400 Rate this article:

Revolutionizing Water Management with SAR: A Hungarian Success Story

Erin Eckles Monday, March 18, 2024

Introduction

In many countries, water management infrastructure is aging, and threats induced by climate change are putting them at further risk. Today, stability and risk mapping through continuous monitoring is becoming an increasingly important requirement to mitigate damage, risk of collapse, and life-threatening accidents.

Any anomalous movement of water management infrastructure must be identified and analyzed in near real-time and at a millimeter-scale. While traditional monitoring methods are highly accurate, they are unfortunately challenging to conduct, time-consuming, expensive, and often cannot provide continuous data on the stability of a structure. Further, ﬂood and water management structures are often located in relatively remote locations, such as on riverbanks or in deep valleys, making onsite surveying difficult. Overgrown vegetation also limits how well water management structures can be accessed and surveyed.

Benefits of InSAR Based Earth Observation Monitoring

When it comes to persistent wide area monitoring, and in particular for difficult to access locations, space-based earth observation provides a wealth of advantages over traditional surveying methods for risk and stability mapping. Satellite based monitoring can overcome limitations of on-site surveying through wide area data coverage and efficient, accurate, repetitive, and cost-effective data collection, processing, and analysis.

Unlike optical satellite sensors that can only operate during daylight, Synthetic Aperture Radar (SAR) satellite systems provide their own energy source, allowing them to collect data day and night, and even “see” through clouds. Regularly acquired and freely available ESA Sentinel-1 SAR satellite data has become an appropriate and officially accepted data source, and Interferometric SAR (InSAR) based earth observation has matured to an established technological solution for persistent surface deformation monitoring from space.

SAR Interferometric Stacking techniques such as Persistent Scatterers (PS) and Small Baseline Subsets (SBAS, or Distributed Scatterers) can obtain very precise measurements of slow surface movements over long periods of time and have proven their reliability in monitoring surface deformation phenomena resulting from tectonic or human activities. Over the last decade, the vast amount of freely available ESA Sentinel-1 SAR satellite data, intensive development of InSAR processing techniques and increased computing capacity, have initiated the development of numerous InSAR monitoring services for all types of applications often covering entire countries and even the whole of Europe.

Figure 1: Location of surveyed water management facilities in Hungary (yellow circles) and installed Corner Reflectors (red circles). Source: Ronczyk et al. (2022).

InSAR Processing Pipeline for the FIR Earth Observation Information System

Recently, a nationwide, operational InSAR monitoring system for strategic water management facilities has been developed and introduced in Hungary for the Hungarian Disaster Management Service (HDMS). Project requirements included monthly monitoring of 63 water facilities, including 83 individual objects, distributed throughout Hungary (Fig. 1), in combination with the development of a near real-time warning system.

Figure 2: Examples of detectable (A1) and undetectable (B1) water management sites from Sentinel-1 SAR imagery, as well as sites where Corner Reflectors can be installed to improve detectability (C1) and their land cover from Google satellite images (A2–C2). Source: Ronczyk et al. (2022).

Work for the monitoring system involved the implementation of a completely new and fully automated InSAR System as a Service (SaaS) which incorporates user requirements, preparatory work, the compilation of an automatic Sentinel-1 processing pipeline, the installation of corner reﬂectors, a special early warning system, and a dedicated, web-based user interface that provides detailed information on the stability of the monitored water facility objects. Fig. 2 shows examples of water facilities that are monitored by this system. They are categorized into eight different types and dominated by sluices, weirs, ﬂoodgates, and dams.

The InSAR based monitoring processing pipeline for critical water management facilities is part of a larger, nationwide Earth Observation system in Hungary, called the “FIR Earth Observation Information System” (Földmegﬁgyelési Információs Rendszer). FIR is a complex national administration earth observation project based on state-of-the-art DevOps processes. FIR operates on other applications in Hungary too, such as forest detection, flood monitoring, and burn area identification.

Technical Details of the InSAR Monitoring System for Water Facilities

An article written by Ronczyk et al. and published in Remote Sensing Journal (2022) provides a comprehensive technical overview of this newly implemented InSAR system to monitor Hungary’s critical water management infrastructure. Such InSAR systems require massive amounts of compute processing power and signiﬁcant time to generate a ﬁnal product. Required hardware demands particularly depend on the number, distribution, and observation frequency of the targeted objects as well as the total size of the monitored area. State-of-the-art hardware and software solutions therefore needed to be implemented as part of this project to cope with technical challenges of near real-time data processing, the large number of objects to be monitored, the different types and sizes of water facilities, and to overcome potential limitations of the Sentinel-1 sensor.

The hardware setup of the FIR Earth Observation Information System consists of more than 1,000 CPU cores, 768 GB of RAM, 2 PB of storage, and two NVIDIA Tesla V100 GPUs that are exclusively dedicated to InSAR processing workﬂows. InSAR calculations are performed using the Persistent Scatterer (PS) technique which is ideally suited to detect objects with well-deﬁned geometry and permanent phase stability, such as man-made water management facilities.

The core part of the implemented InSAR processing technology is realized with ENVI SARscape, the industry leading software solution for the processing and analysis of SAR data. ENVI SARscape enables end-to-end processing and information generation from SAR data, including advanced multi-step processing chains for PS processing. ENVI SARscape can take advantage of GPU computational capacity as well, and through parallelization in network cluster environments, Persistent Scattering workﬂows can be speed-up significantly for big data processing.

The Sentinel-1 Persistent Scatterer processing workflow of ENVI SARscape was integrated into a DevOps environment and a Function-as-a-Service (FaaS) solution was implemented. Docker images realize a sound containerization strategy within a Kubernetes cluster. In a Docker image, the operating system, ENVI SARscape, as well as Python are mounted. Fig. 3 provides an overview of the implemented InSAR processing pipeline for the Hungarian FIR Earth Observation Information System.

Sentinel-1 images acquired in both ascending and descending geometries, with overlapping for each location, are used for separate line-of-sight PS calculations of each water facility object. These are then combined to obtain 2D measurements in vertical and East–West horizontal directions. Thanks to the system design, high-resolution X-band acquisitions can also be processed when images become available.

Tectonic effects only play a minor role on the stability of water facilities in Hungary, which is why the InSAR calculations are only focusing on the water facilities and their immediate surroundings. Sites which were initially identiﬁed as poorly detectable in the PS calculations had their recognizability increased through the use of corner reﬂectors. The InSAR validation procedure consists of multiple methodological approaches, the most important being comparing PS measurements with traditional levelling.

The InSAR monitoring system can process new Sentinel-1 acquisitions within 24 hours of the data being downloaded into the Hungarian Collaborative Ground Segment, which means the end-user receives any possible alarm notiﬁcations within a day of new Sentinel-1 images being acquired. Changes in the temporal evolution of displacement velocities can indicate changes in the behavior of a structure. A threshold value was assigned for each facility, and when this threshold value is exceeded, a warning alert for the individual structure is given. These insights help stakeholders and decision makers to evaluate if tolerances are being exceeded and if closer monitoring is needed.

Figure 3: InSAR processing pipeline of the Hungarian FIR Earth Observation Information System. Source: Ronczyk et al. (2022).

Conclusion

The versatile nature of SAR remote sensing technology allows for the monitoring of areas of interest day and night and through all types of weather. These unique properties make Interferometric SAR technology ideal for persistent monitoring of priority areas such as water management facilities, and for providing critical insights for stability and risk mapping.

Recently in Hungary, a nationwide, operational InSAR monitoring system based on Sentinel-1 SAR data has been launched as part of the national FIR Earth Observation Information System, with the objective to monitor the stability of national hydrological sites. This fully automated continuous interferometric monitoring system processes large amount of Sentinel-1 data using the latest hardware and state-of-the-art software technology, including high performance computing using GPUs, and cutting edge InSAR processing algorithms provided by ENVI SARscape.

Remote sensing stability monitoring using space-based SAR imagery and advanced InSAR processing algorithms can be coupled with traditional ground surveying techniques to provide critical insights for informed interpretation, validation, and remediation. One of the main cornerstones of the implemented system is the “near real-time” processing and warning capability to realize an effective and efficient risk mitigation strategy, further reducing costs and improving operational safety.

Outlook

While traditional geodetic surveying is still the primary method to monitor and review water management facilities in Hungary, the InSAR monitoring system currently acts as preliminary “spatial decision support system”. The authors of this article and those involved in the project, are conﬁdent that the implemented InSAR monitoring system will play an increasingly important role in the large-scale monitoring strategy of critical water management facilities in Hungary.

In the future, the system may be expanded to include Distributed Scatterers (DS, through SBAS) as part of a combined, single processing chain with Persistent Scatterers (PS). This strategy 'connects' the PS and SBAS methods, ensuring consistency of deformation results obtained at point-like (PS) and DS targets. Therefore, it provides better results compared to the approach of executing the two methods independently from each other. Combined PS / SBAS processing is possible through the newly implemented E-PS (Enhanced-PS) and E-SBAS (Enhanced-SBAS) techniques that were released with ENVI SARscape version 5.7.

Hungary’s success story is a compelling example of how SAR technology can transform the monitoring of critical infrastructure. As we navigate the challenges of a changing climate and aging infrastructure, embracing innovative solutions is not just an option – it’s a necessity.

Reference: Ronczyk L, Zelenka-Hegyi A, Török G, Orbán Z, Defilippi M, Kovács IP, Kovács DM, Burai P, Pasquali P. Nationwide, Operational Sentinel-1 Based InSAR Monitoring System in the Cloud for Strategic Water Facilities in Hungary. Remote Sensing. 2022; 14(14):3251. https://doi.org/10.3390/rs14143251