Local Teams, Global Impact: Bathymetry-as-a-Service in Zambia
Ray reservoir - Lusaka August 2024

Local Teams, Global Impact: Bathymetry-as-a-Service in Zambia

Written by: Hessel Winsemius Olivier Hoes Dr. Hubert Samboko and Stephen Mather

Last month, the Open Profile team and the Water Resources Management Authority of Zambia (WARMA) performed field surveys of the bathymetry of a medium sized reservoir in Zambia, using a new, innovative, affordable, and easy to use approach which combines wet and dry bed bathymetry surveys. The survey lasted a few days and resulted in a phenomenally accurate bathymetry map of the reservoir. With a broader training session, the method can be easily replicated by local teams, as equipment used is off-the-shelf and highly affordable as shown in our earlier post.


Wet bathymetric map of Ray reservoir

Open Profile, funded by the World Meteorological Organization, has the mission to establish an affordable, replicable, low complexity method to acquire bathymetry records of streams, rivers and lakes. We do this by combining unique knowledge from the following innovative organizations: Rainbow Sensing, Delft University of Technology, University of Zambia, and OpenDroneMap, Inc., with notable support from the Water Resources Management Authority of Zambia (WARMA). For surveying equipment we decided to focus on the use of light off-the-shelf components that can easily be ordered, replaced and moved around: a fish finder, Do It Yourself GNSS (GPS) equipment, and small uncrewed aerial system (sUAS or drone). Price indications of the main parts are below.


  • Fish finder echo sounder: $400
  • DIY GNSS (GPS) Rover: $425
  • DIY GNSS (GPS) Base Station: $425
  • Small Uncrewed Aircraft System: $1000

Going deeper

By using an approach that combines novel, lightweight equipment to generate wet surveys with UAVs which survey the dry bed, our approach is not only cost effective, it can also be performed under a wider range of conditions. For example, with reservoirs, we are no longer limited to wet season-only surveys, but can complete surveys in the dry season. In our field test, we did a reservoir survey during the worst drought in 20 years time.

We augment the equipment for streams and rivers where higher flow velocities may hamper the use of the fish finder. We add to the fish finder a PVC rig that can be easily assembled. The idea of the base / rover GNSS combination is that, unlike the accuracy of a smartphone on-board GNSS module, all survey materials can receive a very accurate position in the order of a few centimeters. With Android’s developer options on an Android phone, we can easily replace the GNSS position of the smartphone’s GNSS sensor, by the solution provided by the base / rover combination, thus ensuring that bathymetric points as well as the shoreline are highly accurate in space. As surveyed water bodies become smaller, this becomes more and more important. The hardware choices and a full hookup guide for the rig and electronics are found in an earlier post.


From top to bottom: GNSS antenna, GNSS chipset (ublox ZED-F9P on an ardusimple breakout board), Android smartphone and a Deeper Chirps 2+ fish finder. Cabling is for connecting the GNSS antenna to chipset and reading GNSS data on a smartphone.

Building as a local service / heading to the field

We want to prove that our innovation can be turned into a local service, which we are further developing in a European project TEMBO Africa. To this end, the survey method must be accurate and it must be feasible to collect the required data in a reasonable time span. With WARMA, we decided to map the bathymetry of Ray reservoir, just North of Lusaka. WARMA provided a boat, and we went into the field several days with WARMA’s coxswain mr. Chris Ntobolo. The equipment was used to collect the following raw data:

  • Bathymetry points up to a minimum depth of about 0.6 m (the fish finder can measure shallower water, but the boat was too deep for this). A total of 35,106 points were collected at a spacing of roughly 1.5 meters.
  • The location of the shoreline, by walking along the shoreline with a GNSS antenna hidden in our hats (try that with a full sized Trimble or Leica GNSS ??). This is used to have a zero-meter boundary for interpolation of the bathymetry points, and align the RGB-drone terrain data.
  • 1,365 Raw RGB photos covering the dry parts of the reservoir. This was done with an affordable drone setup, here we used a Parrot Anafi and a DJI Phantom 4 Pro. These photos are used in OpenDroneMap’s WebODM to establish a dry-bed terrain dataset through photogrammetry. Due to limited time for ground control, we post-processed the dry bed bathymetry data using an innovative merging through use of much coarser satellite terrain data MERIT Hydro and field data, a process that we will document in a forthcoming blog post.

Some impressions of the field work and the environment are shown below.

RGB orthophoto derived through OpenDroneMap processing


wet bathymetry survey work


Aerial view and drone survey (Left: Boyd Mutale Mutengo - UNZA, right: Hubert Samboko - OpenDroneMap.org


surveyed transects by boat

Back in the office (well, actually….our hotel…) we merged the shoreline and the bathymetry points into one dataset, with zero depth points at the shoreline as above and processed the drone imagery into raw products as below.


Point cloud from WebODM

The wet bathymetry data was interpolated into a 2 meter resolution wet bathymetric raster, combined and with dry bathymetry from the drone after we returned from the field as follows.


Combined bathymetry for Ray reservoir

In many cases, the end product for a reservoir is a stage - volume relationship. This relationship can be used to better understand (changes in) the reservoir’s capacity, or enable a simple way to measure the current volume, with only a gauge plate. Below, we show the resulting stage - area and stage - volume relationships. Note that the spatial accuracy of the GNSS equipment also reveals the location of the original channel and other details. This can e.g. work for mapping of sedimentation if surveys are performed at a more regular basis.


Left: delth - area relationship, right: depth - volume relationship

Our fieldwork has demonstrated that the costs for surveying equipment can be lowered by a factor 10 allowing local people in any economic setting to provide a service with our innovation. A typical single beam echo sounder costs USD 15,000. With the total cost of ownership for materials being this low, the method can already be used by a local service provider by investing at minimum about 1,500 euros in GNSS and fish finder equipment (see the table above) and several practical smaller items such as a tripod, a clamp and small tools and utensils, not included there. This excludes the ability to also acquire dry-bed elevation with the sUAV, which also does not work with a professional echo sounder. We assume a boat and engine is provided by a client, and no spare parts are carried. We highly recommend at least invest double the amount in order to buy spare parts.

Validation

We find it important to not only demonstrate the experience and use cases, but also prove that our depth estimates are accurate. We performed a comparison of a bottom track, collected with an Acoustic Doppler Current Profiler and a fish finder time series, taken at the same time (27 August about 10AM local time, for a one-hour survey). The ADCP is not equipped with a GNSS sensor. Hence matching of the records was done based on the time stamps. The figure below shows a time series comparison side-by-side with a scatter plot of the two observation datasets.

Time series and scatter plot of simultaneous depth recordings by ADCP and Chirps fish finder estimates.

The Chirps dataset depth and positions were matched to the time stamps of the ADCP dataset to be able to make a difference plot in space. The map below shows these differences spatially.


Depth differences between Chirps and ADCP ground track

The results demonstrate that the two platforms give a remarkably good resemblance. The largest differences between the two datasets appear in deeper water, where differences of about 0.2 meters are experienced. We believe that this is due to the fact that the ADCP bottom tracker only has a vertical resolution of 0.2 meters and that the software only adds a 0.2 meter interval once a full 0.2 interval has been observed.

Conclusion

Thanks to the funding from WMO, Open Profile has successfully demonstrated and field tested an affordable, replicable, low complexity method to acquire bathymetry records of streams, rivers and lakes. We expect, particularly with the current water crisis in southern Africa to see an increased demand for the technique we piloted with combined dry and wet bed bathymetry. With a successful pilot completed, we will be demonstrating our findings at the Waternet symposium in Lesotho soon, further develop the service in our ongoing EU project TEMBO Africa, seek funding for commercialization and training, and grow the network of people able to perform these critical services, both on reservoirs as we demonstrated, as well as flowing streams and rivers.

Renaud Lafortune

Entrepreneur, Eng

1 个月

Would be really useful!

回复

要查看或添加评论,请登录

Rainbow Sensing的更多文章