Spatial Research Suite

Loading Spatial Research Suite...

Landsat Engine | Random Forest ML | Cloud Data

Database Extractor
Custom Draw ROI

1. Data Source ⓘ
Choose which administrative boundary dataset to browse — India's admin levels (State/District/Tehsil) or Global country/state boundaries.

India Database (Local)

Global Repository (Natural Earth)

2. Extractor ⓘ
Search or click on the map to select one or more boundary features. Selected features highlight on the map and can be sent to the Clipboard for use elsewhere in the app.

Extraction Level:

Filter State:

Filter District:

Filter Taluka:

Extraction Level:

Filter Country:

3. Selection & Actions

Selected Features:

Merge Selected Boundaries (Dissolve)

Draw your custom area directly on the map using the polygon tool.

Name your Region:

Reference Map

Navigate and click on polygons to extract or draw your own.

Step 1: Study Area ⓘ
Set the area you want to classify. Upload a shapefile or fetch a boundary from the Data Clipboard.

System Note: Dynamic Scaling Active. Safely handles any district or state size.

Upload Target Area (.shp, .shx, .dbf, .prj)

Browse...

Step 2: Landsat Engine ⓘ
Fetches cloud-masked Landsat 8/9 imagery for your boundary and date range — this becomes the input for both spectral indices and the classifier.

Rendered on Google Servers with Pixel-wise Cloud Masking.

Select Date Range:

Scene Cloud Tolerance (%):

Band Combination Setup:

NIR:

SWIR:

Step 3: Ground Truth ⓘ
Draw training points/polygons on the map and label each with a land-cover class (e.g. Forest, Water, Urban) — the classifier learns from these examples.

Select Input Method:

Upload Data Draw on Map

Upload (.shp, .shx, .dbf, .prj OR .geojson)

Browse...

Left-Click to start drawing points or polygons.

Click on the first point to finish the polygon.

Save Shapefile (ZIP)

Step 4: Machine Learning ⓘ
Trains a classifier (Random Forest, CART, or SVM) on your labeled training points, then applies it to classify the whole boundary into land-cover classes. Also computes accuracy, Kappa, and (for RF/CART) a confidence map.

Classifier Algorithm:

Number of Trees (Random Forest):

Generate Spectral Indices:

NDVI NDWI EVI NDBI

Step 5: Spatiotemporal Automation ⓘ
Re-runs your trained classifier across multiple years to show how land cover changed over time, with an auto-generated transition matrix and spatial Change Detection Map.

Generate Temporal Gallery & Auto-Transition Matrix.

Limited to 2013+ (Landsat 8/9) — classification accuracy isn't reliable on older sensors (Landsat 5/7) since the model is trained on Landsat 8/9's specific spectral characteristics.

Start Year:

End Year:

Time Interval:

Select Month:

Spatial map of WHERE land cover changed between the first and last year of your Temporal Gallery (run 'Generate Grid & Matrix' first). Highlighted areas changed class; the rest stayed the same.

Spatiotemporal LULC Change Grid

Temporal LULC Area Statistics

Temporal LULC Area Data

LULC Transition Heatmap

Shows the precise area (Sq.Km) that transitioned from a base classification to a target classification.

Transition Matrix Table (Sq.Km)

Area Distribution Chart

Tabular Zonal Metrics

Random Forest Validation Metrics

Class-wise Accuracy

System Logs

1. Data Source ⓘ
Set the area you want to analyze. Upload a shapefile (.shp+.shx+.dbf+.prj together) or fetch a boundary you previously sent to the Data Clipboard from Shapefile Extractor.

— OR UPLOAD FILE —

Upload Target Area (.shp, .shx, .dbf, .prj)

Browse...

Select Cloud Analysis:

Temporal Aggregation:

Date Range:

2. Run Analytics ⓘ
Fetches satellite imagery for your selected feature (NDVI, LST, etc.) and date range, computes statistics (mean, median, histogram), and renders it on the map. Cached for 15 minutes if you repeat the exact same request.

3. Spatiotemporal Grid ⓘ
Generates a grid of maps — one panel per time interval — so you can visually track how your selected feature changed over multiple years. Automatically widens the search window for years with sparse cloud-free data.

Start Year:

End Year:

Time Interval:

Select Month:

Advanced Zonal Metrics

Index Area Distribution

This chart profiles the landscape by grouping continuous index values into distinct ranges and calculating the total geographic area covered by each range.

Area Distribution Histogram

Spatiotemporal Scientific Grid

Multi-Year Per-Pixel Trend ⓘ
Fits a linear regression per pixel across the years you select, showing whether your feature is increasing (red) or decreasing (blue) over time. Includes a Mann-Kendall test to flag which areas show a STATISTICALLY significant trend (p<0.05), not just noise.

Computes a linear regression slope per pixel across years for the currently selected feature (e.g. NDVI change/year, LST warming/cooling rate), with a Mann-Kendall test for statistical significance.

Start Year

End Year

Month

Show only statistically significant trends (Mann-Kendall, p < 0.05)

Batch Zonal Statistics — Multiple Boundaries at Once ⓘ
Runs one Earth Engine server-side call across ALL boundaries in your current multi-feature selection (e.g. every district in a state), instead of repeating the analysis one boundary at a time. Requires selecting multiple features WITHOUT dissolving them in Shapefile Extractor first.

Runs the feature/date range/aggregation currently selected in Step 2 across EVERY boundary in your current selection in one go (e.g. all districts in a state) — much faster than repeating Step 2 one boundary at a time.

Setup: In Shapefile Extractor, select multiple features without checking 'Merge Selected Boundaries (Dissolve)', send to Clipboard, then import that as your boundary here (Step 1).

Download Batch CSV

Cartography Studio ⓘ
Generates a publication-ready map (with north arrow, scale bar, legend, and coordinates) from any analysis you've already run in GEE Cloud Analytics, Trend Analysis, or LULC Classification.

Generate publication-ready maps with geographic coordinates, north arrow, and accurate scale bar.

2. Map Title:

3. Legend Title:

4. Map Background:

Element Positioning:

Scale Bar:

North Arrow:

Export Options

PREVIEW ONLY: The watermark shown on the map below is for preview purposes only and will be entirely removed in your final downloaded high-resolution asset.

Map Preview

1. Correlation Analysis ⓘ
Tests whether two features are statistically related across your boundary — e.g. does vegetation (NDVI) correlate with surface temperature (LST)? Samples paired pixel values and computes Pearson's r, R², and a scatter plot with a regression line.

Feature X:

Feature Y:

Aggregation:

Date Range:

Sample Points:

2. Confidence Intervals ⓘ
Computes the 95% confidence interval around the mean for your currently selected feature (from GEE Cloud Analytics, Step 2) — tells you how statistically reliable that average is, based on sample size and variability.

3. Outlier Detection ⓘ
Flags boundaries in your Batch Processing results (GEE Cloud Analytics > Batch Processing) whose Mean value is a statistical outlier (|Z-score| > 2) compared to the rest of the group.

Run 'Run Batch Zonal Stats' in GEE Cloud Analytics > Batch Processing first — this analyzes those results.

1. Data Source ⓘ Choose which administrative boundary dataset to browse — India's admin levels (State/District/Tehsil) or Global country/state boundaries.

2. Extractor ⓘ Search or click on the map to select one or more boundary features. Selected features highlight on the map and can be sent to the Clipboard for use elsewhere in the app.

3. Selection & Actions

Draw your custom area directly on the map using the polygon tool.

Reference Map

Step 1: Study Area ⓘ Set the area you want to classify. Upload a shapefile or fetch a boundary from the Data Clipboard.

Step 2: Landsat Engine ⓘ Fetches cloud-masked Landsat 8/9 imagery for your boundary and date range — this becomes the input for both spectral indices and the classifier.

Step 3: Ground Truth ⓘ Draw training points/polygons on the map and label each with a land-cover class (e.g. Forest, Water, Urban) — the classifier learns from these examples.

Step 4: Machine Learning ⓘ Trains a classifier (Random Forest, CART, or SVM) on your labeled training points, then applies it to classify the whole boundary into land-cover classes. Also computes accuracy, Kappa, and (for RF/CART) a confidence map.

Step 5: Spatiotemporal Automation ⓘ Re-runs your trained classifier across multiple years to show how land cover changed over time, with an auto-generated transition matrix and spatial Change Detection Map.

Spatiotemporal LULC Change Grid

Temporal LULC Area Statistics

Temporal LULC Area Data

LULC Transition Heatmap

Transition Matrix Table (Sq.Km)

Area Distribution Chart

Tabular Zonal Metrics

Random Forest Validation Metrics

Class-wise Accuracy

System Logs

1. Data Source ⓘ Set the area you want to analyze. Upload a shapefile (.shp+.shx+.dbf+.prj together) or fetch a boundary you previously sent to the Data Clipboard from Shapefile Extractor.

2. Run Analytics ⓘ Fetches satellite imagery for your selected feature (NDVI, LST, etc.) and date range, computes statistics (mean, median, histogram), and renders it on the map. Cached for 15 minutes if you repeat the exact same request.

3. Spatiotemporal Grid ⓘ Generates a grid of maps — one panel per time interval — so you can visually track how your selected feature changed over multiple years. Automatically widens the search window for years with sparse cloud-free data.

Advanced Zonal Metrics

Index Area Distribution

Area Distribution Histogram

Spatiotemporal Scientific Grid

Multi-Year Per-Pixel Trend ⓘ Fits a linear regression per pixel across the years you select, showing whether your feature is increasing (red) or decreasing (blue) over time. Includes a Mann-Kendall test to flag which areas show a STATISTICALLY significant trend (p<0.05), not just noise.

Cartography Studio ⓘ Generates a publication-ready map (with north arrow, scale bar, legend, and coordinates) from any analysis you've already run in GEE Cloud Analytics, Trend Analysis, or LULC Classification.

Export Options

Map Preview

1. Correlation Analysis ⓘ Tests whether two features are statistically related across your boundary — e.g. does vegetation (NDVI) correlate with surface temperature (LST)? Samples paired pixel values and computes Pearson's r, R², and a scatter plot with a regression line.

2. Confidence Intervals ⓘ Computes the 95% confidence interval around the mean for your currently selected feature (from GEE Cloud Analytics, Step 2) — tells you how statistically reliable that average is, based on sample size and variability.

3. Outlier Detection ⓘ Flags boundaries in your Batch Processing results (GEE Cloud Analytics > Batch Processing) whose Mean value is a statistical outlier (|Z-score| > 2) compared to the rest of the group.

1. Data Source ⓘ
Choose which administrative boundary dataset to browse — India's admin levels (State/District/Tehsil) or Global country/state boundaries.

2. Extractor ⓘ
Search or click on the map to select one or more boundary features. Selected features highlight on the map and can be sent to the Clipboard for use elsewhere in the app.

Step 1: Study Area ⓘ
Set the area you want to classify. Upload a shapefile or fetch a boundary from the Data Clipboard.

Step 2: Landsat Engine ⓘ
Fetches cloud-masked Landsat 8/9 imagery for your boundary and date range — this becomes the input for both spectral indices and the classifier.

Step 3: Ground Truth ⓘ
Draw training points/polygons on the map and label each with a land-cover class (e.g. Forest, Water, Urban) — the classifier learns from these examples.

Step 4: Machine Learning ⓘ
Trains a classifier (Random Forest, CART, or SVM) on your labeled training points, then applies it to classify the whole boundary into land-cover classes. Also computes accuracy, Kappa, and (for RF/CART) a confidence map.

Step 5: Spatiotemporal Automation ⓘ
Re-runs your trained classifier across multiple years to show how land cover changed over time, with an auto-generated transition matrix and spatial Change Detection Map.

1. Data Source ⓘ
Set the area you want to analyze. Upload a shapefile (.shp+.shx+.dbf+.prj together) or fetch a boundary you previously sent to the Data Clipboard from Shapefile Extractor.

2. Run Analytics ⓘ
Fetches satellite imagery for your selected feature (NDVI, LST, etc.) and date range, computes statistics (mean, median, histogram), and renders it on the map. Cached for 15 minutes if you repeat the exact same request.

3. Spatiotemporal Grid ⓘ
Generates a grid of maps — one panel per time interval — so you can visually track how your selected feature changed over multiple years. Automatically widens the search window for years with sparse cloud-free data.

Multi-Year Per-Pixel Trend ⓘ
Fits a linear regression per pixel across the years you select, showing whether your feature is increasing (red) or decreasing (blue) over time. Includes a Mann-Kendall test to flag which areas show a STATISTICALLY significant trend (p<0.05), not just noise.

Cartography Studio ⓘ
Generates a publication-ready map (with north arrow, scale bar, legend, and coordinates) from any analysis you've already run in GEE Cloud Analytics, Trend Analysis, or LULC Classification.

1. Correlation Analysis ⓘ
Tests whether two features are statistically related across your boundary — e.g. does vegetation (NDVI) correlate with surface temperature (LST)? Samples paired pixel values and computes Pearson's r, R², and a scatter plot with a regression line.

2. Confidence Intervals ⓘ
Computes the 95% confidence interval around the mean for your currently selected feature (from GEE Cloud Analytics, Step 2) — tells you how statistically reliable that average is, based on sample size and variability.

3. Outlier Detection ⓘ
Flags boundaries in your Batch Processing results (GEE Cloud Analytics > Batch Processing) whose Mean value is a statistical outlier (|Z-score| > 2) compared to the rest of the group.