Farmers were struggling long before the current health pandemic, but COVID-19 has now further disrupted an already devastating situation for farms. With the COVID-19 pandemic closing down farms’ largest purchasers (e.g. restaurants, schools, etcetera), many farms have been forced to throw away excess produce, dairy products, and eggs, causing extreme food waste and revenue loss. This problem is exacerbated by the fact that temporary foreign workers—which farmers rely on for labor— are unable to travel under the COVID-related border restrictions. Resultantly, many farms lack the labor needed to tend their crops. Thus, the combination of the food waste and labor shortage is causing farms to significantly struggle during this pandemic.
Our solution, FreshHarvest—an online platform that connects farmers, consumers, and gig-economy workers—provides much-needed support to farmers.
When put together, all of these pieces form the bigger puzzle—the much-needed labor support for farmers, a reduction in food waste, and as an added bonus, data-driven insights about farmers’ crops.
Our team decided to tackle this challenge because we love food! But also because of the huge financial toll on farmers, the unfortunate food waste, the lack of workers for farms, and the recent shutdown of many meat-processing facilities.
We started with researching current issues of the food supply chain as a result of COVID-19 and narrowed our solution scope to supporting farmers. We then brainstormed ideas and refined our solution into FreshHarvest, a platform connecting farmers with gig-economy workers, consumers, and data prediction tools. We moved to diagramming the process flow of our product, and after creating a concrete model, we proceeded UX wireframing the three user interfaces. (Fun Fact: none of us knew how to use the design program ‘Figma’ so we taught ourselves during the competition!) Alongside the design process, we were working to create a predictive model for farmers on crop yield based on NASA EarthData of US Midwest TopSoil Erosion. We built and trained our algorithm using Python to predict corn crop yield based off of corn planted per acre and topsoil loss. The result analysis from our algorithm, which was done by comparing our predicted corn yield vs the actual corn yield for each farm was very similar. For a detailed description of how our algorithm works along with a visualization of the data, reference this document: https://docs.google.com/document/d/1nv1qAON-BZf7tdypTExS-0E2299I3W99A3Sw6GORXVw/edit?usp=sharing.
In the future, we will work to implement more crops’ datum (i.e. not just corn) and more parameters (e.g. weather, soil moisture, etc). By predicting crop yields, farmers can make data-driven decisions such as: 1) Exactly how much of each food item they should list on the platform for customers (to ensure that they can put as much food on the platform as they can that maximizes profit without entirely depleting their inventory); 2) Projecting revenue 3) Whether yield should be increased, etc.
A problem we encountered was finding relevant datasets for our project. While NASA has a plethora of agricultural data sets, we were searching specifically for soil moisture data as we wanted to create an algorithm to predict crop yield based on moisture levels. We were unable to find an open-source set on soil moisture data, however, we came across a data set correlating corn crops planted per area and topsoil loss to yield. By adjusting our criteria, we created a predictive model for farmers to predict corn yield on these factors. On the presentation side of our project, we struggled with creating a diagram that conveyed our project clearly but was not overly wordy. By adjusting multiple diagrams we created - some that were too wordy, some that were too vague - we found the Goldilocks in-between that provides just enough details.
Overall, we are beyond ecstatic with what we achieved! Four people, from four countries, across three time zones. Thank you NASA, ESA for this amazing opportunity!
NASA EarthData Midwest Top Soil Erosion https://daac.ornl.gov/daacdata/global_soil/TopSoil_Erosion_MidWest_US/data/