Time-Series Change Analysis

What This Tool Does

Time Series Change Intelligence analyzes a multiband raster time series to detect trends, structural breaks, seasonal behavior, and anomalies. It produces change rasters, a confidence raster, a summary JSON, and an HTML report.

Typical Questions This Tool Helps Answer

Is the change observed in this pixel a long-term trend, a seasonal cycle, a sudden structural break, or a short-lived anomaly?
Which locations show statistically significant long-term decline or recovery versus high-frequency variability that is likely noise?
What is the timing and magnitude of the most significant disturbance event detected in this time series?

When To Use

Forest disturbance and recovery monitoring
Crop stress and phenology tracking
Urban expansion timing checks
Wetland or hydrology change screening

What You Need

Input	Description
Time-series raster stack	One band per acquisition, ordered chronologically.
Optional QA stack	A matching multiband QA stack where positive values mark valid observations.

Key Settings

Setting	Default	Guidance
`algorithm_mode`	`fast`	Use `fast` for screening, `iterative` for breakpoint-focused work, or `bfast` when decomposition stability matters more than speed.
`min_observations`	24	Raise if the stack is sparse or noisy.
`break_threshold`	0.08	Raise to suppress weak break candidates.
`cadence_profile`	`auto`	Let the tool adapt to dense, sparse, or seasonal sampling patterns.

What You Get

Deliverable	Format	Description
`trend_change`	Raster	Trend or change surface.
`breakpoint_count`	Raster	Breakpoint count or dominant breakpoint index.
`breakpoint_date`	Raster	Breakpoint date or date index.
`change_confidence`	Raster	Confidence score in the range 0 to 1.
`summary`	JSON	Summary metrics and QA guidance.
`html_report`	HTML	Human-readable report.

Runtime Output Keys

result.outputs["trend_change"]
result.outputs["breakpoint_count"]
result.outputs["breakpoint_date"]
result.outputs["change_confidence"]
result.outputs["summary"]
result.outputs["html_report"]

Common Questions

Q: Which result should I review first? A: Start with summary.changed_fraction_valid_pixels and summary.mean_confidence, then inspect trend_change and change_confidence together.

Q: What is a common interpretation mistake? A: Treating breakpoint_count as impact severity. It reflects temporal complexity, not necessarily impact magnitude.

Q: Which settings most affect change detection? A: break_threshold, min_observations, and the selected cadence_profile usually produce the largest change-rate differences.

Q: How should teams use the outputs operationally? A: Prioritize areas with high change fraction and stable confidence, then use breakpoint_date timing to schedule targeted verification windows.

Results Delivery Checklist

The input stack is chronological and co-registered
The QA stack, if used, is aligned to the input stack
summary["changed_fraction_valid_pixels"] and summary["mean_confidence"] were reviewed together
Breakpoints were checked against known events or disturbance dates

Operational Notes

Evaluate changed_fraction_valid_pixels and mean_confidence together; high change with low confidence should trigger follow-up QA before action.
Use breakpoint_date for investigation timing, and treat breakpoint_count as temporal complexity rather than impact severity.
For sparse stacks, document the applied cadence profile and adjusted thresholds from summary output before comparing runs.

remote_sensing_change_detection
multi_sensor_fusion_monitoring
sar_analysis_readiness

References

Runtime implementation: wbtools_pro/src/tools/remote_sensing/time_series_change_intelligence.rs
Earth Observation and SAR Operations bundle overview: manual/pro-tools-customer/src/earth_observation_sar/overview.md

When To Use This Workflow

Use Time Series Change Intelligence when you need repeatable multi-date trend and breakpoint screening with confidence-aware outputs for monitoring and reporting programs.

Whitebox Workflows Pro Customer Technical Reference