Layers

When enabled, the globe is split left/right for comparison. Each side can be configured independently.
2024-01-01
XYZ global tiles. Step “0h” = latest.
If TIME is empty, the app will use the latest/last time from capabilities with fallback.

Flower Photos (iNaturalist)

to
—

Vegetation Index (NDVI/EVI) · Nearby Species · Invasives

Queries NDVI/EVI at the selected point (latest available date).
VariableValueNote
NDVI—[−1, 1]
EVI—[−1, ~1]
Date—Composite date

Counts are iNaturalist monthly observations near the selected point for the chosen year.

Aridity index: —
Superbloom prediction: — (rainfall-based)

Default taxa: Lepidoptera (47157), Apoidea (630955). Edit if needed.

Ranks species by 5-year observation counts near the point, flags potential invasives (heuristic or curated list if present).
Rank Common name Scientific 5y Count Flag

Phenology & Prediction

Current location —
Year (POWER daily)
Flower Base Temp (°C) GDD Threshold Predicted Date Accumulated Days
Custom — —
Cherry Blossom — —
Tulip — —
Winter Jasmine — —
Plum Blossom — —
Peach Blossom — —
Rapeseed — —
Wisteria — —
Rose — —

Weather (NASA POWER)

Temp: — °C, Precip: — mm
POWER data lags by 1–2 days; try querying yesterday or the day before.

AI Assistant

Plant ID (Remote Image Recognition)

Upload a flower/plant photo and get species suggestions using an online identification API (Pl@ntNet by default). For conservation projects, use this as a first pass and verify with experts.

NDVI/EVI Forecast (30 days)

We’ll pull ~16-day composites over N years and align with daily POWER weather.
epochs learning rate
If TensorFlow.js is unavailable, a linear baseline fallback will be used.
—
Inputs: historical NDVI/EVI (MODIS/VIIRS) + daily POWER T2M & Precip; engineered features include lag values & seasonal encodings.