Background
Precision agriculture, also known as smart farming or precision farming, is an approach that utilizes advanced technologies, including artificial intelligence (AI), to optimize agricultural practices and improve overall efficiency, yield, and sustainability. These models play a crucial role in precision agriculture by analyzing data, making predictions, and providing valuable insights for decision-making.
Inspiration
TerraMetrix is inspired by agricultural producers and the increasing need to adapt to changing soil conditions and climatic patterns. TerraMetrix allows its users to achieve an informed management scenario with a personalized visualization according to their area of interest (AOI). Often, stakeholders have crop failure simply because of inadequate soil conditions which could have been changed early on. For example, tomatoes require a high nitrogen content in the soil, and if proper testing is done before planting, then fertilizer could be added to meet growing requirements. Rapidly changing climate patterns, variable rainfall, and invasive pests create conditions where stakeholders need a tool that is able to quickly give them baseline information on the viability of their crops.
What it does
Users select a crop that they want to grow and input data on soil conditions including type of soil, pH, nutrient content (nitrogen), soil moisture, and temperature. TerraMetrix then determines based on optimal ranges of soil conditions and precipitation in the specified location. The user is able to upload a shapefile which interacts with our database and returns a visualization of the monthly rainfall across their AOI.
How we built it
We sourced data from the United States Department of Agriculture National Agricultural Statistics Service data set. This is a land cover dataset with various crops represented across the United States. We use this as a general feature the user might be interested in. We sourced data from peer-reviewed publications to determine the range of values for growing conditions for our assessment feature. For our personalized visualizations, we used Prism climate data and through an API and created monthly rainfall rasters for the user’s AOI.
Challenges we ran into
One challenge we ran into was deciding a simple yet effective tech stack we could use in order to integrate our R scripts which collected real time monthly precipitation in the US which stored them in a SQLite database. Sending our data from node.js to R was difficult because of the different ways they would accept/pass data. Surprisingly, it was hard to find a large crop dataset for North America with information about crops and their growing requirements. The USDA NASS dataset was only a land-cover type dataset.
Accomplishments that we're proud of
Despite having different knowledge in languages and development, we were able to collaborate and join together our unique experiences to create different features of the program. Therefore, working together as a team with interdisciplinary backgrounds is something that we’re proud of.
What we learned
We learned how to integrate R scripts with node.js and developed a way to pass data between both. Accessing broad data for which to make inferences from was incredibly challenging and time consuming for a 36 hour hackathon; we had to make pivots to ensure we were able to present a finished product.
What's next for TerraMetrix
As more quality data is collected on soil conditions and additional inputs (such as rainfall and climate), TerraMetrix can evolve to recommend a crop to grow based on existing soil conditions, so the users would not have to amend or change their soil. Additionally, as more users input data into the program, we could track patterns in common mistakes when aiming to grow specific crops. For example, if multiple people have a Nitrogen deficit in their soils when trying to grow corn, we could notify the user of this information so they do not have to measure and input their soil conditions. We plan to integrate a machine learning and computer vision feature to train to detect different crops from aerial imagery. We need ground-truthed data to accomplish this.
Log in or sign up for Devpost to join the conversation.