phenomenals • phenomenals

PhenoMeNals
Phenology Memory signals

📖 Overview

PhenoMeNals is a phenology-aware predictive framework for grapevine yield and quality forecasting, integrating phenology simulation, eco-physiological functions, and statistical modeling.

The framework explicitly accounts for the two-year reproductive cycle of grapevine, combining a dormancy-forcing phenology model with cumulative “memory signals” derived from environmental cues (temperature, radiation, water status, biotic and abiotic stresses). These signals are aligned along a standardized phenological timeline and weighted by their correlation with yield- and quality-related traits, enabling biologically meaningful predictions and interpretation of carry-over effects.

Note: PhenoMeNals is released as an R package with C# routines invoked from R. It is currently available only on Windows, but cross-platform compatibility is under development.

Highlights

🍇 Predicts grapevine yield and quality using phenology-aware environmental signals
🧠 Integrates phenology simulation, eco-physiological functions, and statistical modeling
📈 Accurate forecasting across multiple sites, varieties, and years
🔍 Analyzes carry-over effects on yield and quality
⚙️ Hybrid architecture: R interface + C# computation
🪟 Windows-only (Mac/Linux support coming soon)

Installation

PhenoMeNals is available from GitHub. You need:

R (≥ 4.0)
devtools for installation

install.packages("devtools")
devtools::install_github("GeoModelLab/phenomenals")

This installs the R interface and the compiled C# executable backend (Windows only).

Getting Started

You can start with the included sample datasets:

library(phenomenals)

head(colliOrientali)
head(bbchSample)

Calibrate phenology with:

res <- phenologyCalibration(
  weather_data = colliOrientali,
  referenceBBCH = bbchSample,
  phenomenalsParameters = phenomenalsParameters,
  sites = "ColliOrientali",
  varieties = "Carmenere",
  iterations = 50,
  start_year = 2004,
  end_year = 2015,
  timestep = "daily"
)

Run the full pipeline with:

res <- runPhenomenals(
  weather_data = colliOrientali,
  target_data = targetSample,
  phenomenalsParameters = phenomenalsParameters,
  sites = "ColliOrientali",
  varieties = "Carmenere",
  timestep = "daily",
  target_traits = c("yield"),
  rolling_window = 5,
  evaluation_range = list(c(0, 200)),
  multicollinearity_threshold = 0.8,
  max_phenomenals = 4,
  bin_size = 1
)

Documentation

Use ?phenologyCalibration and ?runPhenomenals to access documentation inside R.

To explore more, visit the pkgdown website:

📚 https://geomodellab.github.io/phenomenals

License

Creative Commons Attribution-NonCommercial 3.0 Unported (CC BY-NC 3.0)