geoflexforecast 0.1.0

Flexible raster time-series forecasting using native and aggregated spatial grids

geoflexforecast

geoflexforecast is a Python package for flexible raster time-series forecasting.

The package allows users to:

• Forecast rasters at native pixel level
• Aggregate pixels into user-defined blocks
• Reduce computational load for massive raster datasets
• Apply ARIMA forecasting across raster time series

Forecasting Modes

Native Mode

Forecast every raster pixel independently.

Aggregate Mode

Aggregate raster pixels into blocks such as:

• 2x2
• 5x5
• 10x10

before forecasting.

This is useful for large rasters and limited computational resources.

Example

from geoflexforecast import forecast_raster_series

forecast_raster_series(
    raster_paths=[
        "NDVI_2001.tif",
        "NDVI_2002.tif",
        "NDVI_2003.tif"
    ],
    output_path="forecast.tif",
    mode="aggregate",
    block_size=10,
    aggregation="mean",
    order=(1, 1, 1),
    min_obs=3
)