# Represent places and changes over time

Use geospatial similarity while retaining the location and observation period.

Author: Murali Sid (https://linkedin.com/in/muralisid)
Source: https://www.agenticarchitectureskills.com/memory/location-and-satellite-data
Published: 2026-09-17
Updated: 2026-10-15
Licence: CC BY-SA 4.0, https://creativecommons.org/licenses/by-sa/4.0/
Cite as: Murali Sid. Represent places and changes over time. Agentic Architecture Skills. https://www.agenticarchitectureskills.com/memory/location-and-satellite-data

---


## The map answers one question; the inspection needs another [#the-map-answers-one-question-the-inspection-needs-another]

A utility finds an area whose vegetation representation changed between years. It is tempting to turn that result straight into a clearance warning. But a changed area does not, by itself, establish which tree is near which conductor, how large the gap is, or whether the observation reflects conditions today.

<TeachingIllustration src="/images/memory/geospatial-space-v1.webp" title="Geospatial data: find similar places or change" alt="Earth patches containing forest, water, fields, and mining ground map into groups by learned surface characteristics rather than physical distance. Yearly representations can be compared, while coordinates, observation intervals, and source observations remain attached." caption="Nearby in the representation does not mean nearby on Earth. Annual patterns do not identify an exact event date." />

Use broad representations to guide investigation, then obtain the resolution, geometry, and freshness required by the operational decision. The value comes from narrowing the search while keeping the limits of the observation visible.

## A representation can describe more than a photograph [#a-representation-can-describe-more-than-a-photograph]

Google's annual Satellite Embedding dataset describes each 10-meter pixel using 64 numbers that encode surface conditions over a year. Its documentation supports comparisons between years. That is useful evidence for exploring broad surface change; it does not supply the exact day an event began.

## Keep the time interval explicit [#keep-the-time-interval-explicit]

Store the location, start and end of the represented interval, dataset version, and original observation references. Learned seasonal behavior and an exact date answer different questions.

## Match resolution to the decision [#match-resolution-to-the-decision]

A representation of a region can help identify an area worth examining. A particular tree's clearance from a conductor requires evidence with suitable spatial detail, geometry, and freshness. Do not treat annual regional change as a real-time clearance measurement.

## Combine with the asset map [#combine-with-the-asset-map]

Join candidate findings to verified network assets and previous inspections. Check coordinate systems and observation quality. Then prioritize a suitable inspection or request better imagery.

## Evaluate against field evidence [#evaluate-against-field-evidence]

Measure how often candidate locations become confirmed findings, what is missed, and whether prioritization improves the existing process. See [vegetation inspection](/use-cases/vegetation-inspection).

## External sources [#external-sources]

* [Google Satellite Embedding dataset documentation](https://developers.google.com/earth-engine/datasets/catalog/GOOGLE_SATELLITE_EMBEDDING_V1_ANNUAL). Accessed 17 September 2026.

**Continue the story:** [Represent how equipment behaves over time](/memory/time-series).