Reviewing Satellite-Based Agricultural Monitoring Systems for Africa
The introduction of satellite Earth Observation (EO) data has immensely broadened and deepened researcher's agricultural monitoring capabilities. These benefits of EO data are dependent however upon having the knowledge to process it into usable products and analyze it. This accessibility and expertise can vary dramatically across countries and regions.
NASA Harvest’s Africa Program Lead, Catherine Nakalembe, recently led a team of researchers exploring this variability by reviewing a number of “open-access operational agricultural monitoring systems” for several African countries. The open access paper, A review of satellite-based global agricultural monitoring systems available for Africa, draws upon the expertise of EO and agricultural experts from NASA Harvest; Geography Department, University of Maryland; World Food Programme, UN; IGAD Climate Predictions and Applications Center; European Commision, Joint Research Centre; Climate Hazards Group, University of California, Santa Barbara; Food and Agricultural Organization, Viale delle Terme di Caracalla, Rome; and GEOGLAM. It compares and contrasts the various systems, the data they use, and their resulting outputs ranging from vegetation indices to crop masks, making it a good resource for individuals and organizations who lack the technical expertise yet are interested in utilizing EO data for agricultural monitoring.
Dr. Catherine Nakalembe demonstrating agriculture monitoring technology in the field.
2019 saw Eastern and Southern Africa experience a series of droughts, floods, landslides, and a desert locust invasion, many of which were compounded by political instability in several countries. These events led to decreases in crop production, but their impacts are rarely properly assessed because traditional methods of agricultural monitoring including field assessment, crop cutting, and farmer surveys are expensive and difficult to scale. Many of these disasters can be forecast and the corresponding effects can be analyzed through the use of EO data, saving significant labor resources in addition to reaping the benefits of early warning that can include soothing markets, allaying food supply crises, and more effectively directing humanitarian assistance.
EO data is big data, consisting of different sources, sensors, and resolutions and therefore requires massive computing capabilities to process and analyze at a significant scale. Recent advances in cloud computing infrastructure like Amazon Web Services, Microsoft Azure, Google Cloud and systems like Google Earth Engine have allowed this processing and analysis to be performed in the cloud instead of a user’s desktop, decreasing financial and labor costs and increasing the development of EO products and applications as well as the possibility to integrate in under-resourced organizations.
These advances have not been equally realized around the world however. Many countries in Sub-Saharan Africa for instance lack the necessary technical expertise and computing infrastructure, as well as the policy frameworks to develop this expertise and infrastructure. This has limited the government’s ability to develop EO-processes that can complement traditional methods of agricultural monitoring like resource-intensive field inspection and farmer surveys. Despite this inability, many governments could, with the right knowledge, improve their agricultural monitoring efforts by leveraging open-access EO-based systems.
This published study provides this knowledge. By reviewing a number of these systems in an accessible manner, Dr. Nakalembe allows agronomists and officials charged with food security who lack the requisite training to understand the different platforms, their underlying datasets, and examples of how they can be applied. Systems selected for review had to be open access and web-based, have basic maintenance handled by the system’s original developers, have manuals available online, and be targeted to agricultural analysts in agricultural ministries or similar officials. This resulted in a list of five systems.
One of the selected systems reviewed by Dr. Nakalembe and co-authors is the Global Agricultural Monitoring System (GLAM) system. Originally developed in 2005 and updated in 2019, GLAM is a “global agricultural monitoring system that provides timely, easily accessible, scientifically validated, remotely sensed data, and derived products and doubles as data analysis tools for crop condition monitoring and production assessment.” In other words, GLAM accesses, processes, and displays previous and current global EO data and resulting products including vegetation condition, temperature, precipitation, and soil moisture.
Users can view the entire globe or zoom into a region of interest. Selecting a specified time period (~7 days) going back to 2002, users can view a number of variables. These variables, or integrated products, include 8- and 16-day NDVI (vegetation condition) datasets from NASA’s MODIS mission, CHIRPS precipitation data from University of California Santa Barbara Climate Hazards Center, Copernicus Land Service’s soil moisture data, and temperature data from NASA’s MERRA-2 mission.
Adding to its utility, results can be limited to areas known to produce certain crops. Applying these “crop masks” will show selected products only in zones known to grow certain crops, which includes maize, soybeans, rice, winter/spring wheat, or a generic cropland. This cuts down on processing time and allows for simplified visualization.
Further the GLAM system permits statistical calculation of the available products. For instance, a user can view how precipitation varies over a two year period in areas where winter wheat is known to grow. This allows users to view how precipitation varied over a given year, while also allowing for comparison between the same months in different years. Based in the cloud, all of this computation and visualization is capable with institutions that lack the computing infrastructure. This ease of use allows GLAM to be integrated into national monitoring systems, including Brazil, Tanzania,and Uganda among others.
GLAM system displaying calculated statistics for NDVI as a time series and histogram.
GLAM additionally is part of the consensus-building process for the GEO Global Agricultural Monitoring (GEOGLAM) Crop Monitor for Early Warning (CM4EW). Crop Monitor is a report produced by the main food security monitoring agencies of the world. Established in 2016, it addresses the “pressing need for enhanced reliable and vetted information over countries at risk of shortfalls in production”. Crop Monitor reports improve agricultural humanitarian aid delivery, better inform food security decision makers, and bolsters policy implementation.
The Crop Monitor for Early Warning May Bulletin showcasing conditions for various commodity crops around the world.
While incredibly valuable, Dr. Nakalembe notes that there remain a number of challenges to implementing agricultural monitoring systems like GLAM. On the user end, many countries may lack the internet connectivity required to access these systems. If physical infrastructure is sufficient, users may also struggle to integrate these systems into current reporting methods. From the developer end, having up-to-date and accurate crop masks can be challenging, especially at the high spatial resolution required for effective analysis. Dr. Nakalembe writes that without concentrated global effort to build the necessary resource, technical, and bureaucratic capacity, the potential of these systems will never be fully utilized in countries who need them the most.
Further discussion of GLAM, along with the other 4 reviewed systems, can be found in A review of satellite-based global agricultural monitoring systems available for Africa.