HackOregon is a non-profit focused on making public data more accessible, or presented in a novel fashion. There is a wealth of data around agriculture available to the public, but it is all siloed in individual government agencies at both the federal and state levels, in inconsistent formats, and potentially difficult to interpret at a glance.
What it does
Crop Compass was designed to provide both a uniform API for accessing crop-related data from disparate sources, and an interactive application that presents this data in an easy to use manner.
How we built it
Crop Compass was an iterative project - we began with a much wider wealth of data than is encompassed in the USDA data hackathon challenge, and endlessly challenged ourselves to ask questions we thought were important. If this is a question about what _technology_ we used to build it, well.. The backend is a Python/Flask app with a Postgres database being served by nGINX/gunicorn, with an Angular-based frontend.
Challenges we ran into
The primary challenge with Crop Compass is the endless amount of data available on any number of agriculture and environmental topics, as well as understanding those datasets and grasping how to relate them to each other. We began with a lengthy list of interesting questions: * What crops are growing at this time of year? * What crops do most farmers in my area grow? * What could I (as a farmer) grow instead that would be more profitable? * Where do our crops go? * How do we keep more of what Oregon grows in state, and reduce our reliance on imports? * How do subsidy dollars affect the crops farmers choose to grow? * How will climate change affect what crops can be grown in Oregon? * Can a farmer differentiate their crops enough to outperform industrial monocrop farming? It turns out that answering a question that seems straightforward -- like, "What can I grow here?" -- is far more complex. What is the climate like? How about the soil? Are there water rights available? How much can I irrigate with groundwater access on my property? At the beginning of the project, we attempted to break down the problem into one of natural resources - water, air, soil, and pollution - but the data is highly specialized and requires intimate domain knowledge to grasp the implications. As an example, we found detailed water rights data that annotated every individual point of access to natural bodies of water and wellhead.. but how much land can those water rights irrigate? How much water does a specific crop use in that particular microclimate? How does precipitation affect the amount of water needed? Eventually we concluded that our initial questions were well beyond what we could explore given the time and expertise our team had available to us. We pared back our scope to a smaller set of interesting facts and focused on these instead: * What crops grow in a particular county? * How much land is dedicated to those crops? * How diverse is the crop selection in a particular county? * How much rain does this county get on average? * How many subsidy dollars does this county receive, and for which crops? You can see that these datasets were inspired by the questions we started with, and that given the time and expertise, one could still explore some deeper concepts.
We'd like to focus on some of the scope we had to cut towards the end of the project: * Import/Export Data * Yield data for crops * Dollar value data for crops * Consistency for commodities between datasets * Location of Organic operations and proportions of production (Organic vs Conventional) * Extrapolation of water required to produce a crop * Pesticide / Fertilizer recommendations for crops We'd also like to find or process data that could be used to provide numbers for: * Farmable land and grading * Production facilities for steps in food production beyond farming * Purchaser data (who is buying these crops / where are they going?) * More detailed import/export data (county, city, regional levels) * Purchaser data * Grouping by classes of regions - Wheat doesn't stop growing at the border of Oregon. What regions can we group by that might provide more insight than political boundaries like state & county lines? * Similarly, what insights can we gain by grouping classes of crops? How has the production of all winter crops changed over time?