Spatial Downscaling of Streamflow Data with Attention Based Spatio-Temporal Graph Convolutional Networks

Sit, Muhammed; Demiray, Bekir Zahit; Demir, İbrahim

doi:10.31223/x5666m

Cited by 5 publications

(5 citation statements)

References 49 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…More recently, researchers have started exploring downscaling approaches for irregularly structured data, such as river networks [18]. In prior work, a graph network model is used to represent and predict intricate network dynamics, while a long short-term memory (LSTM) model is used to capture flow sequences along river paths [141]. These methodologies hold promise for enhancing fine-scale environmental predictions from coarser datasets.…”

Section: Downscalingmentioning

confidence: 99%

Knowledge-Guided Machine Learning

Karpatne¹,

Kannan²,

Kumar³

2022

View full text Add to dashboard Cite

This paper presents an overview of scientific modeling and discusses the complementary strengths and weaknesses of ML methods for scientific modeling in comparison to process-based models. It also provides an introduction to the current state of research in the emerging field of scientific knowledge-guided machine learning (KGML) that aims to use both scientific knowledge and data in ML frameworks to achieve better generalizability, scientific consistency, and explainability of results. We discuss different facets of KGML research in terms of the type of scientific knowledge used, the form of knowledge-ML integration explored, and the method for incorporating scientific knowledge in ML. We also discuss some of the common categories of use cases in environmental sciences where KGML methods are being developed, using illustrative examples in each category.

show abstract

Section: Downscalingmentioning

confidence: 99%

Knowledge-Guided Machine Learning

Karpatne¹,

Kannan²,

Kumar³

2022

View full text Add to dashboard Cite

show abstract

“…In recent years, attention mechanisms have gained remarkable prominence within the realm of DL models, concurrently catalyzing advancements in research and applications within the domain of RS. For instance, Sit et al [38] employed a graph convolutional network with spatiotemporal attention mechanisms to spatially downscale streamflow data, significantly enhancing DL's role in flood forecasting. Additionally, researchers like G. Liu et al [39] introduced a DL model incorporating a terrain-guided attention mechanism, establishing non-linear mapping for downscaling the temperature distribution in the southwestern region of China.…”

Section: Introductionmentioning

confidence: 99%

Spatial Downscaling of ESA CCI Soil Moisture Data Based on Deep Learning with an Attention Mechanism

Zhang,

Lu,

et al. 2024

Remote Sensing

View full text Add to dashboard Cite

Soil moisture (SM) is a critical variable affecting ecosystem carbon and water cycles and their feedback to climate change. In this study, we proposed a convolutional neural network (CNN) model embedded with a residual block and attention module, named SMNet, to spatially downscale the European Space Agency (ESA) Climate Change Initiative (CCI) SM product. In the SMNet model, a lightweight Convolutional Block Attention Module (CBAM) dual-attention mechanism was integrated to comprehensively extract the spatial and channel information from the high-resolution input remote sensing products, the reanalysis meteorological dataset, and the topographic data. The model was employed to downscale the ESA CCI SM from its original spatial resolution of 25 km to 1 km in California, USA, in the annual growing season (1 May to 30 September) from 2003 to 2021. The original ESA CCI SM data and in situ SM measurements (0–5 cm depth) from the International Soil Moisture Network were used to validate the model’s performance. The results show that compared with the original ESA CCI SM data, the downscaled SM data have comparable accuracy with a mean correlation (R) and root mean square error (RMSE) of 0.82 and 0.052 m3/m3, respectively. Moreover, the model generates reasonable spatiotemporal SM patterns with higher accuracy in the western region and relatively lower accuracy in the eastern Nevada mountainous area. In situ site validation results in the SCAN, the SNOTEL network, and the USCRN reveal that the R and RMSE are 0.62, 0.63, and 0.77, and 0.077 m3/m3, 0.093 m3/m3, and 0.078 m3/m3, respectively. The results are slightly lower than the validation results from the original ESA CCI SM data. Overall, the validation results suggest that the SMNet downscaling model proposed in this study has satisfactory performance in handling the task of soil moisture downscaling. The downscaled SM model not only preserves a high level of spatial consistency with the original ESA CCI SM model but also offers more intricate spatial variations in SM depending on the spatial resolution of model input data.

show abstract

“…The advancements in hardware and memory, on the other side, remove the major barrier for large data processing. To date, however, most efforts in simulating and synthesizing RS images focus on a few topics, such as forecast or nowcast of weather variables (e.g., precipitation) Wang 2022, Tuyen et al 2022), image super-resolution and downscaling (spatial enhancement) (Sit et al 2023b;2023c;Harris et al 2022), image time series generation (temporal enhancement) (Requena-Mesa et al 2021, Sit et al 2023a, and image translation between different RS sensors (Zhu and Kelly 2021, Czerkawski et al 2022, Vandal et al 2022, whereas synthesizing optical / radar images that capture Earth surface characteristics has not been well studied.…”

Section: Introductionmentioning

confidence: 99%

Better Localized Predictions with Out-of-Scope Information and Explainable AI: One-Shot SAR Backscatter Nowcast Framework with Data from Neighboring Region

Li¹,

Demir²

2023

Preprint

View full text Add to dashboard Cite

Synthetic Aperture Radar (SAR) provides 10-m weather-independent global Earth surface observations for various tasks such as land cover land use mapping, water body delineation, and vegetation change monitoring. However, the application of SAR imagery has been limited to retrospective tasks by a “first event then observation” rule. Recent studies have proven the feasibility of a one-shot forecast of backscatters of SAR imagery using meteorological driving forces, soil moisture, geomorphic factors, and previous SAR images collected for the target area. Although the approach is promising, spatial connectivity, more specifically, the influence of the status of surrounding areas on the target location has yet to be considered. To fill that gap, this study proposed two nowcasting frameworks that can integrate precipitation and soil moisture data collected from surrounding areas through spatial aggregation (SA) and by processing spatial series (SS), respectively. The catastrophic 2019 Central US Flooding was used as a case study with the goal of predicting backscatters of SAR imagery captured during the event. The results from SA, SS, and a framework that only considers localized input (S0) are compared against each other as well as with the benchmark performance created with persistence assumption. Results show that S0, SA, and SS outperform the benchmark. In addition, considering data from neighboring areas that contribute to the target location further improves prediction accuracy. Comparing the gradients of results considering/not considering additional data indicates that neighboring data can alter the model’s attention on each feature of the localized input matrix. The difference in gradients between SA and SS indicates the way the neighboring information is integrated also matters. The methodology proposed by this study can serve as a building block for more active usage of SAR imagery in forward-looking tasks such as early flood warning and response.

show abstract

Spatial Downscaling of Streamflow Data with Attention Based Spatio-Temporal Graph Convolutional Networks

Cited by 5 publications

References 49 publications

Knowledge-Guided Machine Learning

Knowledge-Guided Machine Learning

Spatial Downscaling of ESA CCI Soil Moisture Data Based on Deep Learning with an Attention Mechanism

Better Localized Predictions with Out-of-Scope Information and Explainable AI: One-Shot SAR Backscatter Nowcast Framework with Data from Neighboring Region

Contact Info

Product

Resources

About