiotemporally continuous soil moisture (SM) data are increasingly in demand for ecological and hydrological research. Satellite remote sensing has potential mapping SM, but the continuity of satellite-derived SM is hampered by gaps resulting from inadequate satellite coverage, snow cover, frozen soil, radio-frequency interference, so on. Therefore, we propose a new gap-filling approach to reconstruct daily time series using European Space Agency Climate Change Initiative (ESA CCI). The developed integrates observations, model-driven knowledge, machine learning algorithm that leverages both spatial temporal domains. Taking China as an example, reconstructed showed high accuracy when validated against multiple sets situ measurements, with root mean square error (RMSE) absolute (MAE) 0.09–0.14 0.07–0.13 cm3 cm−3, respectively. Further evaluation 10-fold cross-validation revealed median values coefficient determination (R2), RMSE, MAE 0.56, 0.025, 0.019 reconstructive performance was noticeably reduced excluding one explanatory variable keeping other variables unchanged removing spatiotemporal domain strategy or residual calibration procedure. In comparison gap-filled based on diurnal temperature range (DTR), bias-corrected model-derived DTRs exhibited relatively lower higher coverage. Application our long-term datasets (2005–2015) produced promising result (R2=0.72). A more accurate trend achieved relative original CCI assessed measurements (i.e., 0.49 versus 0.28, respectively, terms R2). Our findings indicate feasibility integrating fill short- series, thereby providing avenue applications similar studies.

Language: Английский

---