Budburst: Evaluating climate change by analyzing community science data

[easherma]

About the group

Budburst brings together researchers, horticulturists, and community scientists on a shared journey to uncover the stories of plants affected by human impacts on the environment. Budburst tells these stories through data collection, data sharing, education, and personal connections.

The data made by observations is publicly available. In this group we'll be exploring some ways to visualize and clean the data for analysis!

We're interested in exploring a few tasks:

• Making an interactive web map of existing observations over time
• Helping evaluate or answer the question of if an observation is 'early' or 'late'
• Helping automate or streamline manual processes Budburst staff need to do in order to prep data for analysis
• Perform other exploratory data analysis using the Budburst team and their domain knowledge for guidance

Group leaders

Eric Sherman
Dr. Emma Oschrin

Who we're looking for

• Data Scientists (ETL, cleaning, but also analysis)
• GIS folks
• Web developers ( for sharing results publicly)
• People interested in submitting observations

Tools

• Python
• QGIS (if needed)
• anything else!

I've been using QGIS and https://kepler.gl/ for some initial mapping explorations

Relevant Links

https://budburst.org/data

Where we meet

At least for now we're gonna hang with the climate change folks!

[easherma]

Data quality issues, solutions:

• flagging entries when folks manually enter species

• app metadata like time spent on app

• filter by time, but in yearly or seasonal increments

• top 50 most observed species

[easherma]

Next tasks with the Map:

• data to group by
• phenophases during the year

layers for plant part (fruits, leaves, flowers)

• within a layer, there is 7 categories
• also within a given year, time series within a year

[easherma-truth]

All the different plant parts are each in their own layer (except for LU and LV which would be combined, ideally )
LU = leaves unfolding, LV leaves changing in fall, same layer gives an annual picture of what leaves are doing

all of those layers could be filtered by the same time period, say by year.

[easherma-truth]

One outstanding task would be to collect other relevant geospatial layers, such as ecological grow zones

[beautifugift]

Good day @easherma-truth
I will like to contribute to this project.
Let me know how I can help
Thanks

[easherma-truth]

Some additional notes from Dr. Oschrin that I haven’t translated into any discrete technical steps yet:
Look at the data from 2007 to present and see what the inter annual variation of spring leaf and/or flower phenophases are by species (eg sugar maple) or genus (eg acer species).

Phenophases to use: First Leaf or all leaves unfolded would work for leaves; first flower or middle flower would work for flowers.
Taking into account regional variability would be ideal (eg what pattern do we see in the Great Lakes region vs the southwest).

A completed version of this might look like “acer species in the Great Lakes region reached the middle flowering phenophases in mid April in 2009, 2010, 2011 but in 2012-2016 we see average middle flowering time shift to late April or early May” for example, not based on real data.

[easherma]

An update on the map side of things: kepler.gl html export works with a minor tweak mentioned here: keplergl/kepler.gl#2202 (comment)

So a todo for me to is to package up the map we already explored in this fashion to share with the Budburst team.

We also created https://github.com/chihacknight/budburst-data-analysis/tree/main and @varunbhoopalam will be doing some initial exploratory data work there

[easherma-truth]

https://github.com/chihacknight/budburst-data-analysis/blob/main/budburst_filtered_type.zip download to view a map of observations data!

[easherma]

Pausing this group for now.