top of page

Remote Sensing & GIS

Data Driven Pursuit of Regenerative Agriculture

Regenerative agriculture is the most recent in a series of movements in agriculture emulating or guiding natural processes by treating the agricultural unit as an ecosystem. Regeneration is defined in terms of the return of key nutrients to the land through the encouragement of critical chemical cycles in the topsoil, the elimination of certain normative industrial agricultural practices, and through a holistic orchestration of the farm as an ecosystem, emulating nature. This case study seeks to address the following critical question: how do we scale regenerative agriculture commercially beyond individual properties and projects?

The current Regenerative Organic Certification (ROC) framework evaluates a farm’s regenerative capacity with three benchmarks: its soil health and land management, its animal welfare, and labor practices. In an effort to quantify their regenerative practices in preparation for their certification with the ROC, Paradise Farms is partnered with Precision Ecology, a local remote sensing and environmental solutions company, to investigate the potential for multispectral soil spectroscopy as a tool for measuring and assessing soil regeneration.

Over a span of two months Precision Ecology did routine multispectral mapping using the Micasense Altum Photonic Sensor over Paradise Farms in Homestead FL. We captured about 5000 photos for each flyover of Paradise Farms using the Micasense Altum Sensor. This is around total 16 gigabytes of data. This data is processed in our photogrammetry software and then further assessed using different vegetation and soil index calculations. Our focus was on understanding the specific use cases of different analytical tools applicable to multispectral data over farms seeking to implement regenerative systems. Paradise Farms has a history of being a farm that complies with biodynamic and organic certifications.

Pictured Above: Paradise Farms manager David Laws (left) being shown data collected by Precision Ecology operations manager Whitaker Redgate (right).

Regeneration is currently assessed as a mosaic of interconnected criteria due to the complexity of such systems. Thus it is difficult to define the process in terms of specific factors and even more challenging to distinguish between their importance. The current ROC framework, the most comprehensive to date, is fifty-four pages in length. Developed in conjunction with Patagonia, the framework measures regenerative capacity in three levels: bronze, silver and gold. Within these levels are increasingly restrictive requirements which a farm must be in compliance with to demonstrate gold regeneration. It is difficult to define a regenerative system as a tiered measurement. Within the ROC framework there is room for farms to move up the scale of regenerative classifications in future return visits with improved compliance with the items defined. In some cases this requires adding practices and in others it means demonstrating a commitment to the regenerative system.

As Dr. Isaac Mpanga of the USDA National Institute of Food and Agriculture remarked on the importance of data in agriculture, “farming is best done by learning from past mistakes.” Strong records of agricultural data can allow the farmer to be scientific about improving his techniques. Learning from “past mistakes” is important in regenerative agriculture specifically pertaining to the field, i.e., monocropping, pesticides, GMOs, tilling and inhumane treatment of animals and farmers. Data is key to not only demonstrating the effects of regeneration over time, but even further it is the key to learning from its practitioners and improving these practices with time and experience.

Multispectral imaging combines the analysis of a number of wavelengths of light, in our case six, both detectable and undetectable to the human eye. Industry leading MicaSense Altum MSI sensor detects Red, Green, Blue, Infrared, Near-Infrared and Thermal images in each location photographed in its flight. This data shows the photosynthetic reflectance being given off by vegetation’s chlorophyll, demonstrating greater health in terms of greener shades. Where the multispectral imaging captures direct soil data, these same wavelengths display key indicators of soil health by modeling organic versus inorganic compounds. This data can then be analyzed using the computing and machine learning tools described in “how can we standardize this data”.

For data to serve as an effective tool, we must identify and define standardized metrics within many different parts of these complex systems. Therefore it is critical to collect large amounts of data and to look for patterns in the variations demonstrated in order to identify key metrics for standardization. Within the MSI analysis software employed there are a number of critical tools used to identify different aspects of the data being reported in terms of what it means in the soil and vegetation. These include Soil Organic Carbon (SOC) Modeling, Colorized Index (CI), NDVI (Normalized Difference Vegetation Index), NDRE (Normalized Difference Red Edge Index), Soil Adjusted Vegetation Index (SAVI) and GNDVI (Green Normalized Difference Vegetation Index) and RF (Random Forest). Each of these categories is designed to hone in on a specific aspect of soil and vegetation health from the same dataset and will allow the data analysis process to be tailored to the specific needs and circumstances of each landscape system being assessed.

Shown Above: NDVI (Normalized Differential Vegetation Index) map of Paradise Farms

Utilizing the vegetation and soil indices discussed will allow machine learning programs to eventually be employed on large sets of regenerative data, allowing us to learn from our mistakes and identify important factors faster than ever before. With the enormous push simultaneously occurring in remote sensing, these metrics can then be employed on a larger scale to assess the broader impacts and extents of these regenerative systems through such institutions as the World Resources Institute’s Land and Carbon Sensing Lab and the Smithsonian Tropical Research Institute’s Ecosystems and Climate Earth Observatory which assess these same metrics of regeneration on continental and global scales using new innovations in satellite technology.

These powerful machine learning analysis tools being produced by leading researchers and institutions engaged in landscape and soil health studies are providing the framework for spectroscopy to emerge as the most efficient way of collecting and analyzing data on soil health. Even further, as regenerative certifications reach the marketplace, soil spectroscopy will enable the right certifier to make their process both more scientifically quantifiable and more tangible and actionable for practitioners and policy makers.

Soil spectroscopy is the measure of the reflectance of these wavelengths of light from topsoil to determine the health of that soil across a number of categories. This field has been rapidly advancing in recent years. One of the most important tools for SOC monitoring and studying is remote sensing, specifically employing the use of multispectral data containing infrared data. These wavelengths, which are invisible to the human eye, allow us to detect and make decisions based on soil conditions previously inaccessible. This data is so specific that at the local, high resolution scale achieved with our UAVs it can identify soil health based on the relative concentrations of various chemical compounds and availability of nutrients.

Shown Above: Paradise Farms manager David Laws flies a Precision Ecology drone over the property.

Soil organic compounds (SOCs) are the standard metric of nutrient compounds in the soil indicative of soil health. They are identifiable from soil spectroscopy data and indicate the health of soil by the level of concentration of these important organic building blocks. These SOCs are the building blocks of the spectroscopic data being collected on soil through remote sensing around the world, albeit at different scales.

The Consultative Group for International Agricultural Research (CGIAR) recommends both soil spectroscopy and the study of SOC concentration as tools for the advancement of global soil health and soil management. Currently, the analysis of data on soil spectroscopy and SOCs largely occurs at a continental and global scale through the use of satellites. One of the main arguments of many renowned field leaders is that while the effects of soil stewardship can be observed at those large scales, actionable information and helpful mitigation tools must be focussed locally in order to be actionable for practitioners and policymakers. In its most recent framework for assessing healthy soils, CGIAR became the first major body in the community of soil health certifiers to include spectroscopy as a critical tool for assessing soil health. The group cited the efficiency of spectroscopy, which allows significantly larger areas of soil to be analyzed faster and in a cost efficient manner.

Shown Above: RBG Map of Paradise Farms

In the months of September and October, Precision Ecology conducted flights over Paradise Farms to collect weekly date metrics. The data obtained reflected regenerative farming methods and displayed the advantages to regenerating practices. The findings are displayed in the figures below. Precision Ecology built this case study to identify and operationalize these metrics to provide insights on critical barriers to achieving regenerative agriculture at scale.

As a product of the Center for Social Change Miami we have a supportive base of experienced social change makers committed to helping make regenerative soil management a reality at scale. We are a team of driven, innovative individuals who believe that building a comprehensive framework for using remote sensing to standardize the evaluation of regenerative practices will have far reaching benefits in the years to come, as both remote sensing and regenerative agriculture achieve scale.

35 views0 comments
bottom of page