Home
Team members:
Dillon Hicks, Mark Liu, Kathy Qi, Jeremy Smith
Code and subproject-specific documents are stored under gps and canopy_height respectively.
Presentations and project-wide documents are stored in their own directories.
Mangroves are an important part of our environment as they contribute to sequestering carbon from the atmosphere, sheltering the coast from destructive waves, and providing habitats for delicate ecosystems. To aid in the conservation of mangrove forests, precise location and biomass measurements provide valuable information directly related to the estimated economic and environmental impact of the forest. Thus, our projects focused on providing low-cost and accurate methods of acquiring canopy heights estimations and mangrove geolocation.
The two devices in our project, the Real-time kinematic (RTK) GPS Device and the Canopy Height Sensor Pod, were developed to solve two issues in our mangrove field expeditions workflow. The methods to receive both of these measurements were very primitive and resulted in inaccurate measurements. Building these two devices enable us and our collaborators to gain accurate measurements of mangrove geolocation and mangrove canopy height to reduce error in measurements such as mangrove biomass estimations.
In order to take measurements of each mangrove site, scientists need to track where they have been and their current location. Geolocation provides important information about which sites have been surveyed and how far the researchers still have to go. Currently, we are using a portable, handheld Garmin GPS to document coordinates that is easy to carry into the field and water proof. However, this GPS has an error range of about 10 meters which is further exacerbated by working under dense vegetation (Support.garmin.com). Since the measured areas are only about 4 m2, this range yields large deviations from the true result.
Previous studies that monitor and map mangrove forests do not keep precise GPS points of their surveyed sites, with an error range of up to 15 meters (Simard et al, 2012, Jachowski et al, 2013). By collecting accurate geolocation data, we can reasonably track all the surveyed plots that have been conducted in the forest to verify our drone collected data.
We proposed using an RTK GPS in the mangroves to obtain measurements down to centimeter level precision This device is extremely accurate, but it is also susceptible to water and mud (Errors and (RTK), 2019), and is too cumbersome to use in the mangrove ecosystem. Consumer grade outdoor GPS devices can withstand the difficulties of the mangroves, but these devices lack the ability to give an accurate geolocation of mangroves under the thick canopy. Standard survey grade RTK systems Our system stores the device in a water-proof pelican case, protecting the RTK and provides researchers a relatively low-cost, robust method of obtaining real time location.
Canopy height measurements of mangrove forests can be used to estimate the biomass of the forest - a variable with direct impact on the economic and environmental impact of the forest. Despite the importance of biomass measurements, existing methods are prohibitively expensive, unacceptably inaccurate, or unwieldy to use.
One important variable for the biomass calculation is the height of the canopy. To cross-validate the data calculated using the Digital Elevation Model. The current method of measuring canopy height involves the stack of PVC pipes to get an approximation through the summation of PVC pipe length. We would then come up with to approaches to measuring canopy height: top-down and bottom up.
Current methods of estimating canopy height used by other facilities involve using an airborne drone equipped with Light Detection and Ranging (LiDAR) (Simard et al., 2006), (Lee and Fatoyinbo, 2015), (Fatoyinbo and Simard, 2012). However, each study reports root mean squared errors ranging from around 1 meter (Lee and Fatoyinbo, 2015) to 3.55 meters (Fatoyinbo and Simard, 2012). Although our current PVC approach is too unwieldy to continue and fails to scale to canopy heights larger than around 10 meters, it results in a much more accurate measurement, within 1 meter. Additionally, drones powerful enough to deploy large LiDAR systems tend to be too expensive to use on a large scale, and may require up to three operators - the pilot (flies the drone), the data manager (handles the payload and records data), and the visual observer (watches for potential safety issues and assists pilot with general operations). Quality LiDAR drone systems may cost upwards of $50,000, potentially reaching as much as $250,000.
Our work involves proposing a low-cost, low-power, and accurate method to measure canopy heights. We present a sensor pod deployable by telescopic pole to improve upon our existing methods and allow for better scalability.
The RTK GPS Device is a system that can withstand the conditions that exist within the mangrove ecosystem. Typical devices, such as the handheld Garmin GPS, are portable and easy to bring in these environments. However, they yield an error range up to 15 meters and significantly skews our data as the surveyed quadrats are only 4 m2. Our customized RTK GPS system combines the accuracy of an RTK device and encloses it in a water-proof case that can be brought into the mangroves, improving the precision of our coordinates while keeping the entire system robust.
We used an RTK-2 module to continuously obtain GPS coordinates in real time and powered it by an 6600 mAh battery. These components were stored within a laser cut front plate, using buttons to control the system and a Raspberry Pi with a screen to view. The entire system is enclosed within a weatherproof Apache case for protection against mud and water.
We also designed an interface that can be accessed through mobile phones to provide an easy way of tracking current location and marking previously surveyed locations. This component is written in Python, using flask and Leaflet to put the GUI on a webpage with detailed maps. Future work will involve putting the maps offline so they can be used anywhere and including more options for tracking.
The canopy height estimation component of the project involved creating a sensor pod deployable by telescopic pole. The sensor pod acts as a wireless access point, allowing any browser-equipped devices to connect to it. While the operator elevates the pod, a real-time video feed displays the pod’s progress. Once the canopy has been reached, the operator may capture altitude measurements and store them to disk for later processing.
At the center of our architecture, we use a Raspberry Pi Zero W as the controller of the whole system. The Pi provides a relatively low-powered microcontroller capable of interfacing with each of the various sensors. Additionally, the Pi Zero W comes equipped with built-in WiFi capabilities, allowing us to configure the Pi to act as an access point to connect to with external browser-equipped devices.
We use a Lithium Ion Polymer Battery, providing 3.7V at 500mAh, to power our system. Since the Pi expects an input voltage of 5V, we use the PowerBoost 1000 Charger - Rechargeable 5V Lipo USB Boost as a power adapter to step the input voltage up to 5V. This power adapter also allows us to plug in to a charging cable to recharge the battery between uses or during use.
Our altitude measurements are taken by the MPL3115A2 - I2C Barometric Pressure/Altitude/Temperature Sensor. This sensor communicates with the Pi over an I2C interface, and can provide within 0.3 meter accuracy.
We use the Smraza Raspberry Pi Camera Module, a fish-eye camera to provide a real-time video feed from the point of view of the sensor pod.
Finally, we are still in active development of the external structure and docking mechanism. We have prototyped with a wire mesh frame and water bottle, and produced the first iteration of a 3D printed housing with a built-in tapered docking mechanism. Future work will entail 3D printing a more refined structure that better accounts for airflow and protection of the various components before the system is deployed in the field for actual use.
Fatoyinbo, T. and Simard, M. (2012). Height and biomass of mangroves in Africa from ICESat/GLAS and SRTM. International Journal of Remote Sensing, 34(2), pp.668-681.
Lee, S. and Fatoyinbo, T. (2015). TanDEM-X Pol-InSAR Inversion for Mangrove Canopy Height Estimation. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 8(7), pp.3608-3618.
Simard, M., Zhang, K., Rivera-Monroy, V., Ross, M., Ruiz, P., Castañeda-Moya, E., Twilley, R. and Rodriguez, E. (2006). Mapping Height and Biomass of Mangrove Forests in Everglades National Park with SRTM Elevation Data. Photogrammetric Engineering & Remote Sensing, 72(3), pp.299-311.
Support.garmin.com. (2019). GPS Accuracy | Garmin Support. [online] Available at: https://support.garmin.com/en-US/?faq=aZc8RezeAb9LjCDpJplTY7 [Accessed 13 Jun. 2019].
Jachowski, N., Quak, M., Friess, D., Duangnamon, D., Webb, E. and Ziegler, A. (2013). Mangrove biomass estimation in Southwest Thailand using machine learning. Applied Geography, 45, pp.311-321.
Errors, C. and (RTK), R. (2019). Real-Time Kinematic (RTK). [online] NovAtel. Available at: https://www.novatel.com/an-introduction-to-gnss/chapter-5-resolving-errors/real-time-kinematic-rtk/ [Accessed 13 Jun. 2019].