● REMOTE SENSING KNOWLEDGE HUB LAT 20.5937° N   LON 78.9629° E   SCALE 1:250,000
Remote Sensing Knowledge Hub

Satellite data, made understandable.

A free, in-depth knowledge hub for satellite & remote-sensing research. Spatial Research Suite is a reference library of 38+ spectral indices and analyses, interactive browser-based tools, a glossary, dataset guides, and plain-English explainers — built for researchers, ecologists, and planners.

No installation A free, open research reference · runs entirely in your browser
Featured on Product Hunt
Try it live

See a spectral index in seconds.

Tap an index for its formula, the exact satellite bands, and how to read the results — then explore the full reference and the interactive tools.

Five tools, one boundary.

Set a study area once. Every tool below works from that same boundary, so you're not re-uploading shapefiles between steps.

CLASSIFICATION

Land-cover classification

Train Random Forest, CART, or SVM classifiers on your own ground-truth points. Includes per-pixel confidence mapping and a held-out validation accuracy report — not just training accuracy.

TIME SERIES

Multi-year trend detection

Fit a per-pixel linear trend across any year range, with a Mann-Kendall significance test to separate real change from noise. Harmonized Landsat 5/7/8/9 data extends coverage back to 1984.

CHANGE DETECTION

Spatial change mapping

Auto-generated transition matrices and a spatial change-detection map, with minimum mapping unit filtering to remove classification noise between two dates.

STATISTICS

Batch and correlation analysis

Run zonal statistics across dozens of boundaries in a single server-side call. Test correlations between any two variables — vegetation and temperature, for example — with a scatter plot and Pearson's r.

From boundary to publication figure.

The same four steps whether you're running a quick NDVI check or a full classification.

01

Define your study area

Upload a shapefile, draw a boundary directly on the map, or extract one from built-in administrative datasets (India down to village level, or any country globally).

02

Run an analysis or train a classifier

Pick from vegetation health, surface temperature, precipitation, nighttime lights, and a dozen other satellite datasets — or label training points and train a land-cover classifier.

03

Get statistically validated results

Every result comes with the numbers behind it: accuracy and kappa on held-out test data, Mann-Kendall significance, confidence intervals — not just a colored map.

04

Export a publication-ready output

Download a cartographic map with a north arrow, scale bar, and legend, plus an auto-generated methodology report listing exact datasets, parameters, and citations.

How it works, under the hood.

Every analysis is transparent, method-based, and grounded in peer-reviewed remote-sensing science — so your results hold up in a dissertation, a government report, or a grant proposal.

Cloud-based satellite analysis

Modern remote-sensing analysis runs on cloud platforms that hold decades of satellite imagery — Landsat, Sentinel-2, MODIS and more — as multi-petabyte archives no single workstation could store. Processing happens next to the data on distributed servers, so continent-scale analysis that once took weeks can run in minutes. Our guides and reference explain how these workflows work, end to end.

Land Use / Land Cover (LULC) classification

Our LULC classification engine turns raw spectral imagery into labelled thematic maps — water, forest, cropland, built-up land, bare soil. Supervised machine-learning classifiers (such as Random Forest and Support Vector Machines) learn the spectral signature of each class across the visible, near-infrared, and shortwave-infrared bands, so a flooded paddy field is distinguished from open water. Every classification is paired with a formal accuracy assessment — an overall accuracy figure and a Kappa coefficient from a confusion matrix — so you can report exactly how reliable your map is.

Coral health & reef monitoring

For coastal and marine researchers, GISforus offers coral health and bleaching-detection workflows. Healthy and bleached coral reflect light differently, and this contrast can be tracked from space using sea-surface temperature anomalies, water-column-corrected reflectance indices, and multi-date change analysis. By comparing imagery before and after a thermal-stress event, the platform highlights reef areas that have likely lost their symbiotic algae — a wide-area early-warning layer that directs, rather than replaces, in-water survey.

People who work with satellite data.

Researchers

Reproducible methodology reports and defensible accuracy metrics, for work that needs to hold up in peer review.

Urban planners

Track land-use change and urban heat over time across a district or state, without commissioning a custom analysis.

Environmental consultants

Batch-process zonal statistics across many boundaries at once, and export client-ready cartography directly.

Graduate students

Run a full classification workflow — training, validation, accuracy assessment — without setting up a Python or GEE development environment.

The details that make results defensible.

Specifics that matter if you're going to cite this in a report or a paper.

  • Validation accuracyAccuracy and kappa are computed on a held-out 30% test split, not on the same data the classifier trained on — with at least 20 training points. Below that, results are clearly labeled as training-data (resubstitution) accuracy, not held-out.
  • Multi-sensor harmonizationLandsat 5, 7, 8, and 9 bands are aligned to a common convention, extending analysis back to 1984 where Sentinel-2 (2015+) alone wouldn't reach.
  • Trend significanceMulti-year trends include a Mann-Kendall test, so you can distinguish a statistically significant change from year-to-year noise.
  • Cloud maskingSentinel-2 and Landsat imagery is cloud-masked at the pixel level before any statistic is computed.
  • Auto-generated citationsEvery export includes the exact datasets, date ranges, spatial resolution, and classifier parameters used — plus citations for the underlying data sources.

Who built this.

Anant Kumar Pathak

Dr. Anant Kumar Pathak — PhD, former Assistant Professor, working in GIS and spatial analysis. His doctoral research at Jamia Millia Islamia examined coastal tourism sustainability in Sindhudurg, Maharashtra, combining land-use change analysis with carrying-capacity assessment; his applied work spans carbon footprint mapping and Coastal Regulation Zone (CRZ) analysis — the same methods this knowledge hub explains in depth.

LinkedIn

Technology in service of ecological balance.

Our mission is to democratise planetary observation — to make rigorous Earth analysis accessible to every researcher, ecologist, and steward, regardless of coding ability or budget, so that better data leads to better care for the natural world.

Our values are organised around six commitments. We chose six deliberately: across cultures and disciplines it stands for harmony, balance, and responsibility — the equilibrium between people and the environment that sits at the centre of everything we build.

1 · Stewardship
Every tool should help protect and restore natural systems, not merely observe them.
2 · Rigour
Results must be scientifically sound, method-based, and reproducible — never a black box.
3 · Access
Planetary-scale analysis should reach those without code or capital, not only specialists.
4 · Transparency
Every algorithm, formula, and data source is documented and open to scrutiny.
5 · Balance
We help decision-makers weigh development against ecological limits with real evidence.
6 · Longevity
We build for the long horizon — the decades over which ecosystems actually change.

Questions worth answering upfront.

Do I need a GIS background to use this?

No. The interface walks you through each step (set a boundary, pick a feature or train a classifier, review results). That said, the outputs — accuracy metrics, significance tests, confidence intervals — are built for people who'll recognize and use them, so a research or planning background helps you get the most out of it.

How accurate are the classification results?

Accuracy and kappa are computed on a held-out test split (30% of your training points, never seen during training) whenever you provide at least 20 points — not on the same data the classifier trained on, which is a common shortcut that inflates the number. Below that threshold, the report tells you explicitly that you're seeing training-data accuracy, not held-out validation.

Is my uploaded data safe?

Boundary and shapefile data you upload or draw lives only in your active browser session — it's used to run the analysis you request and isn't permanently stored or shared with third parties. See the Privacy Policy for details.

What satellite data sources does it use?

Landsat 5, 7, 8, and 9 (harmonized back to 1984), Sentinel-2, CHIRPS precipitation, SRTM elevation, VIIRS nighttime lights, and WorldPop population density — all drawn from public satellite-data archives.

Can I cite results from this in a paper or report?

Yes — every export includes an auto-generated methodology note listing the exact datasets, date ranges, spatial resolution, and classifier parameters used, plus citations for the underlying data sources, so your methodology section is reproducible.

Everything you need to understand the methods.

Explore the Resources →