📍
Welcome to agragent
To get started, go to Satellite Maps and upload a KML/KMZ file or draw a polygon over your field. The dashboard will update automatically with real data from the location.
Satellite Maps
Google Earth Engine · Sentinel-2 · Real ImageryGoogle Earth Engine
Conecta con tu cuenta GEE para visualizar imágenes Sentinel-2 reales
💡 Sentinel-2 SR Harmonized · 10m resolución · Cloud masking automático (QA60) · Solo área del polígono
Índice:
GEE Activo
Cargando imágenes GEE...
🗺️
Suelta el archivo KMZ/KML aquí
NDVI — Índice de Vegetación
0.0 (Sin veg.)0.30.60.80.9 (Alta)
Fórmula: (NIR − RED) / (NIR + RED) · Bandas B8, B4
Capas activas
0 capas
Sube un KMZ/KML o dibuja para agregar capas.
🛰️ Sentinel-2 · OpenStreetMap base · Conecta a GEE para imágenes reales
Climate Data
Sube un polígono para ver datos climáticos
Upload a KML/KMZ or draw a polygon to load climate data for the location.
Annual GDD
—
Growing Degree Days (base 10°C)
Chill Hours
—
Hours 0–7.2°C (Apr–Sep)
Frost Days
—
Days with Tmin ≤ 0°C
Heat Waves
—
≥3 consecutive days >35°C
Dry Months
—
<30mm precipitation
Avg Humidity
—
Relative humidity (%)
Total Solar Radiation
—
MJ/m² (season total)
Total ET₀
—
Evapotranspiration (mm)
Monthly Temperature (Min/Max) — 12-Month Season
Monthly Precipitation (mm)
Monthly Relative Humidity (%)
Monthly Solar Radiation (MJ/m²)
Monthly Evapotranspiration ET₀ (mm)
Growing Degree Days (GDD) Accumulation
Historical Analysis (2015–2026)
Multi-season comparisonAnnual GDD Comparison by Season
Annual Precipitation by Season (mm)
Average Temperature by Season (°C)
Frost Days per Season
Heat Stress Days per Season (>35°C)
Season Comparison Table
Upload a polygon to load historical data.
Genomic Analysis
RNA-seq Differential Expression — Vitis vinifera phenological stages (Altimiras et al.)RNA-seq Processing Pipeline
Raw Reads
68 samples
→
FastQC
✓ QC
→
Trimmomatic
✓ trimmed
→
Salmon
✓ quant
→
tximeta
✓ import
→
edgeR
✓ DE
→
DEGs
✓ 3,603
→
ML Classification
✓ done
Total Reads
1,336M
68 RNA-seq samples
Data Processed
87.0 GB
FASTQ files
Phenological Stages
8
E-L 3 → E-L 41
Unique DEGs
3,603
FDR < 0.05 (edgeR)
Phenological Stage Progress
68%
Season
Progress
Progress
Veraison
Stage 35 — Modified E-L Scale
EL 3
Bud
Bud
EL 15
Inflor.
Inflor.
EL 27
Berry
Berry
EL 35
Veraison
Veraison
EL 38
Ripen
Ripen
EL 41
Senesc.
Senesc.
DEGs per Developmental Stage (vs. E-L 3 baseline)
Sample Composition by Variety
DEG Accumulation Across Stages
Top Differentially Expressed Genes by Stage
| Gene ID | Log2FC | FDR | Stage | Regulation |
|---|---|---|---|---|
| VIT_03s0063g01440 | +7.71 | 1.89e-16 | Shoot & Inflor. | UP |
| VIT_01s0011g00140 | +6.76 | 1.37e-14 | Shoot & Inflor. | UP |
| VIT_19s0014g03940 | +5.62 | 5.15e-15 | Shoot & Inflor. | UP |
| VIT_10s0003g02070 | +4.93 | 1.74e-24 | Shoot & Inflor. | UP |
| VIT_02s0025g00200 | +4.00 | 1.70e-14 | Shoot & Inflor. | UP |
| VIT_12s0142g00360 | +3.99 | 1.68e-21 | Shoot & Inflor. | UP |
Differential Expression Summary — All Comparisons vs. E-L 3 (Baseline)
| Comparison | Major Stage | Tissue | Replicates | Down-regulated | Up-regulated | Total DEGs |
|---|---|---|---|---|---|---|
| EL 3 → 5 | Shoot & Inflor. | Flower bud | 6 | 0 | 0 | 0 |
| EL 3 → 15 | Shoot & Inflor. | Flower bud | 8 | 11 | 84 | 95 |
| EL 3 → 17 | Shoot & Inflor. | Flower bud | 5 | 91 | 80 | 171 |
| EL 3 → 27 | Berry dev. | Flower | 2 | 304 | 214 | 518 |
| EL 3 → 31 | Berry dev. | Grape berry | 6 | 455 | 590 | 1,045 |
| EL 3 → 35 | Ripening | Grape berry | 13 | 1,052 | 596 | 1,648 |
| EL 3 → 36 | Ripening | Grape berry | 4 | 2,259 | 634 | 2,893 |
| EL 3 → 38 | Ripening | Grape berry | 15 | 1,675 | 661 | 2,336 |
| EL 3 → 41 | Senescence | Fruit | 3 | 2,563 | 840 | 3,403 |
Image Analysis
Embrapa WGISD · Wine Grape Instance Segmentation Dataset📤 Analizar imagen propia
Arrastra o haz clic para subir imagen de viñedo
Acepta .jpg · .png · .tiff — Máx 50MB
Yield Prediction
Extra Trees Regressor — Season 2025/2026Predicted Yield
9.8
tonnes per hectare
Confidence Interval: ± 0.8 t/ha (9.0 – 10.6)
Extra Trees Regressor
R² = 0.972
MAPE = 3.8%
Feature Importance
Parcel Predictions
| Parcel | Area | Predicted | ±CI | Trend |
|---|---|---|---|---|
| Parcel A | 4.2 ha | 10.3 t/ha | ±0.7 | ▲ +0.4 |
| Parcel B | 3.8 ha | 8.9 t/ha | ±1.0 | ▼ −0.6 |
| Parcel C | 5.1 ha | 10.1 t/ha | ±0.8 | ▲ +0.2 |
| Parcel D | 2.9 ha | 9.2 t/ha | ±0.9 | — 0.0 |
| Parcel E | 6.3 ha | 10.6 t/ha | ±0.6 | ▲ +0.5 |
Historical Yield: Predicted vs Actual (2020–2025)
Model Comparison
| Model | R² | MAPE (%) | RMSE | Train Time | Best |
|---|---|---|---|---|---|
| Extra Trees | 0.972 | 3.8 | 0.41 | 2.3s | ★ |
| CatBoost | 0.969 | 4.1 | 0.44 | 18.7s | |
| Random Forest | 0.964 | 4.5 | 0.47 | 3.1s | |
| XGBoost | 0.958 | 5.0 | 0.52 | 6.4s | |
| SVR | 0.931 | 6.8 | 0.66 | 0.9s |
References & Credits
Author, publications, and data sourcesAuthor
FA
Associated Publications
Transcriptome Data Analysis Applied to Grapevine Growth Stage Identification
Altimiras, F. et al. — Agronomy, 14(3), 613, 2024
Source of genomic data: RNA-seq pipeline, DEGs, phenological stage comparisons (Tables S1–S5)
A Computational Framework for Crop Yield Estimation and Phenological Monitoring
Altimiras, F. et al. — Progress in Artificial Intelligence, EPIA 2024.
LNCS vol. 15400, Springer, 2025
Datasets & Data Sources
| Resource | Description | Usage in agragent |
|---|---|---|
| WGISD | Wine Grape Instance Segmentation Dataset (Embrapa). 300+ images, 5 varieties, YOLO bounding boxes. Santos, T.T. et al. Computers and Electronics in Agriculture, 2020 |
Image Analysis section |
| Sentinel-2 SR | COPERNICUS/S2_SR_HARMONIZED — 10m multispectral satellite imagery via Google Earth Engine | Satellite Maps (NDVI, NDRE, MSAVI, TCARI) |
| Open-Meteo | Free open-source weather API — historical daily temperature and precipitation | Climate Data section |
| RNA-seq (ArrayExpress) | 68 Vitis vinifera samples across 8 E-L stages. Processed with Salmon + edgeR pipeline | Genomic Analysis section |
License & Source Code
This project is open source under the MIT License.
Source code: github.com/fjaltimiras/agragent