Designing a System for Digital Soil Assessments by Integrating Soil Sensor Technologies with Unmanned Aerial Systems

点击翻译按钮获取中文摘要

Pedometrics is an emerging field of soil science research that leverages existing technologies in remote sensing, Big Data analytics, and machine-learning to discover the environmental controls on soil variability and support sustainable soil management. My Soil-Landscape Analysis & Modelling Laboratory is the leading Canadian lab in this field, and our research has had national and international impact. A strong understanding of soil variability and soil security is anchored to our ability to address food security, water security, climate change mitigation, and sustainable resource management;43 and the provision of accurate soil information is linked to eight UN Sustainable Development Goals (SDG). Hence, the long-term vision and objective of the research program is to innovate technologies and apply data-driven approaches to ensure sustainable soil management across multiple spatial scales (e.g. farm, regional, and national-scales) using pedometric approaches. To address SDG 2: Zero Hunger, precision agriculture technologies are continuously adopted to reduce management costs and the environmental impacts of agriculture. Maximizing the potential of these technologies requires accurate and precise measurement and modelling of field-scale soil variability to develop predictive soil maps (PSMs) that inform sustainable soil management practices and variable rate technologies by optimizing the efficiency of agricultural inputs (e.g. fertilizers, water, and seeds) on a location-by-location basis. Farm-scale, pedometrics research has typically focused on ground-based sensors. However, the emergence of unmanned aerial vehicles (UAVs), coupled with the increasing accuracy and decreasing cost and size of sensors, provides a unique opportunity to revolutionize the modelling and monitoring of soil variability. The short-term objectives of this project are as follows: 1. To integrate, optimize, and evaluate multiple sensors on a UAV in order to predict soil attributes at the field-scale and to compare its effectiveness against ground-based sensor surveys. 2. To evaluate and compare geostatistical and machine learning techniques for producing high-resolution PSMs of key soil health indicators over contrasting cropping systems and landscapes. This project will innovate an automated sensor platform using hyperspectral, LiDAR, and gamma radiometric sensors that are mounted on a UAV system to perform high-resolution surveys of agricultural test areas. Using a range of machine-learning approaches, the sensor platform will predict the spatial patterns of soil health indices. This research will have economic and environmental benefits by minimizing agricultural input costs and environmental impacts. Canadians will benefit from cost-effective tools that document environmental goods and services; furthermore, this research provides the means to assess regenerative agricultural practices that generate carbon offsets by farmers.