PREP - Oct 8 Meeting with Prof. Ray and Dr. Kay (2:30-3:00 PM): Read Before the Call

Everyone reads this before the Thursday, Oct 8, 2:30-3:00 PM connect with Prof. Ray and Dr. Kay. Dr. Kay asked us to come prepared to share what we understand so far and our next steps. This entry has all three.

• THE 30-SECOND ANSWER: We are building FluxView, a dashboard that shows the farm's flux tower data (air temperature, rainfall, humidity, soil moisture, and carbon exchange) so anyone can see what happened and when, like a weather app with history. Forecasting comes second, using the same data.
• WHAT WE HAVE: 8 flux towers on the PVAMU research farm. Daily and 30-minute records from 2023-05-10 to 2026-09-15 (about 3 years). Each tower has one 30-minute file and one daily file, plus QC flags for data quality (see the Flux Tower Data Dictionary entry).
• WHAT WE UNDERSTAND (plain words): a flux tower does not measure one plot. It measures air exchange over a changing area (the footprint), which depends on wind, tower height, and the surrounding field.
• Negative carbon exchange (NEE) means the land is taking CO2 out of the air (a sink). Positive means it is releasing CO2 (a source).
• The plots had different farm management over three years: ploughing, planting, weeding, harvesting, fertilizer, and biochar. Changes in the data need to be matched to when these happened.
• Plot layout (from Dr. Kay): 6 plots and 8 towers. Towers do not map one-to-one to plots, and some towers feed two plots. See the "Shell reply (Dr. Kay): Plot and Tower Mapping" entry for the full table.
• Sponsor priorities: (1) the visualization dashboard first, with weather, soil moisture, and carbon exchange; (2) forecasting later on the same data, tested honestly by training on earlier data, holding back a later period, predicting forward, then comparing with what actually happened.
• OUR PLAN THIS WEEK: Rory loads the daily data into a database, then the 30-minute data. Raylen checks the loaded data against the original files. Alex sketches the two dashboard pages. Charts will be labeled by plot, with the towers that feed each plot shown. The first working views are due Oct 14.
• QUESTION 1 - Plot mapping: which tower best represents each plot? How should we handle Towers 5 and 6, which each feed two plots? Dr. Kay said the mapping may not be fully accurate, so what needs confirming first?
• QUESTION 2 - Management timeline: can we get the dates of ploughing, planting, weeding, harvesting, fertilizer, and biochar application for each plot? Without them we cannot explain changes in the data.
• QUESTION 3 - Quality flags: what does each flag value mean (for example, what does 1 stand for)? Are missing values filled in, or left blank?
• QUESTION 4 - Time zones: the TIMESTAMP column is in UTC, but the other timestamp columns are local time. Which should charts use?
• QUESTION 5 - Units: please confirm air temperature in degrees C, soil moisture as a fraction, rainfall per interval in mm, and carbon exchange in micromol CO2 per m2 per second.
• QUESTION 6 - Access: who should be able to see the dashboard (public, PVAMU only, or team only)? Is there a data-sharing rule for hosting Shell or PVAMU data on outside services?
• QUESTION 7 - Success: what does a good first demo look like by the end of October or November? Which variables must it include?
• QUESTION 8 - Forecasting: what should we predict first (for example, soil moisture or carbon exchange), how far ahead (for example, 3 months), and how should we split training and test data?
• QUESTION 9 - Access: can we get the eShare folder and the plot and intervention photos Dr. Kay mentioned? Can we see Shell's internal visualization tool (Hemshikha)?
• QUESTION 10 - Field trip: what should we look for on the farm that the data cannot show?
• WHO SPEAKS: Raylen opens with the 30-second answer, then our next steps and questions 1, 2, and 7. Rory covers where the data loading stands, plus questions 3, 4, and 5. Alex covers the two dashboard pages and what a user should see first. Everyone else listens and takes notes.
• IF YOU DO NOT KNOW: say so and write it down. "We will confirm that" is a good answer. Do not guess at plot names, flag meanings, or dates.