Field scouting: a 14-day recording and review plan

Overview

# Field scouting: a 14-day recording and review plan

## Goal and required inputs

Produce a repeatable observation log and a reviewed next-visit plan. Before scheduling, identify your real field, crop and recorded growth stage; agree safe access, an observer, revisitable stops and a local counting method. Have a way to record time zones, count denominators and uncertain observations. This is a work-plan template, not an observed farm result.

## Procedure

1. Use Adapt to my program to create a private farm program without a public Hub profile; edit steps, days and approved inputs, then preview selected fields/start date before creating tasks. Day offsets are 0, 1, 3, 7 and 14; adapt dates for access and local practice. Created tasks are planned work, not completed observations.
2. On day 0, assign stop IDs and describe how another observer can revisit them privately. Record route changes rather than silently moving a sampling point.
3. On day 1, record directly visible symptoms and the number of plants examined at each stop. Keep yellowing and visible damage separate; a symptom is not a diagnosis.
4. On day 3, check missing times, units, denominators and ambiguous descriptions. Leave unknown values empty and route uncertain records to a named reviewer.
5. On day 7, repeat the same method. On day 14, review comparability and approve the next observation tasks.

## Deliverables and acceptance criteria

Keep a private `field-log.csv` with `observation_id,field_id,stop_id,timestamp_with_timezone,crop,growth_stage,plants_examined,leaf_yellowing,visible_leaf_damage,observation_quality,method_note,observer_alias`. Keep `review-log.csv` with `observation_id,issue,reviewer_alias,status,next_task`.
Accept the cycle only when record IDs are unique, every observation has a field/stop/time, counts have a denominator, and every unresolved issue has an owner or an explicit pending status. Missing data must never become zero or an invented observation. Compare symptom counts only where the method and denominator are comparable.

## Worked example and next action

Download [synthetic scouting labels](/hub/agridigi/synthetic-scouting-labels). The fictional `scout-a-01` record has yellowing, visible damage and uncertain quality, so its disclosed toy rule sets `review_required=true`. Use it to test a review queue; do not import it as a real field visit. Join `field_id` to [sandbox boundaries](/hub/agridigi/sandbox-field-boundaries) and inspect the dataset's executable checks. For your real cycle, create a follow-up task only after a person reviews the observation and records the reason.

## Privacy, provenance and license

Strip names, exact stop coordinates and photo GPS/EXIF from public examples; obtain permission for photographs and farm records. Store an observer alias rather than a public employee name. AgriDigi authored this template. [UC IPM](https://ipm.ucanr.edu/what-is-ipm/) is a reference on monitoring and identification, not a source of copied data. Original text and tasks are [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/): credit AgriDigi, link the original resource and license, and state changes. External references retain their terms. No yield claim, diagnosis, treatment rate or automated intervention follows from this plan.

Action plan

Crop: Crop independent · Duration: 14 days

1. Define safe revisitable stops and the recording method

   Day 0

   Confirm permission and safe access to your own real field. Assign field/stop IDs, crop, recorded stage and observer alias; write the local counting method and route. Deliver a private route note and blank field-log.csv. Accept only if another authorized observer can revisit the stops and all required columns are defined; sandbox polygons are not real fields.

2. Record the first visit with count denominators

   Day 1

   At each stop record the actual time with zone, plants examined, directly visible yellowing/damage and observation quality. Keep causes and diagnosis separate from symptoms. Deliver completed baseline rows with unique observation IDs. Accept only if every count has a denominator and missing values remain missing; remove photo GPS and personal details before sharing.

3. Resolve missing fields and assign uncertain records

   Day 3

   Check field/stop IDs, time zones, units, denominators, duplicate IDs and ambiguous labels. Mark unresolved values unknown rather than zero. Deliver review-log.csv with each issue, reviewer and pending/resolved status. Accept when every unresolved item has an owner; use synthetic labels only to test the schema and review queue.

4. Repeat the agreed stops and document method changes

   Day 7

   Revisit the baseline stops using the same counting method. Record actual weather/access deviations and any change of stop or denominator. Deliver a second dated observation set and a comparability note. Accept when records can be matched by field/stop and changed methods are explicit; two visits do not establish disease cause, yield or treatment need.

5. Close the recording cycle and approve the next tasks

   Day 14

   Review completeness, duplicates, comparable symptom counts and unresolved issues. Request local expert review for new or unexplained symptoms. Deliver an approved next-visit list with responsible people and reasons, or an explicit pending decision. Accept only actual completed records as evidence; scheduling does not prove completion and this plan creates no chemical application.