We are delighted to announce that we have just released the GEMS Grid code library, where the code is under the open source Apache 2.0 license, which allows anyone to use the code for commercial or non-commercial purposes – you simply need to provide attribution to GEMS Informatics when you use or modify it.

Just before we at GEMS Informatics started developing Application Programmer Interfaces (APIs) in agriculture for GEMS Exchange, GEMS geospatial expert Jeffery Thompson worked with others in the GEMS team and colleagues at NSIDC to develop the GEMS Grid, a hierarchical discrete global gridding system. This Grid has allowed us to provide data sets at different resolutions ranging from 36 km to 1 m, and still have them remain functionally interoperable. The interoperability is possible because we have written the code to allow users to project data onto the grid, aggregate data to coarser resolutions, and, notably, also disaggregate data to finer resolutions. The latter operation is ordinarily a difficult problem, but is made easier, as I discuss below, since we enable the users of our code to thoughtfully address it in a standardized, replicable way.

Many problems in agriculture (e.g., understanding the spatial location of crop production) require equal area parcels of land to do proper calculations. Working with strict lat-lon coordinates won’t suffice as areas near the equator are significantly different in size as areas near the poles. The GEMS Grid preserves equal-area assumptions as it divides land, so you can do these calculations with confidence, and preserve aggregation-disaggregation consistency in the data, even if you are not a GIS expert.

So let’s look at the 5 options for disaggregation that GEMS geospatial developer Olena Boiko included in the GEMS grid toolbox, schematically described in the figure she developed above.

Option 1. Value transference. 
In this case, if you were to subdivide a 3 km² resolution grid cell into 9 x 1 km² cells, this option would be appropriate for any value that is deemed roughly constant throughout the area applied. Examples would be rainfall in inches or grain yield in bushels / acre.

Option 2. Even value division. 
Sometimes the quantity measured in a cell represents a cumulative value for the area in which it is reported. In this case, if the parent cell is homogenous, then splitting it up into 9 equal-area pieces would require that you divide the value in each equivalent cell by a factor of 9. Examples where this selection makes sense include grain production in bushels, crop acreage, and population.

Option 3. Value transference with a mask. 
This one is similar to Option 1 except we are no longer making the assumption that the distribution of values in the parent cell is spatially homogeneous. For example suppose you were measuring grain yield, but you knew that 3 of your nine cells had buildings occupying them (see white areas in the Figure). In this case you only transfer your values to 6 remaining cells (colored peach) that have arable land. Cells are binary with this option (i.e., either allowed a value or not).

Option 4. Even value division with a mask. 
Analogously, you can mask out cells in the value division case when you know that your daughters cells are not all equal. This is just like the case in Option 3, except you divide your parent-cell value evenly by the number of viable daughter cells. In this pictorial example, there are 6 viable daughter cells, so each gets a value of 900/6 = 150. This would be appropriate if you were computing grain production in bushels and you had a total value that needed to be split up across the 6 arable daughter parcels.

Option 5. Flexible division with a mask. 
This scenario is the most flexible, and allows the user to create a master mask with arbitrary weights at each daughter cell. It allows you to block off daughter cells entirely, and prescribe the relative weights of all remaining daughter cells. This is ideal for situations where you are allocating crop distributions and you want to avoid certain land use features (e.g., lakes, forests, housing) and probabilistically distribute the remaining crop areas (e.g., with higher probability near soils with a high SSURGO National Commodity Crop Productivity Index).

I’m confident these flexible disaggregation tools will provide much easier, more accurate, and replicable solutions for your particular spatial analytic problem. So please give them a try!

​This activity supported in part by MnDRIVE Global Food Ventures, University of Minnesota

Assessing BMP Effectiveness for Water Quality Outcomes with Remote Sensing

Virtually everyone in Minnesota and bordering states whose profession has a direct stake in Water Conservation assembled in Downtown St. Paul’s RiverCentre October 17-18 for the annual Water Resources Conference. I had the honor of chairing a Special Session at the conference titled “Assessing BMP Effectiveness for Water Quality Outcomes with Remote Sensing.” The session featured two very foundational efforts that we are attempting to bridge together in partnered research at GEMS: (1) The Minnesota Department of Agriculture’s (MDA) Minnesota Agricultural Water Quality Certification Program (MAWQCP), directed by Brad Jordahl Redlin, and (2) the Department of Forest Resources’ remote-sensed Water Quality assessment for our 10,000+ lakes, led by Leif Olmanson.

The 90-minute session was organized so that each of the three speakers (one had dropped out last minute for logistical reasons) had 20 minutes to speak plus two minutes for burning questions. At the end the audience had 15 minutes for questions with the entire panel of speakers. There were about 30 very inquisitive people in the audience, who peppered the speakers with questions throughout.

Updates on the MDA’s certification program

Brad Jordahl Redlin kicked off the session with an outstanding talk that outlined the principles on which the certification program he founded and directs was created, and highlighted the progress that has been made. Literally over a million acres have now been certified, spread across over 1400 producers. In exchange for regulatory certainty for a period of time, these producers allow his government agency to (1) assess their farm practices, (2) suggest new management practices (sometimes with additional incentives such as subsidized equipment from a grant), and (3) carry out audits of their new practices. The program, which has been operational for nearly a decade, is well beyond the stage of corralling early-adopters. New participants span the spectrum of producer demographics in age, farm size, and more.
 

Scaling lake quality measurement

Leif Olmanson stepped up to review an activity that he first spearheaded twenty years ago – using remote sensing to measure the water quality of Minnesota lakes. In the interim, he has overseen steady progress using new generations of satellites (from Landsat to Sentinel), multiple metrics (clarity, chlorophyll, organic matter), and the frequency of data reporting (from once every 5 years to ~weekly lake snapshot and monthly pixel-level composites of all 10,000+ lakes). Data is made available to the casual user via a friendly Lake Browser interface, and in numeric form pixel-by-pixel for data wizards on GEMS Exchange.
 

GEMS bridging the two programs

David Porter rounded out the session discussing the improvements he has made in creating algorithms that properly remove clouds and aerosols from the satellite images. This is a prerequisite for feeding all of the regularly spaced (~5 day) images of a field through a growing season into a sophisticated Recurrent Neural Network (RNN, a type of Machine Learning algorithm) that his colleague Anubha Agrawal has been preparing. As described in this case study, David, Anubha, and I intend to use the RNN trained on known, certified management practices (e.g., strip till / no till, cover crops) from the MAWQCP to make a prediction of what practices growers are making on uncertified farmland. We then will trace these producer fields to the downstream lakes in each watershed, and see what covariates (e.g., slope, soil type, distance to waterway) affect the correlation of % farm practice adoption vs. downstream lake quality in a watershed.

​This activity supported in part by MnDRIVE Global Food Ventures, University of Minnesota

GEMS’s new NSF award

Training ag and food scientists how to make the most out of supercomputers

Do you write code (maybe in R, maybe in Python)? Maybe you’re an agri-food scientist in a lab, or in a similar field in industry. Perhaps you’ve tried to write some code to get your work done, figuring you can run it on your laptop. But when you churn on that data you find after two weeks of keeping your laptop open to run the code that it’s probably not going to finish any time soon… Epiphany. You need to do this in a smarter way! Well you’re in luck – GEMS was just awarded a grant from NSF titled “Cyber Training: Pilot -- Breaking the Compute Barrier, Upskilling Agri-Food Researchers to Utilize HPC Resources” and you are exactly the audience we wish to reach with the courses we are developing.

The Team

The course material is being developed by an interdisciplinary team including:

• Instructor Joe Axberg, who works on supercomputing infrastructure by day but is also a part-time instructor after hours in the IT Infrastructure Program in the College of Continuing and Professional Studies.
• Project mastermind, Ali Joglekar, an agricultural economist by trade with real-world experience working with smallholder farmers in East Africa
• Kevin Silverstein, a cofounder of GEMS Informatics and expert bioinformaticist with specialties in molecular plant-microbe interactions and large-scale informatics problems
• Jesse Erdmann, systems architect and guiding force behind much of the inner workings of GEMS infrastructure.
• Ben Lynch, Director of MSI, with a long track record of innovating in the scientific computing domain

Scope

The proposal identifies 7 separate course modules that each contain a significant number of concepts, representing a broad swath of content to help up-skill agri-food researchers to use HPC:

1. Introduction to High-Performance Computing (HPC) for Agri-Food Researchers
2. Introduction to Cloud Computing for Agri-Food Researchers
3. Hands-On: Use HPC to Analyze More Data and Faster
4. Leveling-Up I: Expanding HPC Skills for Agri-Food Research
5. Computer Science for the Agri-Food Researcher
6. Leveling-Up II: Advanced HPC Concepts
7. High Performance Workstations and Servers for Agri-Food Researchers

It is expected that the number of modules proposed and the content within each module may change as we build out the course and modules during the development phase(s). The goal of our iterative course development process is to refine this broad set of concepts and craft cohesive modules each with a duration of 2-6 hours of lecture materials (depending on the module). Some modules will span multiple weeks. Several rounds of purposeful learner feedback is integral to our course development process. 

The course modules will also be “stackable” allowing learners to have some ability to “pick and choose” modules in line with their competencies and interests.

Timeline for delivery

As the diagram indicates, we are planning to offer each HPC for Agri-Food Researchers course over a 10-12 week period of time. The proposed course structure, in line with our other GEMS Learning offerings, includes asynchronous and synchronous instructional elements. Lecture materials will be delivered via videos and readings, which students are expected to review ahead of the weekly 1.5 hour class session. This instructor-led class time is primarily dedicated to synthesizing and reviewing the content being taught. To solidify learning objectives, students will be expected to engage in regular assessments and hands-on learning exercises outside of the weekly class session. We estimate that students will be responsible for 6+ hours/week of self-directed learning activities, depending on their skill-levels.

You can influence our content!

The good news is we are still in development, so you can influence these course offerings. Please fill out this survey now to make your voice heard!

​This activity supported in part by MnDRIVE Global Food Ventures, University of Minnesota

Behavioral economists are studying “nudges” to prompt behavior change in agriculture. Accounting for production economics can help make nudges more effective for reducing agricultural run-off.

Water pollution caused by agricultural nitrogen runoff is a pervasive problem worldwide. In a recently published paper (Food Policy), GEMS researchers (Yuan Chai and Philip Pardey) teamed up with University of Western Australia collaborator (David Pannell) to provide a new slant on this age-old problem. Combining insights from both production economics and behavioral science opens up new and practical avenues for nudging farmers towards eco-friendly practices that can mitigate water pollution. 

Production Economics Insights

The authors highlight three key empirical findings that offer opportunities to reshape farming practices and reduce nitrogen pollution. First, many farmers tend to apply more nitrogen than necessary, driven by yield maximizing or other motives. This can lead farmers to use more fertilizer than is economically optimal, which not only incurs avoidable costs but also exacerbates the water pollution problem. Second, farmers may imagine that, contrary to the evidence, using more fertilizer is a risk-reducing strategy. Thirdly, the profit-to-fertilizer-rate relationship is flat near the optimum, suggesting that farmers can curtail their fertilizer usage without incurring significant, if any, private costs (through foregone income).

Nudges

Leveraging these three insights, the authors draw on emerging findings from behavioral economics to identify a number of potentially transformative approaches to curbing agricultural pollution. The authors make specific suggestions for ways in which behavioral science could contribute in these areas, including understanding farmers’ perceptions about nitrogen, yield and profit (detecting biases), understanding the influences of these biases on behavior, understanding farmers’ perceptions of and preferences for different policy responses, and designing behavioral nudges that capitalize on the specific insights we have highlighted regarding the production economics of nitrogen fertilizer. Nudges could include strategies that aim to correct biases in perceptions (e.g., about yield increases or risk reductions from applying N),or tap into social and psychological factors that influence farmers’ decision making.

Behavioral changes that are deemed nudgeable (i.e., where farmers bear little or no cost) have a lot of merit. The nudges revealed through this research could go a long way to cleaning up water bodies in ways that do not undermine food security or farmer livelihood concerns: a rare win-win result.

​This activity supported in part by MnDRIVE Global Food Ventures, University of Minnesota

Making Sense of Agriculture from Minnesota to Malawi

GEMS Sensing has coupled world-class engineering design and build expertise with first-rate agricultural science smarts to realize a robust, extensible sensing system for agri-food research anywhere in the world.

Since early 2020, GEMS Sensing has partnered with the Centre for Agricultural Transformation (CAT) in Malawi to pilot the local deployment of GEMS research-grade weather stations and its associated data visualization and sharing tool. Piloting a network of weather stations across Malawi involves much more than working through the logistics of manufacturing, delivering and deploying the necessary hardware and software. It also involves identifying partners willing to trial the weather stations on their research stations/farms and training local CAT personnel to enable the deployment, particularly in remote rural areas with uncertain cellular connectivity. Working through the CAT provides a unique opportunity to engage with private and public agribusiness throughout Malawi to test GEMS Sensing technologies in a low-income, smallholder farmer, tropical environment. 

Over the past five years, the GEMS Sensing team has used feedback from our field partners to develop ever-improving scientific-grade, plug-and-play sensing devices coupled with a secure web-based informatics service. Weather stations aren’t new, but GEMS Sensing has created a highly customizable system that is built specifically for the varied sensing needs of researchers and their collaborators. The system's core logger box can be used for scientists with vastly different sensing requirements, from a full-stack meteorological station supporting a host of on-station or in-field crop performance studies to more specialized applications aimed at monitoring the below ground conditions that affect plant growth and health. The system's configurable and easy-to-use dashboard enables the integration, exploration, and sharing of real-time sensor data from multiple, perhaps distant, sites through a single portal.

Tracking the real-time, micro-level weather and edaphic patterns impacting agricultural production is crucial for a whole host of reasons, including advancements in breeding, agronomy, and extension efforts and the livelihoods of the farmers responsible for this production. These types of data are used for modeling weather events over large areas (e.g., rainfall and temperature) or informing management decisions like when and how much to irrigate a crop. However, being able to collect, store, clean, explore, share, and analyze these data from multiple sites, isn’t always straightforward. With this motivation in mind, the team has coupled world-class engineering design and build expertise with first-rate agricultural science smarts to realize a robust, extensible sensing system for agri-food research anywhere in the world.

GEMS Sensing in sub-Saharan Africa

For the Malawi pilot project, we worked with a South African-based manufacturer to produce the GEMS Sensing stations. The full-stack weather stations are configured to collect eight parameters every 15 minutes: air temperature, humidity, barometric pressure, rainfall, soil moisture, soil temperature, wind speed, and solar radiation. Data are sent via cellular network to a cloud-based database, processed for quality, and made accessible to partners via a user-friendly, web-based portal.

The primary objectives of the pilot exercise were to better understand:

1. Tropical environmental impacts on the sensing hardware;

2. Total cost of ownership from a user perspective (i.e., costs of acquisition, installation, and maintenance) involved in deploying digital ag weather sensors in a low-income context with limited infrastructure support; and

3. Malawian agribusiness’ demand, willingness to pay and potential use cases for networked weather stations (or related real-time sensing needs).

To date, the CAT team has deployed nearly 40 solar-powered weather stations across the spatially and temporally variable agro-ecologies that affect agricultural outcomes throughout the country. The stations continue to stream real-time data from experimental research stations, commercial farms, and smallholder farms. The pilot project is set to wrap-up in December 2023, though the CAT may continue to operate some of the weather stations beyond that time.

Learnings and Lessons

Scaling up a pilot project in Malawi from our operations based in Minnesota during the height of the global pandemic was not without its challenges. The original game plan was to travel with a handful of sensors to Malawi so that our technical team could work closely with CAT personnel and deployment partners to scope out the entire constellation of technical, human capital, cell connectivity, local logistics, device assembly and deployment partner protocols and other factors that could likely affect the odds of a successful outcome. Based on this test deployment we had envisaged tweaking the technical specs, revising our training and related deployment protocols before scaling up deployment throughout the country.

COVID-19 had other plans. It shut down all travel to Malawi, disrupted supply chains for computer chips that were critical to fabricating the GEMS designed sensors, and complicated shipping logistics. Undeterred, we pivoted our whole operation to enable a remote deployment, which, despite our best efforts, was not without its own complications.

Persistence ultimately paid. As we began scaling field deployment in April 2022, our technical team became adept at diagnosing issues long distance, ably supported by a great ground team in Malawi. Fluky cell connectivity issues seemed to be associated with variable soil properties that were vastly improved by simply raising the logger box further off the ground. Unexpected down-time issues were eventually tracked to battery depletion problems that arose from locally assembled devices that sat turned on for lengthy time before travel resumption enabled field deployment to proceed. The fix, change out the run-down batteries.

All told, our GEMS Sensing devices exceeded our expectations, even withstanding Tropical Cyclone Freddy that hit Malawi in February 2023!

Our team visited Malawi in May and traveled with the local CAT sensing team throughout the Southern and Central regions to get first-hand feedback from a host of public and private partners who participated in the GEMS Sensing pilot project. We made site visits with DARS research scientists, commercial agrifood businesses such as Bayer, Pxyus and Global Seeds, and One Acre Fund which services smallholder farmers directly. These partners were uniformly enthusiastic about the ready access to automatically streamed and easily accessible localized weather data. They also highlighted numerous opportunities for leveraging the streaming weather data streams with various backend analytics to tailor informatics solutions to the particular and varied agri-food production problems they faced.

GEMS sensors are currently being deployed with success across four continents. However, this represented our first widespread test of the system in a logistically challenged, low-income African agricultural setting. Prior to the pilot we had all sorts of questions as to whether or not the system would even work, and if so, how reliably and securely. Importantly, would anyone find practical value in such a sophisticated, real-time sensing system in a country dealing with many infrastructure and technological constraints.

Weather matters a lot for agriculture, whether you are farming in Minnesota or Malawi. While Minnesota farmers have access to all sorts of weather related data products, locally accurate, real-time data is still relatively scarce. In Malawi, rurally-relevant weather data of any sort is largely absent. The last time we checked, NOAA (National Oceanic and Atmospheric Administration) live stream weather data from just one site in Malawi, the Blantyre International airport! This pilot demonstrated the potential for changing this reality for Malawian farmers and the numerous public and private agencies that support their operations.

Even with its many hurdles and holdups, this pilot exercise revealed the feasibility and the latent demand for access to locally sensed weather data. With the lessons learned from our pilot deployment we are excited about the possibility of using GEMS Sensing’s data generation and analytical capabilities to support data-driven decisions that strengthen existing and emerging value chains in Malawi, as well as unlock new agri-food value-chains throughout the world.

This posting was produced as part of the Centre for Agricultural Transformation, an effort led by Land O’Lakes Venture37 and funded with a grant from the Foundation for a Smoke-Free World, Inc. (“FSFW”), a US nonprofit 501(c)(3) private foundation. The contents, selection and presentation of facts, as well as any opinions expressed herein, are the sole responsibility of the authors and under no circumstances should they be regarded as reflecting the positions of FSFW.

• Centre for Agricultural Transformation
• Land O’Lakes Venture37
• Foundation for a Smoke-Free World

​This activity supported in part by MnDRIVE Global Food Ventures, University of Minnesota

What skills should you be building for a career as a data scientist in Digital Agriculture?

Everything we do has a spatial and temporal component. These days GIS skills are a big plus!

I often have students come to me saying “I’m passionate about the agri-food sector and I want to do data science. What skills do I need to land a satisfying job, and ultimately a career, in this area?” So I figured I’d take the opportunity to share what I typically say to them here, for the benefit of all those out there with similar goals and interests. Key items are in bold.

First off, there is so much unstructured data out there, in disparate formats and locations that you aren’t going to get far without a programming language under your belt. And in this field, that really boils down to Python and/or R. Sure, Chat-GPT can write code, but trust me, it’s not there yet.

Next, everything we do has a spatial and temporal component. These days GIS skills are a big plus! You don’t have to be a GIS expert, but you should know what a coordinate reference system and datum are, understand the limitations and practicalities of aggregation and disaggregation to different levels of resolution, and be facile with vector and raster manipulations.

Data science applied to any domain involves statistics and modeling. Moving beyond point estimates with p-values and understanding Bayesian statistics will get you far. And an understanding of databases (e.g., relational, graph, or columnar) can also be useful.

Finally, many datasets are incredibly large, so analyses often can’t be performed on your laptop. So familiarity with doing analyses on High-Performance Computing (HPC) infrastructure can be critical. These systems have mechanisms in place for you to schedule jobs to be run across multiple processors in tandem with other people’s jobs, and there are conventions and rules of etiquette for that.

Depending on the positions you have in your wish list, you may want to be sure to have either a Masters level degree or Ph.D. Masters should suffice if you’re happy to have someone else identify and devise the scope of the problems you work on. If you want to do pure R&D and define your own problems, a Ph.D. will likely be necessary.

If you're lacking some of these skills or qualifications, there are numerous data science degrees offered across the country, as well as short instructional modules such as those included in GEMS Learning.

​This activity supported in part by MnDRIVE Global Food Ventures, University of Minnesota