A Flexible, Cloud-based Sensor Data Service Platform
TimeTuesday, July 246:30pm - 8:30pm
DescriptionPlant phenotyping research has grown rapidly in the past decade. With the wide adoption of GIS enabled sensors, devices, and instruments, the amount of data collected have grown rapidly as well, especially data collected in the fields. On the other hand, the transfer and management tools for such field data are lagging behind. Many researchers still rely on the old way of saving data to disks from devices in simple text files or spreadsheets before copying them to their desktop or department server for analysis, leading to inaccurate or lost data and low level of productivity. To help filling this gap, we collaborated with a plant phenotyping sensor research group at Purdue University to develop a cloud-based data service platform for field sensor data ingestion, management, processing, and analysis. Deployed on XSEDE’s JetStream VM environment, the platform is built based on an open source software stack, including MongoDB for sensor data management, Node.js for geospatial data preprocessing and data access REST API, and an online visual analytics application implemented as a web component on a HUBzero platform. End users can query, aggregate and visualize plant health data such as chlorophyll, fluorescence, nitrogen, water, normalized difference vegetation index (NDVI), nitrogen reflectance index (NRI), and soil and plant analyzer development (SPAD) at different temporal and spatial scales in real time. The platform is flexible and can be applied to support similar use cases that utilize field data collected by mobile devices, a practice being adopted by a growing number of research projects. In this poster, we will describe the overall system architecture, software implementation, and how it has been used in supporting plant phenotyping sensor research in the field and decision making by farmers and stake holders.