Exploring the communities which grow food for us here in the United States.
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.



Exploring the communities which grow food for us here in the United States.

@conroywhitney's entry to the USDA Innovation Challenge 2015

View website → http://whogrowsfoodfor.us


The aim of this project is to view the United States through the lens of the regions that produce our food. This starts by looking at where food is grown here in the United States.

To that end, this project has two key pieces of functionality:

  • To allow users to view which regions around the country are involved in the production of specific agricultural products (both crop- and animal-based), and to what degrees.
  • To allow users to view metadata about the distribution of agricultural production as a nation and within a given region.

The goal is to allow farmers, researches, and consumers to gain a better understanding of our food system through the lens of a particular agricultural product.


The data for this project is derived from the National Agricultural Statistics Service (NASS) QuickStats API as accessed through the Microsoft Farm Data Dashboard.

Data Disclaimer

The data used in this application is a filtered version of what is available from the NASS QuickStats API. Specifically, products, query options, and result values are omitted from the final dataset for reasons of brevity (not all products need to be shown), simplicity (conflating multiple variables such as areas planted vs. harvested), and practicality (removing (D) values since they are too low to be considered statistically valid).

For this reason, if you look closely, you will realize that not all of the numbers add up. Specifically, the state numbers will not add up to the national numbers, and the county numbers will not add up to the state numbers. This is because the NASS QuickStats API omits values on the county level which it includes in the state level, and omits some state values that are counted in the national number. When in doubt, click the View in NASS QuickStats API link in the application to view the raw query which produced the data shown.

Data Processing

The data used for this project can be downloaded in its original format. Additionally, you can view the code used to transform from the public format into the datasets in the data folder which is used by the visualizations. Here are some steps to do that:

  1. npm run clean: Start by cleaning all of the data that was packaged with this git repository.
  2. Make any changes you want to the filters which will later remove certain products or options. These regular expressions can be found in the productFilter and filterOption methods of src/product_helper.js. Note the corresponding tests in test/product_helper_spec.js which will help explain how those functions are expected to behave.
  3. gulp product-list: Download a list of all product names from the NASS API. Saves into data/raw/product-list.json.
  4. gulp product-metadata: Download information about possible query string options for each product in the list (class_desc, statisticcat_desc, unit_desc, etc.). Saves into data/raw/{product}/{option}.json.
  5. gulp product-concat: Combine all the separate raw option files into a single file which lists all possible option combinations. Saves into data/raw/{product}/options.json.
  6. gulp product-combinations: Filter and create the cartesian product of all option combinations. These are then combined to create a querystring property which will be used to send a request to the NASS QuickStats API per the contest rules. Creates a unique name property which is a human-readable summary of what the query represents (e.g., lettuce_(romaine)_acres_area_harvested_county). Note: This step removes options which are not directly related to this application to reduce the number of query combinations. Saves output to data/raw/{product}/queries.json.
  7. gulp product-jsonlint: At this point, it's a good idea to sanity-check that all of the files we've been saving are indeed valid JSON. This does not save or delete any files, just outputs any parser errors it encounters. Ideally, there will be no output or effect. If you see any errors, run npm run clean and start over.
  8. product-download: Run each of the queries from step #6 by hitting the NASS QuickStats API with the querystring. Save the raw JSON data directly to output file data/raw/{product}/{query_name}.json. Note: This method may fail based on timeouts from the API. For this reason, you may need to loop between Step #8 and Step #9 a few times.
  9. gulp product-jsonlint-clean: Like Step #7, we run all existing files through jsonlint to find any formatting errors. However, this time, we delete offending files so we can try re-downloading them again. After running this step, return to Step #8 until you don't see any output or errors in either Step #8 or Step #9. Note: this may take a handful of tries before it all downloads, but it will eventually resolve and download everything.
  10. product-clean: Now that we have the API cached locally, filter and combine the raw query responses from Step #8 into smaller files which represent the information we want to work with. Specifically, we need to know the product name, class, and data values for National (FIPS 00000), State (FIPS XX000), and County (FIPS XXYYY) levels. Look in the getCleanJSON method of src/product_helper.js to see the transformation from raw API data to smaller JSON files. Check out corresponding tests in test/product_helper_spec.js to see how the function is expected to behave. Saves to data/products/{product}.json.
  11. product-combine-all: Combine all product JSON files into one giant file to be included with the application. Note: Some applications may just want a handful of product JSON files, or may choose to load them individually to save time or bandwidth. Saves to data/products.json.
  12. product-jsonlint: One last time, sanity-check that all the files we have downloaded and transformed are indeed valid JSON objects. This ensures that we can use any or all of these files in our application without worring about causing any runtime parsing errors.

Thank You!

This project would not be possible without the contributions, feedback, and encouragement of the following people:

  • The open-source community, and @mbostock in particular, for d3.js and a plethroa of examples and libraries which make projects like this possible. You da real MVP!
  • Stuart Allan of Allan Cartography for his explanations and feedback on how to make a map communicate what you are trying to say in the simplest and most effective manner.
  • The Dunbar Farms crew for their inspiration, encouragement, and knowledge base about investigating and participating in the food production system here in the United States.
  • The Zeal team for the co-working space, technical feedback, and encouragement.
  • Last but not least, Tanya Green, for her patience and feedback as I discussed this project ad nauseam for weeks.