HowTo: Reporting interpretations annually (25%)

Priority: high
Updating: once per year

This HowTo describes how we report to NYSDEC about patterns among pesticide and metabolite detections in groundwater and lakes.

Change log:

When	Who	Comment
2024 09 17	Sp17	First draft version. Skeletal, particularly about the data interpretation techniques which are in flux still entirely within the Cornell project personnel.

Related HowTos:

1. Objectives of annual interpretive reporting:

Mobilize annual interpretations of data across all sites.
Inform NYSDEC about raw results and cross-site interpretations.
Discover patterns that may apply beyond the groundwater sites where we sample.

2. Timing of Reports:

There were minimalist annual reports for 2021 and 2022, the former about cooperator recruiting plans and the latter containing the full categorical roster and the first analytical data. These were essentially extended versions of the quarterly project reports.
The first substantive report will be in 2024Q3, covering 2022 and 2023 data which pass the critical mass needed for interpretation of patterns among categorical sites. The 2023 sampling at long term sites is becoming usable and data from these sites is folded in with categorical site data.
These reports had an original contract schedule that did not take into account the sampling and analytical calendars. The sampling calendar ends in late fall, before Thanksgiving, with the second round of categorical samples and long termers that took the longest effort to reach. The last 2023 samples were sent to the lab in December 2023. The data became available in May 2024.
The first substantive report is also taking longer while we do first experiments with multichemical interpretation.
We hope that the next report factoring in 2024 analytical data will be possible in spring 2025, if we can receive all analytical data by March or April 2025.
With the end of a five year contract in December 2025, we may need to omit fall sampling at categorical sites in order to have lab analytical results back well before the end of 2025 to produce a last interpretive report.

Note that these reports are all cumulative to the first samples taken, in 2022.

3. Content of annual interpretive reports

The 2023 interpretive report outline is approximately:

1. Introduction
  1.1 Purpose and context
  1.2 Prior work
  1.3 Project objectives, priorities, and coverage
  1.4 Pesticide use in 2022
  1.5 This report

2.  Sites and protocols
  2.1 Design, approach, and recruiting
  2.2 Recruits
  2.3 Sample collection and analytical methods
  2.4 Analyte selection

3.  Basic results
  3.1 Categorical sites
  3.2 Long term sites
  3.3 Lakes

4.  Interpretations and discussion
  4.1 Additional data
  4.2 Cross tabulations
  4.3 TGUS
  4.4 Machine Learning
  4.5 Future refinements to interpretations

Appendices

The sparse lakes sampling limits interpretations to a comparison with groundwater sites, in section 4.2. 4.3 and 4.4 are entirely about groundwater sites, dominated by the categorical sites which have been best characterized.

4. Data interpretation techniques

(to be filled in)

4.1 Cross-tabulation

4.2 Modeling of leaching with TGUS index

4.3 Machine learning

4.4 Redaction of detailed data for release to NYSDEC and the public

5. Building the report

For the 2023 reports we began composing using the Quarto framework as we do with quarterly reports and 52 annual individual confidential reports. Those reports draw directly from the database and create graphics and tables from data retrieved from the database.

We began an annual interpretive report using Markdown and Quarto; just after a first partial version we abandoned this in favor of the more familiar, but slower, indirect crafting in Microsoft Excel of charts and tables, instead of generating them directly from the database via R code, and we are composing narrative interspersed with figures and tables in Microsoft Word.

graph TD;
E((Text<br>editor)) --> Q[.qmd file] --> Quarto((Quarto app<br>and<br>R interpreter)) --> R[Preview HTML<br>in browser]
Quarto --> D[DOCX file export] --> W((MS Word partial))
DB[(Project<br>tabular<br>database)]  --> Quarto

DB --> CSV[CSV joined export]
CSV --> Ex((MS Excel))

W --> W2((MS Word final editing))
Ex --> W2
W2 --> Draft[Final draft version DOCX]
Draft --> DEC((NYSDEC review)) --> W3((last Cornell mods))  --> Final[final]

We extended the project database to include chemical attributes like field degradation half lives, pesticide label application rates, and soil properties of the cooperative sites. Sources and processing are discussed in the Tabular Database HowTo

6. Distributing reports

We customarily email PDFs of quarterly reports to NYSDEC.

The interpretive report can also be posted on the project website.