Case Study: Water control / drainage using UAV imagery

Case Study: Water control / drainage using UAV imagery


Throughout AgSky Technologies Inc.’s primary operating area of western Manitoba and eastern Saskatchewan, excess moisture has always been an issue, and this crop year was no exception. During the last several weeks, we’ve been busy with post-harvest field imaging for the purpose of drainage ditching and water control. As a result, we are able to now present a brief case study documenting one client’s use of generated Elevation Data to reduce potential problem areas from his field.


Step 1: Capture initial UAV imagery

Google Earth rendering of flight and data capture locations

On October 11th, 2014, the AgSky crew conducted a flight mission to obtain RGB (natural light) imagery.  Parameters were set to approximately 300ft altitude in a East-West grid pattern across the 160-acre quarter section, and the resulting flight plan took about 30 minutes to capture 336 images from the on-board Canon 16.0MP digital camera.  Raw data captured totaled about 3.02 gigabytes for this one field.

Step 2: Process captured images

Digital point-cloud generated in Postflight Terra 3D

Back at the office, these 336 JPEG images were processed using Pix4D’s Postflight Terra 3D v3.2.69 on a Core i7-4770K Windows 7 x64 desktop system with 32GB RAM.  After approximately 4 hours of automated processing, the multiple individual images were combined into two large GeoTIFFs with a 3.86 average GSD (Ground Sampling Distance).  The first GeoTIFF was a 1.8gb orthomosaic providing a visual and geo-referenced map of the field.  The second GeoTIFF was a 830mb DSM (Digital Surface Map) representing the triangulated elevation map of the field.  A full 3D point-cloud was generated as well with approximately 74 million data points, and the overall project files totaled 12.6gb.

Step 4a: Post-Process using GIS

Next, the GeoTIFFs are loaded into a GIS program (AgSky currently uses Quantum GIS v2.4.0).  An accelerated overview of the post-processing is shown in the video above, and represents about 1 hour of analysis time, although this can vary greatly depending on the complexity of a field and type of reporting desired.

 Step 4b: GIS algorithm and analysis

At this stage, various calculations and algorithms are applied to provide easy-to-read Elevation maps, Hillshade renders, and Watershed analysis.  A special algorithm called Fill Sinks XXL (Wang & Liu) is used in this case to identify potential areas that will be impacted by excess moisture and need attention to ensure better drainage. (Click tabs to view data layers)

Step 5: Reports generated and delivered


Report generation screen

Using QGIS Report Composer tool, PDF reports were prepared showing various layers of data and delivered to the client in both a printed and digital format.  Below you can download Sample reports which are similar to what was actually used by the client.

SampleReport_Elevation.pdf (Adobe PDF, 11,344kb)

SampleReport_Hillshade.pdf (Adobe PDF, 10,757kb)

SampleReport_WatershedSinks.pdf (Adobe PDF, 8,243kb)

Step 6: Client works field using report data


Data transferred via USB on to GreenStar 2630 for review and guidance in the field

Data is translated using John Deere’s APEX software and prepared for viewing on the GreenStar 2630 screen from USB memory stick.  Using this guidance and the paper map copies for overview, the client is able to work problem areas with scrapers and other implements.  In this particular case, the client took several hours on November 2nd, 2014, to work up the identified acres.

Step 7: Revisit and processing

Once the client finished the quarter, the area was revisited by AgSky and another flight mission was run.  Data was captured on November 9th, 2014, using similar flight parameters as on October 11th.  Once the 304 images (2.04gb) processed the resulting orthomosaic and DSM were transferred into QGIS for post-processing.  Analysis was performed and showed the following results: (Click tabs to view data layers)


Upon delivery and review of the maps, the client was very pleased with the information provided.  Paper maps were used to plan out general ditching before going to the field.  Using the data directly on the tractor’s GS 2630 display screen allowed him to then quickly and efficiently locate and work potential problem areas, and move less dirt in the process.

With the confirmation of a second flight mission, it’s clear that most potential problem areas were greatly reduced.  The integrity of the data and analysis is shown in the southeast corner of the field (bottom right corner) were the client did not do any direct drainage work.  As a result, the problem areas highlighted in the first flight were noted again in the same areas that weren’t worked.

Of course, the true test would be to compare yield maps from one season to the next, and an update will be posted if/when this information is available.