Skip to Content

Flight #2 – June 27, 2015

Iowa River – Marengo

  • Water above level registered on Google Earth – Riparian trees helped to delineate water line
  • Perpendicular photo makes it possible to evaluate water level among trees with good precision
  • Orthogonal photo makes it possible evaluate water location on larger areas – trees and obstacles can cause a fairly precision
  • Aerial vision of drone allowed observation of flooded field hidden from road by trees
  • No inundation models in Marengo to compare with verified wet areas

Data:

  • 8:45 am
  • Discharge = 10,100 ft^3/s
  • Stage = 16.31 in

Legend for inundation maps:

  • Yellow – verified wet areas

Untitled

Untitled

 

Des Moines River – Des Moines

  • An orthogonal photo allows to assess river border even with trees
  • James W. C. Soccer Park partially flooded – easy to delimitate water reach

Data:

  • 10:50 am
  • Discharge = 49,000 ft^3/s
  • Stage = 28.5 in

Legend for inundation maps:

  • Blue – verified wet areas
  • Black – blind areas due to obstacles (trees, overpasses, etc.)
  • Red – model expected wet area, actually observed as dry

DCIM100MEDIA

Untitled

Untitled

 

Raccoon River – Des Moines

  • Possibility to observe where water did not reach
  • Fleur Dr. was closed due to flooding. Use of drone provided a good vision of road blocked by flooding and its surrounding areas

Data:

  • 11:12 am
  • Discharge = 48,300 ft^3/s
  • Stage = 28.56 in

Legend for inundation maps:

  • Blue – verified wet areas
  • Black – blind areas due to obstacles (trees, overpasses, etc.)
  • Red – model expected wet area, actually observed as dry
  • Yellow – model expected dry area, actually observed as wet

DCIM100MEDIA

Untitled

Untitled

Untitled

 

Raccoon River – West Des Moines

  • Isolated trees and buildings reached by water were excellent reference points
  • Wet areas observed by drone (yellow) compared with wet areas forecasted by model (blue) with 27.5 in. Model forecasted water lines reaching regions above was possible to assert with drone photos.

Data:

  • 11:20 am
  • Discharge = 48,300 ft^3/s
  • Stage = 28.56 in

Legend for inundation maps:

  • Blue – verified wet areas
  • Black – blind areas due to obstacles (trees, overpasses, etc.)
  • Red – model expected wet area, actually observed as dry

DCIM100MEDIA

Untitled

Untitled

 

Conclusion:

As observed in Flight #1, tree coverage results in lower accuracy on delimitation of water line. In some situations, taking perpendicular photos can be helpful for that propose.

We did not find a way to obtain the angles (vertical and horizontal) in which each photo was taken, but the coordinates automatically registered meta-information of each photo that allowed photos, when imported into a Picasa web album, to be geo-located automatically, making it easy to recognize each group of photos (important on a trip in which 8 stops were made consecutively and the amount of images obtained was huge).

The use of drones along with the Google Earth capability of generating polygons to represent the water line and its overlaying with the waterline generated by flood models allows a fast way to assess the model’s accuracy. Sometimes it is possible to observe a different color of grass on Google Earth were the water line in flooding reaches due to moisture above a common level.

It was possible to check that most of the waterlines forecasted by models were positively assessed on aerial photos.

Tips for future flights on assessing water lines at flood events:

  • Take 4 (or more) orthogonal photos at each point so a 360-degree vision can be observed
  • Assessment of where water did NOT reach is also important during anomalous river activity
  • Taking an additional perpendicular photo above water lines can help where there are too many trees and no reference points
  • Isolated trees and buildings inside water bodies are good reference points. Look for them while taking photos
  • Importing photos into Picasa can save much time in positioning them on the map and identifying each assessed region
  • KML files describing flood model waterlines being used by IFIS do not load properly on Google Earth. How to solve: create a map in Google Maps, import the KML and export the KML. Google Maps will make it compatible with Google Earth after exporting
  • Drone’s average field of view is a circle with a 500m radius. This can be taken into consideration when planning flying locations
  • Lens correction can be easily performed using Photoshop .lcp files. Looking for the specific drone’s lens correction profile file on the web and applying it is a good way to improve image understanding. The bad part: during correction, some information near photo borders are lost
Last modified on August 27th, 2015
Posted on August 25th, 2015