OPheno/Weed-phenology

OPheno: Curated Weed Emergence & Phenology Dataset

Overview

The OPheno Curated Weed Emergence & Phenology Dataset provides harmonized, high-quality weed emergence and phenology observations collected from research studies, field trials, and scientific publications.

It contains standardized temporal observations of weeds growing within various crop systems, enabling:

• modelling of weed emergence timing
• development of weed phenology models
• cross‑site and cross‑species comparison
• machine learning applications
• agronomic research and decision-support development

This curated dataset is continuously expanded as new submissions are reviewed and validated.

---

Load the Dataset

If you want to download directly the .csv file of the data set click here

With Python

from datasets import load_dataset

dataset = load_dataset("OPheno/Weed-phenology")
df = dataset["train"].to_pandas()

With R

library(reticulate)

datasets <- import("datasets")
dataset <- datasets$load_dataset("OPheno/Weed-phenology")
df <- py_to_r(dataset$train$to_pandas())

For more information about setting up your environment click here

Data Dictionary

Column	Type	Description
trialNumber	string	Unique identifier for each experimental trial.
countryCode	string	ISO 3166-1 alpha-2 country code of the observation site.
crop	string	Crop species identified using EPPO code.
cropEstablishmentMethod	string	Method of crop establishment (e.g., direct seeding, transplanting).
tillageSystem	string	Tillage practice used at the site (e.g., no-till, conventional tillage).
typeOfTrial	string	Type of experimental setup (e.g., field, greenhouse, lab).
seedingDate	string (ISO 8601)	Date the crop was seeded (YYYY-MM-DD).
targetCode	string	Target weed species identified using EPPO code.
date	string (ISO 8601)	Date of the observation (YYYY-MM-DD).
assessment	string	Type of observation or assessment performed (e.g., cumulative emergence, weed growth stage).
measurementUnit	string	Unit associated with the observation (e.g., %, BBCH).
value	float32	Numerical value recorded for the observation.
latitude	float32	Latitude of the site in decimal degrees (WGS84).
longitude	float32	Longitude of the site in decimal degrees (WGS84).
photoPeriod	float32	Photoperiod (day length in hours) at the observation date.
dayOfYear	int32	Day of year for the observation (1–366).
daysAfterStart	int32	Days since the start of the trial or seeding date.
cumulativeEmergencePercentage	float32	Cumulative emergence expressed as a percentage over time.
contributor	string	Name(s) of the dataset contributor.
organization	string	Affiliated institution(s) responsible for the data.
doi	string	Digital Object Identifier linking to the original publication or dataset.
publicationStatus	string	Publication status of the data (e.g., published, unpublished, in review).

Data Creation

The curated dataset is built from researcher contributions and from published studies.
To harmonize diverse submissions while keeping the process open and flexible, OPheno applies a lightweight curation workflow:

1. Ingest raw submissions from the OPheno Submission Portal and literature sources.
2. Standardize names, formats, and units where appropriate (e.g., ISO‑formatted dates, consistent column naming).
3. Map observations to the curated schema, including:
   • country_code, latitude, longitude, source
   • crop, crop_establishment_method, tillage_system, seeding_date
   • weed, date, type_of_value,value, measurement_unit
   • photo_period, day_of_year, days_after_start
   • optional cum_emergence and cum_emergence_percentage when derivable.
4. Perform basic validation, such as:
   • checking that dates parse correctly
   • verifying coordinate plausibility
   • catching obvious type inconsistencies
5. Enrich the dataset with environmental context variables where possible, such as photoperiod, day‑of‑year, and days‑after‑start.

If a submission is not yet ready for curation (e.g., incomplete metadata), it remains accessible in the submission portal for transparency and future revision.

Although the curated dataset undergoes quality checks, users should still apply any domain‑specific validation required for their analysis.

Contributions, Data Scope and Requirements

We welcome contributions to OPheno! If you have relevant weed emergence or phenology data to share, please go to our submission portal. Contributions help enhance the dataset's scope and utility for the entire community. To ensure the utility of OPheno for modeling baseline weed dynamics, the current version focuses on specific data types.

Inclusion Criteria (Examples):

• Assessments from untreated control plots.
• Data under standard regional agronomic practices.
• Pre-treatment measurements or data before confounding events occur.

Exclusion Criteria (Current Curated Version):

• Weed-control treatments that directly modify weed development or emergence patterns.

Please note that these criteria apply only to the current curated release.
We still welcome submissions containing such data because:

1. they remain scientifically valuable,
2. they help us understand the broader landscape of available phenology and emergence datasets,
3. they may be incorporated in future expanded versions of OPheno.

In short, even if your data fall outside the scope of the curated dataset, we encourage you to submit them, as they contribute to community knowledge and help guide future development of the project.

Metadata is Crucial: Please provide detailed metadata regarding management practices and environmental conditions, especially any treatments applied (include this information in the submission portal.

Future Scope: We aim to potentially expand OPheno in the future to include data explicitly studying factors like weed control effects. If you have such data, please provide it in the submission portal.

Allowed Values and Units

PERC_PLANTEMERGED_PLOT
Represents cumulative emergence already provided as a percentage.
For each date:
emergence(t) = average_value(t)
→ Shows the percentage of plants that have emerged by each date, as directly measured.

NUM_PLANT_M2
Represents relative emergence based on the final plant density.
For each date:
emergence(t) = (average_value(t) / final_average_value) × 100
→ Shows how much of the final density has emerged by each date.

NUM_PLANTEMERGED_M2
Represents cumulative emergence over time (new plants per date).
For each date:
emergence(t) = (sum of average_values up to date t / sum of all average_values) × 100
→ Shows cumulative emergence progress as a percentage.

Calculation logic

Time-series definition

• Each calculation is performed per time series
• All dates belonging to the same time series are processed together.

Data validation

Before any calculation:

• All observations of a given assessment_type must use the same unit.
• If mixed units are detected, an exception is raised.

Date aggregation

• If multiple observations exist for the same date (e.g. replicates or subsamples):
  • Values are averaged first. After averaging: Only one value per date is retained.

Citation Information

@misc{opheno_dataset_2026 title = {OPheno: An Open-Source Weed Phenology Database}, author = {{[Espejel Padilla, Jorge Alberto; Akhter Muhammad Javaid; Kromminga, Heiko Hopke and Hoffman, Holger]}}, year = {2026}, publisher = {Hugging Face}, journal = {Hugging Face Hub}, howpublished = {\url{https://huggingface.co/OPheno}} }

License

The data is distributed under the Creative Commons Attribution 4.0 International (CC BY 4.0) license. You are free to share and adapt the data for any purpose, provided you give appropriate credit, provide a link to the license, and indicate if changes were made. See the LICENSE file for full details

Maintainer

OPheno is maintained by Heiko Hopke Kromminga , Muhammad Javaid Akhter , Jorge Alberto Espejel Padilla, and Holger Hoffman. Please use the Hugging Face Hub's Community tab for questions or open an issue on the repository [if using GitHub mirror etc.]

Acknowledgments

The maintainers acknowledge the use of large language models for assistance in drafting documentation sections, generating code examples, and refining text presented in this repository. All final content, was reviewed, validated, and finalized by the maintainers