winter 2022 - 2023

Project Overview
Agriculture
GHG's
Remote-Sensing
I joined Regrow as the lead designer for the Sustainability Insights (SI) product. The primary purpose of SI is to use remotely-sensed data to allow clients to plan and assess how the adoption of more sustainable agricultural practices would affect GHG emissions and other important outcomes. The core of my work on SI was to find ways to create a narrative about the landscape, crops, and emissions and environmental outcomes. 

A project that embodies this work is the Geopatial Explorer tool. The purpose of this tool was to provide clients with a highly visual, map-based way to understand the effects of adopting (or failing to adopt) regenerative agriculture practices on their croplands. These effects could be widespread across regions, crop types, and even time periods. My primary responsibilities for this project included the full end-to-end design process, including qualitative research, conducting internal interviews, visual and UX design work, and documenting the full process in a product spec to be shared with engineering collaborators.
BACKGROUND CONTEXT

old tool, new questions

There had previously been a map-based tool in SI called Geospatial Analysis, which had been deprecated prior to me joining the team. However, clients clearly craved and made requests to our customer success team for a more visual way to explore their data. And, as time went on, the quantity and types of data Regrow had available continued to grow. With Regrow’s clients being some of the largest food and agriculture businesses in the world (ex. Cargill, General Mills), users were interested in analyzing and making management decisions for potentially millions of acres of crops at a time.

A new mapping tool would need to be able to show insights for 12 possible key performance indications (KPI’s) that each fell into one of two categories: agricultural practices and outcomes. Clients needed to be able to draw insights from combinations of these KPI’s in order to understand the effects of transitioning to more sustainable land management.

Also crucial to this process would be designing a visualization that could expand and accommodate new data sets as Regrow expanded outside of the world of carbon into other realms of sustainability, such as water and yield.

EARLY IDEATION

key questions:
Creating a user journey

After conducting an audit of the deprecated tool and interviewing Regrowers about their experiences with the product, it was clear that the tool needed to parse different geographic scales. For example, viewing data between states and also for counties or watersheds.

To get a sense for other ways that clients might want to manipulate data, I interviewed 8 Regrowers across the Product, GTM, Data Science, and CS teams. It became clear that users were interested in:

  1. Evaluating success of existing regenerative ag. programs: Success can be measured in terms of meeting absolute targets or, if there has been improvement relative to neighboring areas.
  2. Planning out areas to execute new programs: Identify areas that stand out from the pack; which areas have the largest potential for improvements?
  3. Identifying landscape-level trends: Which areas are hot spots for different environmental risks?
User journey, goals, & relevant data types
User journey and goals
PROJECT DEFINITION

CREATING A SCOPE 
OF DESIGN

A map is an immensely complex tool; entire companies exist for the exclusive purpose of mapping! For the sake of keeping a neat scope for this project, I worked closely with product to create boundaries and aim to achieving the following goals:

  • Allow clients to compare a maximum of up to two KPI’s: one practice and one outcome. This reduces potential visual noise, and keeps exploration focused.
  • Allow for a few customizations in the form of layers that can be toggled on or off.
  • Incorporate at least one publicly available data set, to test how the tool might integrate with other areas of climate science.
  •  Make clear connections between the map’s visualization of data, and other ways to manipulate data in the rest of the SI platform
EARLY DESIGNS

CORE layout & VernaculaR filtering

Using the above matrix of each agricultural practice and outcome metric combination, I tried to imagine how a user might want to filter that information by scale, region, and crop type. The result seemed overwhelming at first, but I was able to establish a set of rules to govern how users would move through the data.

After exploring a few options for allowing users to quickly select KPI’s, I decided to adopt what I called a
‍“vernacular filter,” where all of the different filtering dimensions were written in the format of a sentence. This way, users would immediately have to frame their navigation in the form of a question and make more intentional search choices, rather than being forced into a free-for-all scroll through the map.

The next layout question was how to allow users to access additional details, namely more detailed metrics for whatever KPI they were analyzing. After trialing a few designs, I ultimately decided to utilize a floating side panel that could contain nested views.
Early designs exploring navigation and layout. Designs arranged oldest to newest, starting from the far left.
FINAL VISUALS

COLOR, Accessibility, simplification

For the design of the map portion, I referenced other mapping tools created by government agencies or that used publicly-available climate and agriculture data sets for inspiration. Initially, I explored using a combination of choropleth coloring and symbols to represent different KPI’sI. For example: cover crop adoption could be represented by coloring on gradient scale, and water risk rates could be overlaid on top of those counties as proportionally-sized circles in those respective colors. However, upon a second round of internal user feedback, it seemed this was still quite visually intense.

Exploration for using choropleth coloring and symbol sizing to show the relationship between KPI's

To simplify this, I adopted a bivariate choropleth color scheme. In this model, agricultural practice KPI's are represented by one color gradient (yellow), and outcome KPI's are represented by another (blue). The two are combined the create a third color (green) where there is overlap. With this scale, I needed to prioritize accessibility. The color scale would need to have enough visually-distinct colors that someone could pick apart different regions easily, with enough contrast, and with clear boundaries. The coloring also could not conflict with any additional layers. Through a series of explorations, I landed on the final designs shown here:

Top: color scheme for having one KPI selected, for practices and outcomes. Bottom: coloring for combination of 2 KPI's

Another key change for this final iteration of the visual design was to simplify the side panel. I reduced some of the color blocking in favor of more subtle uses of color, and pared down the statistics being shown to the most crucial pieces of information. In order to add context on whether results were "good" or "bad," I added small tags to each summary statistic, indicating whether each value was relatively low or high.

Zoomed-in view of an individual county
FINAL VISUALS

PROGRESSIVE reveal

With the more aggregate-level view simplified to only the most crucial pieces of information, users still needed a way to view individual areas (in this example, counties) in greater detail. If the user clicks on an area in the side panel or on the map, or, if they manually zoom in on the map, they are able to see a more detailed view.

In this view, the user can see landscape features, such as city name and waterways, at a more granular level. In the side panel, they are able to see the specific number of farms, total land area, and crop types for that geography. They are also able to their selected KPI metrics broken down by those additional dimensions, as well as volume for each crop on that piece of land.

VISUAL DETAILS

Refinement & Layers

The final element to this visualization was allowing users to customize their map view by toggling different layers.

In my research into other tools, I found some common sets of features, including labels, roads, and boundary lines. These all were standard features out-of-the-box. The complexity began when we started to consider how to incorporate new data sets, such as information about water scarcity. Questions about water usage were top of mind for our clients, and for the agritech industry as a whole, but due to technical limitations Regrow couldn’t yet offer remotely-sensed insights.

So, the best way for us to incorporate this information was using publicly available data. To do this, we referenced the Aqueduct Water Risk Atlas, focusing on water quality indicators. I worked with our geospatial engineer to create a simple visual indicator in the waterways that would utilize this data and allow users to see contamination levels.

All of the map layers included in the final version of design
Sample of visual indicators for contaminated waterways
CONCLUSION

FINAL THOUGhts, Next Steps

With these designs, I conducted a second round of internal interviews to validate the designs further. Overall, the qualitative results were resoundingly positive, with customer success managers eager to share the designs with their clients. Having entered into the project with a very specific scope in mind, we as a product team began to brainstorm the next features to explore, including:

•  Allowing users to add custom points, shapes, and boundaries
•  Adding the ability to take and share snapshots of the map
•  Adding more public data sets for things like livestock and yield resilience

However, at this point in time I left Regrow as the first version of the map was at just the beginning of its development and unfortunately did not get to oversee the final implementation. Over the course of this project I was certainly challenged; I was new to designing for map-making, which is both an art and a science. In particular, I enjoyed the technicality of this design and the finesse of having to create simple and reusable patterns out of the chaos of data and geoscience!

Updated JUne 2024
LinkedinEmail