Skip to content

RA · Foundations

Learning geospatial analysis, Earth observation and GeoAI: a guide to the field

A learning guide to geospatial analysis, Earth observation and GeoAI: core ideas, an order of study, open data and standards and how expertise is shown.

As of
17 November 2026
पठन
11 minutes
साक्ष्य
10 sources, cited in order
स्तर
Introductory to intermediate
अध्ययन समय
15 minutes
NEXUS TWIN · WF-1 · GLOBESEED 04ab8c74RA-F04.DISCIPLINE.GEOSPATIAL-GEOAIxyzSCHEMATIC · GLOBEST-01 · 0.484 0.532ST-02 · 0.758 0.719
WF-1 · globe Schematic wireframe twin (globe, climate system). An illustration, not data.

What you will be able to do

  • Distinguish vector and raster data and name one use of each.
  • Explain why harmonisation and interoperability matter in Earth observation.
  • Describe the layers of geospatial work.
  • Write a validation note that reports error against reference data.
स्तर
Introductory to intermediate
अध्ययन समय
15 minutes
Before you start
The learning guide to the Faculty of Risk Science and Methods, or equivalent background
अभिलेख
Since 1972
Landsat has continuously acquired images of the Earth's land surface2
Coordination
1984
year CEOS was established to coordinate space-based Earth observation3
मानक
80+
open geospatial specifications in the OGC standards listing4

01What geospatial analysis, Earth observation and GeoAI covers

What geospatial analysis, Earth observation and GeoAI covers

Geospatial analysis, Earth observation and GeoAI is the Academy discipline of geospatial analysis, Earth observation and GeoAI, one of the disciplines of the Faculty of Risk Science and Methods. It covers how data tied to places are stored and analysed, how satellites and other sensors observe the Earth, and how machine learning is applied to the resulting data.

Earth observation is the supply side. Copernicus describes itself as the Earth observation component of the European Union's Space programme, looking at the planet and its environment to benefit European citizens, with information services that draw from satellite Earth observation and in-situ data 1. The European Commission manages the programme. Member states, the European Space Agency, EUMETSAT, ECMWF and other bodies take part in implementing it 1.

Geospatial analysis is the method side, and GeoAI joins it to machine learning. The discipline sits beside AI and machine learning for risk analysis, catastrophe, exposure and vulnerability modelling and risk indicators and composite indices.

02Why place matters in risk work

Why place matters in risk work

Risk is local. A flood, a fire or a drought affects some places and not others, and exposure depends on what is built or grown where. Maps are how those facts are seen together.

Long records make change visible. The Landsat programme is jointly managed by NASA and the US Geological Survey. Since 1972 its satellites have continuously acquired images of the Earth's land surface and provided an uninterrupted data archive 2. A learner who understands such a record can ask what has changed at a site and over what period.

Open access has widened the field. The Copernicus Data Space Ecosystem offers access to open datasets from the Copernicus services to support environmental monitoring, disaster response and climate analysis 6. NASA operates platforms that collect Earth science data used to support scientific and societal applications 7. These sources are not a substitute for care. A learner must still ask what a sensor can see, at what resolution and how often.

03How the field fits together

How the field fits together

A useful map has four layers. Data covers vector and raster formats, coordinate systems and metadata. Observation covers sensors, orbits, resolution and revisit time. Analysis covers overlays, change detection, classification and validation. Learning covers machine learning models trained on imagery and other layers. Figure 2 shows the layers.

Standards hold the layers together. The Open Geospatial Consortium (OGC) describes its standards as internationally recognised specifications that let different systems exchange information, organised by functional area. It lists more than 80 specifications and says implementers are encouraged to use the newer OGC API standards in place of older web service standards 4. The SpatioTemporal Asset Catalog (STAC) specification is described as a common language to describe geospatial information so that it can more easily be worked with, indexed and discovered 5.

Coordination between space agencies has a long history. The Committee on Earth Observation Satellites (CEOS) was established in September 1984. Its original function was to coordinate and harmonise Earth observations to make it easier for users to access and use data, and it initially focused on interoperability, common data formats, inter-calibration of instruments and common validation and inter-comparison of products 3. These concerns remain the heart of the craft.

Learners often move straight to the model and skip the earlier layers. That is the most common source of error in this field. A classification that looks impressive may rest on mismatched dates, an unknown cloud mask or a coordinate system that was never checked. Working up from data and geometry guards against this. It also makes the later work easier, because a learner who knows what is in a product can judge what a model can safely learn from it.

Fig. 02Four layers of geospatial work

Four layers of geospatial workFour stacked layers: data in vector and raster formats, observation by sensors and satellites, analysis including change detection and validation, and learning with machine learning models, with standards drawn as the link between them.01अधिगमModels trained on Earth data02AnalysisChange, classes, validation03अवलोकनSensors, orbits, resolution04डेटाFormats, coordinates, metadata
A teaching map of the layers of the field, with standards linking them 4,5.4,5Sources: OGC; STAC; Academy desk teaching map. Figure CC BY 4.0. As of 17 November 2026.
Figure 2 in words

Four layers stacked from top to bottom. Learning: machine learning models trained on imagery and other layers, highlighted. Analysis: overlays, change detection, classification and validation. Observation: sensors, orbits, resolution and revisit time. Data: vector and raster formats, coordinate systems and metadata. Standards such as OGC specifications and STAC link the layers.

04A suggested learning order in four stages

A suggested learning order in four stages

The Academy describes study at four levels: introductory, intermediate, advanced and integrative. The four stages below follow the same arc for geospatial analysis, Earth observation and GeoAI. They are a suggestion, and a learner can enter at any stage.

Stage one: data and geometry. Start with how place is represented. Learn vector and raster data, coordinate reference systems and the idea of metadata. Read the STAC about page to see how a catalogue describes assets in space and time 5.

Stage two: sensors and products. Next, study what satellites measure. Read the CEOS background for why harmonisation matters 3, the Landsat description of a long record 2 and the NASA Earthdata pages on platforms and data 7. Note the difference between a sensor, a product and a derived indicator.

Stage three: analysis and validation. Then study analysis. Work with open data from the Copernicus Data Space Ecosystem 6 and practise change detection and simple classification. Always check results against reference data and record how. Keep a short log of each processing step, with the tool, version and settings, so that another person can repeat the work and reach the same map.

Stage four: GeoAI and its limits. Finally, study machine learning on Earth data. Read the review by Reichstein and colleagues on deep learning and process understanding for data-driven Earth system science 8, and the ECMWF account of a data-driven forecast model trained on reanalysis data 9. Practise asking what a model has learned from, and where it should not be used. A model trained on imagery from one region and season may fail in another, and the only way to find out is to test it there.

The order is a suggestion. A practitioner may begin at stage three and return to stage one later. What matters is that each new idea has the ideas beneath it in place.

Fig. 01Four stages of study in geospatial analysis, Earth observation and GeoAI

Four stages of study in geospatial analysis, Earth observation and GeoAIA rising ladder of four steps: data and geometry, sensors and products, analysis and validation, and GeoAI and its limits, each with the sources and programmes a learner meets at that step.01Data and geometryVector, raster, metadata02Sensors and productsCEOS, Landsat, Earthdata03Analysis and validationChange, classes, error04GeoAI and its limitsLearning from Earth data
The suggested learning order of this guide, with the sources that anchor each stage 5,3,2,7,6,8,9. The order is a suggestion, not a required curriculum.5,3,2,7,6,8,9Sources: STAC; CEOS; NASA and USGS Landsat; NASA Earthdata; Copernicus; Nature; ECMWF. Figure CC BY 4.0. As of 17 November 2026.
Figure 1 in words

Four steps rise from left to right. Step one, data and geometry: vector and raster data and catalogues such as STAC. Step two, sensors and products: CEOS, Landsat and NASA Earthdata. Step three, analysis and validation: open data from the Copernicus Data Space Ecosystem. Step four, highlighted, GeoAI and its limits: deep learning for Earth system science and data-driven forecasting.

Fig. 03From a satellite pass to a validated map

From a satellite pass to a validated mapA four-stage cascade: a satellite acquires an image, the product is processed and catalogued, an analysis detects change, and the result is validated against reference data before it is used.ACQUIRESatellite imageCATALOGUEProcessed productमेटाडेटाANALYSEDetect changeVALIDATEReference dataReport error
A teaching sequence for a change map built from open satellite data 2,6.2,6Sources: NASA and USGS Landsat; Copernicus Data Space Ecosystem; Academy desk teaching map. Figure CC BY 4.0. As of 17 November 2026.
Figure 3 in words

Four stages run from left to right: acquire, catalogue, analyse and validate. A satellite acquires an image of the land surface. The product is processed and catalogued with metadata. An analysis detects change between dates. The result is checked against reference data before anyone relies on it.

05Six core ideas every learner should hold

Six core ideas every learner should hold

Six ideas carry most of the weight in this discipline. Each is simple to state and takes practice to use.

Data differ in resolution and spectral coverage. NASA describes Landsat as medium-resolution satellite data and says broad spectral coverage, beyond visible and near-infrared light, is essential to accurately study Earth's surface 2.

Records are only comparable if harmonised. CEOS began with interoperability, common formats and inter-calibration for this reason 3.

Metadata make data usable. STAC exists so that assets can be indexed and discovered 5.

Standards are tools, not decoration. OGC specifications let different systems exchange information 4.

Validate against reference data. A map from a model is a hypothesis until it is checked.

Machine learning learns from what it is given. The ECMWF AIFS was trained on a subset of the ERA5 reanalysis for 1979–2018, and its authors say what that does and does not cover 9.

06Canonical references and data to know

Canonical references and data to know

A learner should know a short list of references and what kind of evidence each holds. Read primary documents first.

For programmes and data: Copernicus 1, the Copernicus Data Space Ecosystem 6, Landsat 2 and NASA Earthdata 7. For coordination: CEOS 3. For standards: OGC 4 and STAC 5. For learning: the NASA ARSET trainings 10. For GeoAI: the Reichstein review 8 and the ECMWF account of the AIFS 9. Learners should record the product name, version, date and licence of every dataset they use.

07Skills that make the knowledge usable

Skills that make the knowledge usable

Knowledge becomes useful through a small set of skills. The discipline treats these as its practical core.

Handling coordinate systems. Know when data are in different projections and how to convert them. Reading a product description. Find the resolution, date, processing level and known limits. Building a catalogue. Describe assets with metadata that others can search. Detecting change. Compare two dates and separate real change from noise. Validating. Compare outputs with reference data and report error. Explaining a map. State what it shows, what it does not and how confident it is.

08Open courses and pathways

Open courses and pathways

The NASA Applied Remote Sensing Training Program (ARSET) describes trainings that build skills to use remote sensing data for applications including air quality, public health, agriculture, disasters, ecological conservation, water resources, climate resilience and wildland fires. Its page lists trainings with filters from Fundamentals to Advanced levels, including a Fundamentals of Remote Sensing training 10. The page was opened on 17 November 2026. It does not say whether a training leads to a qualification, and this guide does not either.

Open data are a pathway in themselves, and learners who work through a real product learn faster than those who only read about it. The Copernicus Data Space Ecosystem and Landsat archives let a learner work with real products 2 6. A suggested order is the ARSET fundamentals, then a small project with open data, then a written note on validation.

The Academy catalogue also lists a Geospatial and Risk Data Science studio, which this guide does not describe as open. No other course is named, because none was opened and confirmed as open to the public for this edition.

09How expertise is shown

How expertise is shown

Expertise in this field is shown by work that others can check. The Academy records learning, not credentials.

Finishing a course gives a completion record. It is not a certificate, an accreditation or a qualification, and it does not state that a person is competent in a profession. Achievements on the Academy recognise assessed learning only.

The Academy catalogue lists two courses for this discipline: Geospatial analysis, Earth observation and GeoAI: foundations and practice, at the intermediate level, and Geospatial analysis, Earth observation and GeoAI: frontier methods, at the advanced level. On 17 November 2026 the catalogue page shows the first as in authoring and the second as planned. Nothing is open for enrolment until it is published in the NXS Learning catalogue, so this guide does not point to an Academy course as a place to enrol.

Outside the Academy, the strongest evidence of expertise is a body of work: a documented map or change analysis with its data sources and versions, a validation note that reports error, or a catalogue entry that follows an open specification. Learners are encouraged to keep such a record from their first course.

10Emerging areas to watch

Emerging areas to watch

Several areas are changing what a geospatial analyst needs to know. Each is a place to read on once the core is in place.

Data-driven models of the Earth system. Reichstein and colleagues examine deep learning and process understanding for Earth system science 8, and ECMWF reports continuing work on its data-driven model 9. Newer standards. OGC encourages use of its newer API standards 4. Open data ecosystems. The Copernicus Data Space Ecosystem offers flexible access, integration and deployment options for Sentinel data 6. Programme changes. The Copernicus main website says it is now an archive, with important pages moved to the eu-space.europa.eu site, so learners should check where current material lives 1.

Related work across the Nexus network gives wider context. See geospatial and Earth observation research at the Global Centre for Risk and Innovation, climate and nature risk at the Global Risks Alliance, climate and nature governance at the Global Risks Forum and the proposed Nexus Standards dataset metadata profile. These pages describe programmes and proposals. A link to them implies no endorsement of the sources this guide cites.

What this guide does not claim: it is suggested learning, not a curriculum that anyone must follow, and it is not professional, legal, medical or technical advice. It names programmes, courses and bodies as places to learn and implies no partnership with any of them. It reports figures as their issuers published them and draws no new estimates from them. Risk Academy does not certify competence, and a completion record is not a professional qualification. The Academy records no partners, cohorts or learner outcomes for this faculty, and this guide implies none. The Nexus Standards are proposed and in draft; nothing here treats them as adopted.

प्रश्न

प्रश्न और उत्तर

Where should a complete beginner start in geospatial analysis and Earth observation?

Start with stage one on data and geometry, and the NASA ARSET Fundamentals of Remote Sensing training described on its page.

What is the difference between Earth observation and GeoAI?

Earth observation gathers data about the planet from sensors. GeoAI applies machine learning to such data, often with other spatial layers.

Do I need to program to follow this guide?

Not for stages one and two. Stages three and four are easier with a scripting language and a geospatial library.

Why do standards matter?

They let different systems exchange and find data. OGC describes its standards as specifications that let different systems exchange information.

Do courses in this discipline lead to a qualification?

No. A completion record shows that you finished a course. It is not a certificate or a professional qualification.

शर्तें

शब्दावली

पृथ्वी अवलोकन
The gathering of information about the planet from satellites and other sensors.
Raster
Data stored as a grid of cells, each with a value, as in a satellite image.
Vector
Data stored as points, lines and polygons with coordinates.
मेटाडेटा
Data that describe other data, such as source, date, resolution and licence.
Revisit time
How often a sensor observes the same place.
GeoAI
The use of machine learning and related methods on geospatial data.

Check your understanding

  1. Why did CEOS focus on interoperability and common formats?

    So that users could access and use observations from different satellites and compare their products.

  2. What does STAC do?

    It gives a common way to describe geospatial assets so they can be indexed and discovered.

  3. Why validate a map from a model?

    Without reference data it is only a hypothesis, and errors cannot be measured.

Continue with a course

स्रोत

संदर्भ

10 sources, numbered in the order they are first cited
  1. European Union, Copernicus programme (2026). About Copernicus. www.copernicus.eu. Accessed 2026-11-17.
  2. NASA and U.S. Geological Survey, Landsat programme (2026). About Landsat. landsat.gsfc.nasa.gov. Accessed 2026-11-17.
  3. Committee on Earth Observation Satellites (2026). Background. ceos.org. Accessed 2026-11-17.
  4. Open Geospatial Consortium (2026). OGC Standards. www.ogc.org. Accessed 2026-11-17.
  5. STAC community (2026). SpatioTemporal Asset Catalogs (STAC). stacspec.org. Accessed 2026-11-17.
  6. European Union, Copernicus programme (Data Space Ecosystem) (2026). Copernicus Data Space Ecosystem. dataspace.copernicus.eu. Accessed 2026-11-17.
  7. National Aeronautics and Space Administration (NASA Earthdata) (2026). Earthdata Learn. www.earthdata.nasa.gov. Accessed 2026-11-17.
  8. Nature (Reichstein, Camps-Valls, Stevens and Jung) (2019-02). Deep learning and process understanding for data-driven Earth system science. doi.org. Accessed 2026-11-17.
  9. European Centre for Medium-Range Weather Forecasts (2024-01). AIFS: a new ECMWF forecasting system. www.ecmwf.int. Accessed 2026-11-17.
  10. National Aeronautics and Space Administration (NASA Earthdata) (2026). Applied Remote Sensing Training Program (ARSET). www.earthdata.nasa.gov. Accessed 2026-11-17.

अभिलेख

Editorial record

Facts and course states are given as of 17 November 2026. This guide is reviewed at least every 270 days and whenever a cited report publishes a new edition or a course of the faculty opens.

Published by the Academy desk of Risk Academy. To report an error or request a correction, write through the Risk Academy contact page; corrections are dated and listed with the page. The Academy's statement on AI use explains how AI assistance is used and reviewed. Text, visuals and data are licensed CC BY 4.0, except where a cited source states other terms.

Cite this article

Academy desk, Risk Academy (2026, November 17). Geospatial analysis, Earth observation and GeoAI guide. Risk Academy. https://risksacademy.com/risk-science/learning-geospatial-analysis-earth-observation-and-geoai-guide/

Republish

Republish or translate this article free of charge, in print or online, under CC BY 4.0: keep the byline and Risk Academy, link to the original and note any change. Figures, maps and twins made by the institutions carry the same licence; third-party data keeps its own credit line.

Attribution to paste
<p>Learning geospatial analysis, Earth observation and GeoAI: a guide to the field — Academy desk, Risk Academy, Risk Academy. First published 17 November 2026 at <a href="https://risksacademy.com/risk-science/learning-geospatial-analysis-earth-observation-and-geoai-guide/">https://risksacademy.com/risk-science/learning-geospatial-analysis-earth-observation-and-geoai-guide/</a>. Republished under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/).</p>

हिस्सा

EmailLinkedInBlueskyMastodon