0

Your Cart is Empty

Flood Monitoring From Space

November 05, 2021 5 min read

Monitoring Flood monitoring from space using earth observation / remote sensing

 

 Subscribe here

 

Join the Newsletter

Subscribe to get our latest podcast content by email.

    We won't send you spam. Unsubscribe at any time.

     

    Our guest today is Shay Strong, Ph.D. Dr. Strong is the recently appointed Vice President of ICEYE, based out of Espoo, Finland. While she is somewhat new to the world of SAR, she has vast experience in remote sensing, and brings valuable cross-disciplinary insight via her P.hD in astrophysics from the University of Texas at Austin. Beginning a shift out of academia, she worked for Johns Hopkins University, developing national security satellite systems. This evolved into an interest in cloud-based imagery data pipelines, leading to her current role with ICEYE where the focus is optimizing these data pipelines, and how to efficiently extract the most useful information from them.  ICEYE is a SAR satellite company that utilizes their own constellation of 14 satellites to respond to and monitor flooding events. 

     

    Monitoring Flooding - Why Is It So Hard?

     

    There are quite a few challenges in the pursuit of monitoring and mapping flooding.For ICEYE, these challenges begin with determining where collection needs to be to begin with. The nature of catastrophic events is that they happen quickly, and this reduces the timeline available to activate resources. Considering that boots on the ground have the time-based priority of protecting themselves and others in immediate danger, ICEYE has built their own team of meteorologists and analysts to help predict where and when the next event may be.

     

    Having an idea of where and when the next flooding event may be is invaluable, but it is still difficult to make predictions reliably enough to capture the peak of an event. The peak of a flooding event would look like the deepest and widest extent of flooding, or destruction of critical infrastructure. 

     

    Determining locations with the greatest potential for impact can be further enabled by referencing other classic GIS resources, such as demographics and maps of existing infrastructure. Analysts can combine this information with historical data and weather models to understand where the areas of highest risk are. This means that when there are multiple potential events that need to be monitored, it is easier to make the difficult decision of where resources should be allocated. 

     

    The key element of monitoring a natural disaster is being able to monitor change. In a flooding event, this change happens quickly. In order for information to be useful and valuable, it must also be timely. Repeated collections over an area allows the creation of time-series products, allowing more accurate and relevant solutions for stakeholders.

     

    SAR and Flooding

     

    SAR, synthetic aperture radar, is highly versatile and capable of providing timely and valuable information about the geography of a study area. As an active system,SAR has the benefit of being able to penetrate clouds and haze, as well as being able to continue collection through both the day and night. There are a variety of image collection options out there, including optical, infrared, thermal, etc. but this round-the-clock collection aspect of SAR makes it especially useful for responding to emergencies. 

     

    Collecting SAR data is only the first part of the puzzle. Once the imagery is collected, it must enter the data pipeline to be assimilated, processed, and analyzed before useful information can be derived from its products. The trick is that these are real-life events being monitored, with very high stakes. The derived information is most needed within 24 hours of the event to inform responders and the public, creating added pressure on analysts for swift and accurate work.

     

    Before one can evaluate the results of the collected data, it must first go through ground truthing.Ground truthing is the process of relating collected imagery to its real life location. In GIS, we frequently see this as georeferencing.  Validating SAR data in a flooding event is quite challenging. Flood gauges that already exist are most likely near water bodies, such as rivers, which have likely overflowed as part of the event, and are not visible, or producing useful information under the circumstances. Additionally, previously easily identifiable locations, such as street or building corners, maybe inundated by the flood, and no longer act as useful landmarks for ground-truthing. 

     

    Any information collected via SAR is only as useful as the DTM (digital terrain model) it is overlaid on. This is complicated by the fact that high-quality DTMs are expensive, and narrowly available in many locations. In rough areas with even moderate variability in terrain, the standard 30m DTM is going to be minimally informative. SAR raster data can have a resolution of up to 25 cm. In order to calculate the 3D extent of the flood (the depth) the DTM must be subtracted from the ground-truthed SAR readings, so detail in the terrain model is imperative to getting useful results.  

     

    The Business of Monitoring Flooding

     

    Satellite collection in general carries a hefty price tag, so it is important that there be a consumer market for imagery out there to support it.When it comes to flooding, in particular, those who see the most value so far are insurance agencies and governments. Insurance companies will use the products to help evaluate claims and update the risk level of various territories. Governments will primarily find value in the context of emergency aid response, and more so, to assess future risks due to climate change. 

     

    A big part of understanding where the viable market is for SAR data on flooding is knowing where the information will be most valued. For some places, like India in the monsoon season, flooding is seasonally routine, and residents have adapted to it for thousands of years. For these reasons, they probably will not see as much value in satellite monitoring as, say, a tourist district in southern Louisiana. 

     

    Another high-cost point in the flood imaging pipeline is the manual labor involved.Although automation has come a long way, there are still a significant amount of human eyes (and brains) needed to complete an analysis. It is likely we will see the number of humans in the mix reduced, but never eliminated. 

     

    As we explored earlier, one of the difficult aspects of making SAR data usable in a flood situation is ground-truthing. One way the team at ICEYE has been able to bypass this is by tapping into social media resources. Videos posted in places like YouTube or TikTok of flooding events can be used to help verify collected imagery. Considering how far sensors in cell phones have come, there is a lot of room for improving how citizen scientists can contribute to collection efforts. 

     

    Overall, when looking to create a geospatial solution, it is vital to make sure you are answering the right questions. Do not be afraid to be specific, as broad questions can have even broader answers. A business of this kind will have technical, as well as cultural challenges, and the more you understand them before encountering them, the better. 




     Subscribe here

     

    Join the Newsletter

    Subscribe to get our latest podcast content by email.

      We won't send you spam. Unsubscribe at any time.

       



      Also in THE MAPSCAPING PODCAST

      aerial imagery
      Collecting And Processing Aerial Imagery At Scale

      November 11, 2021 5 min read

      Imagery collection is a stage of the GIS process that has been taken for granted for some time. Commonly, consumers are limited to coarse government-provided satellite imagery, or must commission a UAV/UAS collection of the area they are interested in to obtain high quality data products that suit their business needs.
      Read More
      Labels Matter
      Labels Matter

      November 05, 2021 6 min read

      Deep learning and machine learning are both types of artificial intelligence (AI). Ultimately, the goal of AI is to help humans solve problems more efficiently. In order to do this we need to train the algorithm and to do that we need training data ( labeled data )
      Read More
      Geospatial startups
      Getting Your Geospatial Start-Up Started

      November 05, 2021 5 min read

      Building a startup from the ground up is no doubt an intimidating process, so it can be an excellent jumpstart to have specialized resources at hand when beginning such an adventure. There are organizations out there that exist to help startups in their early stages.
      Read More