Refining drainage mapping using low-cost devices

A Research and Innovation Story

In the previous blog post, we explored how detailed geometric drainage data can be collected with smartphones using a community mapping approach. The data was collected in Dar es Salaam, Tanzania, in the years 2017–2019. In 2020, the mapping initiative was expanded to Mwanza at the shore of Lake Victoria. In this blog post, we outline how the mapping methodology was improved to allow for even more accurate drainage data, acquired with locally available technologies.

By Iddy Chazua, Elia Dominic, Louise Petersson, Hessel Winsemius

Drainage mapping in Mwanza used the data model adopted from the Dar Ramani Huria project with updates and modification of some issues that were identified during the drainage mapping in Dar es Salaam. Specifically, the elevation measurements were significantly improved. In Dar es Salaam, the bed levels of drains were estimated by measuring the vertical distance between the drain bottom and the nearest road at the upstream end of each drain segment. Then, the absolute elevation of the drain bed level was calculated in relation to a digital terrain model. For the Mwanza mapping initiative, the team used a survey-grade technique which is Real-Time Kinematic (RTK) with a low-cost GNSS receiver called Ublox unit.

The Mwanza drainage mapping was conducted on two different tracks.

• Drain line and drain points
• Elevation Points

Caption: Drainage mapping in Mwanza

Drain lines and drain points

This was the initial phase, of which most of the procedure was adapted from the Dar Ramani Huria methodology with minor updates considering connectivity issues in the Dar es Salaam data set. Like in Dar es Salaam, the drainage line tracing was done from the downstream end of the drainage to its upstream end using the Open Data Kit Collect Android application (ODK). The main difference in the workflow compared with Dar es Salaam was that the team started a new geotrace in the ODK application every time a drain connected to a side-branch, instead of mapping the drain as a continuous geotrace, to avoid a hydraulically disconnected data set.

Prior to the surveying, the team identified, surveyed and set up a base station that was acting as a reference point and streaming corrections using the internet to the rovers. During surveying, the rover connected to the smartphone and ODK interface used for surveying but using the corrected location. The elevation data was collected as drain points at the starting point, endpoint and at the middle of the drain if necessary. Additionally, extra points were taken throughout the drains of for example waste blockages and other irregularities.

For quality assurance of the drain lines and drain points, the team applied several methods:

• Field papers during data collection. The team used field paper to ensure that data was completely collected in the area of interest. Field papers helped in monitoring the data collection and distribution of the work to all surveyors.
• Using QGIS for aligning drains segments with aerial imagery. After data collection, most of the drain segments do not fully follow the physical drains because of the GPS accuracy and human error while tracing. Because of this, updated imagery with the satisfying resolution was used for correcting the position of the drain lines in relation to the aerial imagery.
• Hydro-OSM model for data quality checking. This is used for final data check to unveiling the unseen disconnected drains in the data which cannot be seen by eyes. The tools help in checking the connectivity of the network, identifying disconnected drains.

Caption: Drainage mapping in Mwanza

Elevation Points

For elevation, Precise Point Positioning (PPP) methodology was used to survey the base stations and Real-time Kinematic (RTK) for surveying elevation points around the city. PPP is an established positioning technology to obtain high-precision coordinates that rely on observations recorded by a single receiver with errors as small as a few centimeters under good conditions. The base stations surveyed for more than 8 hours to reach the convergence time. Before surveying, the team located the area for the base station by considering the following factors:

• The maximum distance from the base station to the farthest point in the survey area should not exceed 8 km.
• Base GPS receivers were placed in locations with a clear line of sight to the sky in all directions. This factor was highly considered with regard to the mountainous structure of the Mwanza city.
• The base GPS receiver is located in a location with an equal distance range in all directions and provides full coverage of the site.

Then the following procedures employed for surveying the base station:

• Stationary data was collected for more than 8 hours
• The raw data was converted from the original file format i.e UBX to RNX using a simple python script
• Then the RNX files were uploaded to nrcan, the Canadian Geodetic Reference Systems Tools for accurate positioning, heights and coordinates transformations for processing.
• The output coordinates were referred to as the estimated position which was then used for ocean tide loading values correction. With these values in BLQ file obtained from tide, data was uploaded again to nrcan for final output
• From nrcan, the final coordinates were collected and used as reference points for the base station.

Caption: Setting up base station

After the base station was established, the team configured two rovers for starting the survey. Each rover set up contains a U-blox ZED-F9P receiver connected to the antenna, Raspberry Pi and battery. The base station streams data in real-time to the rover through an Internet Protocol (NTRIP). The rover is connected to an Android mobile phone and by using the Android application PPM Commander on the phone, the rover antenna position is used as the phone’s location. Then the team used the steps below to start surveying.

• The rover must be powered and connected to any nearby internet source because it uses Networked Transport of RTCM via Internet Protocol (NTRIP) to receive the correction from the server.
• Then the mobile phone connected to the Rover through USB On-The-Go (USB OTG), and the mock location enabled in the mobile phone setting allows the use of the external GPS.
• Open Data Kit (ODK) is used for surveying. The initial stage is to download the survey form from the server. The survey includes all attributes to be collected including the point, elevation and accuracy of the GPS.
• Before this task, the team has already surveyed and traced all the drains — collecting points to show if the drains are blocked, outflow or any other identified features based on the data model. With this information, the field paper (simple map to show all the mapped features) is produced and the team relied on the map to revisit all the previous mapped drains for elevation surveying.
• For each drain, the team surveyed elevation points at the beginning and end of the drainage and if necessary at the middle. The points are collected as closed as possible to the drains.
• PPM Commander application is used for monitoring the GPS solution. Throughout the process, the surveyors were supposed to make sure the GPS status is in FIX mode not otherwise.
• Also in PPM Commander, the Differential Global Positioning System (DGPS) which helps to tell the strength of internet connection on either side was closely monitored. DGPS is the time for message transmission from base to rover and it should be between 1–3 seconds.
• The surveyor was also supposed to measure the height of the rover rod. During data processing the rover height will be deducted from the surveyed height to get accurate measurements.

Caption: Mwanza drainage

Caption: Mwanza drainage

Apart from the drainage mapping, the flood depth collection method was adapted and improved. When we were conducting flood mapping in Dar es Salaam, the flood depth was measured by considering respondents’ height. i.e. if the water was ankle-deep, knee-deep, chest-deep, etc. This method proved some weaknesses as people have different heights. In Mwanza, the team applied a reviewed and systematic approach where measurements were taken from different respondents to determine their height with reference to the human height categories. The improved flood depth measurement method is described here.

To find out more, check out omdtz — OpenMap Development Tanzania.