Raster Outputs

The plugin generates two types of GeoTIFF rasters: change frequency and change hotspots.

Change Frequency Raster

File: change_frequency.tif

A single raster showing how many times each pixel changed class across all time intervals.

Specifications

Property	Value
Data type	Int16
NoData value	-1
Value range	0 to (n-1), where n = number of input rasters
CRS	As specified in Options
Resolution	Same as input rasters

Values

Value	Meaning
-1	NoData (pixel invalid in one or more years)
0	No change (same class across all years)
1	Changed once
2	Changed twice
...	...
n-1	Changed in every interval

Example

For a 3-raster analysis (2010, 2015, 2020):

• 0: Pixel was class X in 2010, 2015, and 2020
• 1: Pixel changed either 2010→2015 OR 2015→2020
• 2: Pixel changed both 2010→2015 AND 2015→2020

Styling in QGIS

Recommended symbology for visualization:

1. Right-click the layer → Properties → Symbology
2. Choose Singleband pseudocolor
3. Set Min to 0 and Max to your maximum change count
4. Choose a sequential color ramp (e.g., Viridis, YlOrRd)
5. Click Classify

Example color scheme:

Value	Color	Description
0	Light gray	Stable/persistent
1	Yellow	Low change
2	Orange	Moderate change
3+	Red	High change

Use Cases

• Identify stable areas: Pixels with value 0
• Find dynamic zones: Pixels with high values
• Prioritize monitoring: Focus on frequently changing areas
• Validate classifications: High change frequency might indicate classification errors

---

Change Hotspot Rasters

File: change_hotspot_{year0}_{year1}.tif (one per interval)

Kernel density surfaces showing concentrations of change locations.

Specifications

Property	Value
Data type	Float32
NoData value	-9999
Value range	0 to maximum density
CRS	As specified in Options
Resolution	Same as input rasters
Kernel	Gaussian
Radius	1000 map units

How Hotspots Are Generated

1. All pixels that changed class are identified
2. Their centroids are converted to point locations
3. Up to 50,000 points are sampled (for performance)
4. Kernel Density Estimation (KDE) is applied
5. Result is a smooth density surface

Values

• 0: No nearby change
• Higher values: Greater concentration of change
• Values are not normalized between intervals

Relative Comparison

Hotspot values are relative within each raster. To compare across intervals, consider normalizing or using consistent symbology.

Styling in QGIS

Recommended visualization:

1. Right-click the layer → Properties → Symbology
2. Choose Singleband pseudocolor
3. Set interpolation to Linear
4. Use a heat-style color ramp (e.g., Magma, Inferno)
5. Set Min to 0 and let QGIS calculate Max

For consistent comparison across intervals:

• Note the maximum value across all hotspot rasters
• Set the same Min/Max for all layers
• Use identical color ramp settings

Use Cases

• Visualize change clusters: See where change is concentrated
• Identify change fronts: Urban expansion, deforestation edges
• Compare intervals: Which period had more concentrated change?
• Support narratives: "Change was concentrated in the northeast"

---

Working with Raster Outputs

Viewing in QGIS

Rasters are automatically added to your QGIS project. If not visible:

1. Layer → Add Layer → Add Raster Layer
2. Navigate to output directory
3. Select the .tif file

Exporting for Other Software

GeoTIFF is widely compatible:

• ArcGIS Pro / ArcMap
• Google Earth Engine
• R (raster, terra packages)
• Python (rasterio, GDAL)

Combining with Other Data

import rasterio
import numpy as np

# Read change frequency
with rasterio.open('change_frequency.tif') as src:
    freq = src.read(1)
    meta = src.meta

# Create binary mask of high-change areas
high_change = np.where(freq >= 2, 1, 0)
high_change = np.where(freq == -1, -1, high_change)

# Write output
meta.update(dtype='int16')
with rasterio.open('high_change_mask.tif', 'w', **meta) as dst:
    dst.write(high_change.astype('int16'), 1)

Zonal Statistics

Combine with polygon boundaries:

1. Load change frequency raster
2. Load administrative boundaries (polygon layer)
3. Use Processing → Raster Analysis → Zonal Statistics
4. Calculate mean change frequency per zone

Creating Maps

For publication-quality maps:

1. Style raster layers appropriately
2. Add basemap or context layers
3. Use Project → New Print Layout
4. Add map, legend, scale bar, north arrow
5. Export as PDF or image

---

Technical Notes

Memory and Performance

• Block-based processing keeps memory usage bounded
• Large rasters process in chunks (256×256 pixels)
• Hotspot generation samples up to 50,000 points for performance

Coordinate Reference System

Raster outputs use the CRS specified in Options:

• Default: Project CRS
• Can be set to any valid CRS
• Reprojection uses nearest-neighbor (preserves categorical values)

NoData Handling

• Pixels that are NoData in ANY input year get NoData in change frequency
• Hotspots only consider pixels valid in both years of the interval