<?xml version="1.0" encoding="UTF-8" ?><!-- generator=Zoho Sites --><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><atom:link href="https://www.apaceo.com/blogs/author/angela-manchester/feed" rel="self" type="application/rss+xml"/><title>APACEO - Blog by Angela Manchester</title><description>APACEO - Blog by Angela Manchester</description><link>https://www.apaceo.com/blogs/author/angela-manchester</link><lastBuildDate>Fri, 24 Jul 2026 05:45:17 +1000</lastBuildDate><generator>http://zoho.com/sites/</generator><item><title><![CDATA[Earth Observation Just Turned 54]]></title><link>https://www.apaceo.com/blogs/post/earth-observation-just-turned-54-what-it-means-to-work-in-an-industry-still-finding-its-feet</link><description><![CDATA[<img align="left" hspace="5" src="https://www.apaceo.com/landsat-1.png"/>There is something strange about working in a field this young. Medicine, teaching, engineering, the trades, these all feel like they have simply alwa ]]></description><content:encoded><![CDATA[<div class="zpcontent-container blogpost-container "><div data-element-id="elm_GWcMP8L-TiCkGBfotNPDaw" data-element-type="section" class="zpsection "><style type="text/css"></style><div class="zpcontainer-fluid zpcontainer"><div data-element-id="elm__-_6sc2HT9e-Tsps1HE9Ew" data-element-type="row" class="zprow zprow-container zpalign-items- zpjustify-content- " data-equal-column=""><style type="text/css"></style><div data-element-id="elm_HWXdf8BCT7GrQ7wTCMT1ag" data-element-type="column" class="zpelem-col zpcol-12 zpcol-md-12 zpcol-sm-12 zpalign-self- "><style type="text/css"></style><div data-element-id="elm_xnKs3Um1RASLBFzpqi7GNA" data-element-type="heading" class="zpelement zpelem-heading "><style></style><h2
 class="zpheading zpheading-align-center zpheading-align-mobile-center zpheading-align-tablet-center " data-editor="true">There is something strange about working in a field this young</h2></div>
<div data-element-id="elm_xReoUWfDTd-Z3drk23GQvQ" data-element-type="text" class="zpelement zpelem-text "><style></style><div class="zptext zptext-align-center zptext-align-mobile-center zptext-align-tablet-center " data-editor="true"><p></p><div><p>There is something strange about working in a field this young. Medicine, teaching, engineering, the trades, these all feel like they have simply always existed, stretching back centuries, their foundations so old they seem carved into the ground itself. Earth Observation does not have that luxury. It has a birthday. It has a rocket launch on record. It has a single point in time before which nobody on the planet had ever seen their own world this way.</p><p><br/></p><p>That day was July 23, 1972, and as of today the industry has just turned 54.</p><p><br/></p><p>That youth creates a strange kind of double existence. The industry is old enough to have a lineage, eight satellites deep, with archives stretching back nearly 50 years and methods that have been tested and refined across decades. Yet it is also young enough that the tools practitioners learned on ten years ago can feel almost unrecognisable today. Machine learning, cloud computing, and open data platforms have reshaped the field so quickly that a career spanning just fifteen or twenty years can span multiple entirely different eras of the same profession.</p><p>Most industries get to grow up slowly, generation by generation, absorbing change gradually enough that the ground never feels like it is shifting beneath the people standing on it. Earth Observation does not get that gift. It is simultaneously establishing its foundations and rebuilding them in real time, which means the people working in it are not just practitioners of a craft, they are also its historians, its first generation, and its architects, often all at once.</p><p>To understand how that all started, it helps to go back to the very beginning.</p><p><br/></p><h3>The Satellite That Started It All</h3><p>On July 23, 1972, a Delta 900 rocket lifted off from Vandenberg Air Force Base in California, carrying a satellite that had no idea it was about to become a legend. It was called ERTS-1, the Earth Resources Technology Satellite, a name only an engineer could love. NASA would later rename it something that stuck: Landsat 1.</p><p>Nobody watching that launch could have known they were witnessing the first heartbeat of an industry, Earth Observation, one that would still be alive and expanding 54 years later.</p><p><br/></p><h3>A Camera Built to Fail, and a Backup Built to Save It</h3><p>Landsat 1 carried two instruments into orbit. The star of the show was supposed to be the Return Beam Vidicon, a camera system built on analogue TV tube technology that had already proven itself on weather satellites. Riding along as an afterthought was something experimental: the Multispectral Scanner System, or MSS.</p><p>Within days of reaching orbit, a defective power switching circuit caused surges that knocked the Return Beam Vidicon offline for good. The mission's primary instrument was dead before it had really begun.</p><p>It was the backup that saved everything. The MSS, the scrappy secondary system nobody had fully trusted, turned out to be superior in almost every way. It became the beating heart of the mission, and within days it proved its worth by capturing imagery of an 81,000 acre wildfire burning in the isolated wilderness of central Alaska. A view of Earth that had simply never existed before was suddenly possible.</p><p>Landsat 1 was built to last one year. It lasted six, quietly orbiting the planet every 18 days and eventually imaging about 75 percent of Earth's entire surface before it was finally decommissioned on January 6, 1978.</p><p><br/></p><h3>Was It Truly the First?</h3><p>Not exactly, and that nuance is part of what makes the story interesting. Weather satellites had already been watching Earth from space since 1960. What set Landsat 1 apart was purpose. It was the first Earth observing satellite launched with the explicit intent of studying and monitoring the planet's landmasses, rather than tracking storm systems. It became the first Earth observing satellite explicitly designed to study planet Earth itself, and in doing so it launched an entirely new discipline.</p><p>Its founding mission was to gather facts about the natural resources of the Earth using satellites carrying sophisticated remote sensing instruments. That single sentence, written more than five decades ago, has quietly held true through eight satellites and counting.</p><p><br/></p><h3>A Legacy That Just Kept Growing</h3><p>Since that July morning in 1972, the Landsat program has continuously acquired images of Earth's land surface, building an unbroken visual record of a changing planet that exists nowhere else in human history. The archive it left behind spans nearly 50 years and is freely available to anyone in the world.</p><p><br/></p><p>There is something quietly poetic in how it all began. The instrument nobody expected much from ended up carrying the entire future of the field. A whole industry now exists because a backup camera worked better than the one it was meant to support.</p><p>Every satellite that has ever looked down and measured a forest, tracked a coastline, or mapped a drought traces its lineage back to that one Delta rocket and its stubborn, unglamorous scanner. Fifty four years on, the story is still being written, one orbit at a time.</p><p><br/></p><p>#EarthObservation #RemoteSensing #GIS #SatelliteImagery #DisasterResponse #GeospatialData #SpaceForGood</p><p><img src="/landsat-1.png"/></p></div><p></p></div>
</div><div data-element-id="elm_RcMS8vK4Q6mXGte_9vDowQ" data-element-type="button" class="zpelement zpelem-button "><style></style><div class="zpbutton-container zpbutton-align-center zpbutton-align-mobile-center zpbutton-align-tablet-center"><style type="text/css"></style><a class="zpbutton-wrapper zpbutton zpbutton-type-primary zpbutton-size-md " href="javascript:;" target="_blank"><span class="zpbutton-content">Get Started Now</span></a></div>
</div></div></div></div></div></div> ]]></content:encoded><pubDate>Thu, 23 Jul 2026 09:40:41 +1200</pubDate></item><item><title><![CDATA[The Defensibility Test]]></title><link>https://www.apaceo.com/blogs/post/the-defensibility-test</link><description><![CDATA[<img align="left" hspace="5" src="https://www.apaceo.com/images/ga99617092e0cd84db29ec3b4e60635951415fc014ada1ac9cb229a9244893774a8914b702751989f29b0342aae9b436b6f4214df92bec69d7ead556408c36a28_1280.jpg"/>AI can turn a folder of satellite imagery into a finished map in seconds. The real question is whether you could defend that answer. A four-question test for checking any AI-generated EO output, right data, right assumptions, traceable evidence, known failure point, before you put your name on it.]]></description><content:encoded><![CDATA[<div class="zpcontent-container blogpost-container "><div data-element-id="elm_0Nea-NDHQIO7SBDve1bf-g" data-element-type="section" class="zpsection "><style type="text/css"></style><div class="zpcontainer-fluid zpcontainer"><div data-element-id="elm_YKyN6jmNS0qzKaPJzJkcwg" data-element-type="row" class="zprow zprow-container zpalign-items- zpjustify-content- " data-equal-column=""><style type="text/css"></style><div data-element-id="elm_gCG4EF-oS26KQ4ZBjkbPyg" data-element-type="column" class="zpelem-col zpcol-12 zpcol-md-12 zpcol-sm-12 zpalign-self- "><style type="text/css"></style><div data-element-id="elm_17Ipvr_bSj68l02rS3hh2g" data-element-type="heading" class="zpelement zpelem-heading "><style></style><h2
 class="zpheading zpheading-align-center zpheading-align-mobile-center zpheading-align-tablet-center " data-editor="true"><span>How to Trust an AI-Generated EO Output</span></h2></div>
<div data-element-id="elm_1bxxOKybRySKKxFzTUAQTg" data-element-type="text" class="zpelement zpelem-text "><style></style><div class="zptext zptext-align-center zptext-align-mobile-center zptext-align-tablet-center " data-editor="true"><p><span><span>AI tools that touch Earth Observation data are everywhere now.&nbsp;</span></span></p><p><span><span>Feed a folder of imagery in, ask a plain-language question (&quot;show me the areas that need attention&quot;),&nbsp;</span></span></p><p><span><span>and get a finished map or answer back in seconds.&nbsp;</span></span></p><p><span><span><br/></span></span></p><p><span><span>The speed is real. What is less settled is whether the answer is right, and whether you could stand behind it if someone pushed back.</span></span></p></div>
</div><div data-element-id="elm_ydLBAs8MBXRWxh4DxOMfow" data-element-type="text" class="zpelement zpelem-text "><style></style><div class="zptext zptext-align-left zptext-align-mobile-left zptext-align-tablet-left " data-editor="true"><p></p><div><p style="margin-bottom:16px;">That is the actual bottleneck right now, and it is worth naming clearly: the limit is not access to data (there is more free EO data available than most people use) and it is not really interpretation either (AI is often decent at that). The limit is trust. Specifically, whether a professional can defend the output: which data it used, which assumptions it made, and whether those choices were the right ones for this question.</p><p style="margin-bottom:16px;">This matters more in EO than in a lot of other domains, because the outputs feed real decisions: where to send a maintenance crew, whether a flood claim is valid, whether a piece of land has changed use. An answer that looks clean on screen can still be built on the wrong buffer distance, the wrong cloud mask, or an assumption that quietly does not hold for your area.</p></div><p></p></div>
</div><div data-element-id="elm_ooD0w6yj1GgZFCl9ebkkrA" data-element-type="heading" class="zpelement zpelem-heading "><style></style><h2
 class="zpheading zpheading-style-none zpheading-align-left zpheading-align-mobile-left zpheading-align-tablet-left " data-editor="true"><span>The four-question defensibility test</span></h2></div>
<div data-element-id="elm_QkljW-UYg4U0115LWmtcyA" data-element-type="text" class="zpelement zpelem-text "><style></style><div class="zptext zptext-align-left zptext-align-mobile-left zptext-align-tablet-left " data-editor="true"><p></p><div><p style="margin-bottom:16px;">Before you present or act on any AI-generated EO output, run it through four questions. This applies regardless of which tool produced it, this is not about one platform over another.</p><p style="margin-bottom:16px;"><strong>1. Which datasets did it actually use, and are they the right vintage and resolution for this question?</strong> An AI tool will happily use whatever imagery is available in the folder or the default source it was pointed at. That is not the same as the right imagery. A land-use answer built on a two-year-old scene, or a flood answer built on 30m resolution when the flood boundary needs 10m, will look confident and still be wrong.</p><p style="margin-bottom:16px;"><strong>2. Which assumptions, buffers, filters, or thresholds did it apply, and would you have chosen them yourself?</strong> Every &quot;show me the areas that need attention&quot; question hides a dozen small decisions: how wide a buffer around infrastructure, what counts as &quot;change,&quot; what threshold separates signal from noise. An AI tool picks something. Your job is to find out what, and check it against how you would have done it.</p><p style="margin-bottom:16px;"><strong>3. Can you point to the specific pixels or features that support the conclusion?</strong> If the answer is &quot;the model said so&quot; and you cannot trace it back to something visible in the data, you cannot defend it in a meeting. You should be able to open the source imagery and show the evidence, not just the output.</p><p style="margin-bottom:16px;"><strong>4. What would change the answer, and have you checked whether that condition holds?</strong> Every EO conclusion is conditional on something: a cloud-free scene, a stable sensor calibration, a correct date range. Name the condition. Then check it, the same way you would check metadata before trusting any other satellite product.</p></div><p></p></div>
</div><div data-element-id="elm_LfaKrjWjpgjg9Yt8wQYZbA" data-element-type="text" class="zpelement zpelem-text "><style></style><div class="zptext zptext-align-left zptext-align-mobile-left zptext-align-tablet-left " data-editor="true"><p></p><div><h2 style="margin-bottom:16px;">Why this is the actual skill</h2><p style="margin-bottom:16px;">None of this is anti-AI. AI in EO workflows is only going to get more common, and used well it saves real time. The skill that matters now is not learning to prompt it better, it is learning to interrogate what it hands back before your name goes on it. That is the same discipline that already applies to any EO output, human or automated: treat every result as something you have to defend, not something you get to trust by default.</p></div><br/><p></p></div>
</div><div data-element-id="elm_N0l8iyTFVCEu0MGYZknMQg" data-element-type="text" class="zpelement zpelem-text "><style></style><div class="zptext zptext-align-left zptext-align-mobile-left zptext-align-tablet-left " data-editor="true"><p></p><div><div><span style="font-weight:bold;">Open question</span></div><div>What is one AI-generated EO output you have seen, or built yourself, that you would not have signed off on without checking the underlying data first?&nbsp;</div><div>What did you actually check?</div><div>This is free, standalone thinking, part of the Earth Observation community's vendor-neutral approach to EO education.</div></div><p></p></div>
</div><div data-element-id="elm_wtqUTod3TBmK1ePLnwfYBg" data-element-type="button" class="zpelement zpelem-button "><style></style><div class="zpbutton-container zpbutton-align-center zpbutton-align-mobile-center zpbutton-align-tablet-center"><style type="text/css"></style><a class="zpbutton-wrapper zpbutton zpbutton-type-primary zpbutton-size-md zpbutton-style-none " href="https://www.skool.com/earth-observation-6970/about" target="_blank"><span class="zpbutton-content">Join the Earth Observation Community</span></a></div>
</div></div></div></div></div></div> ]]></content:encoded><pubDate>Wed, 01 Jul 2026 18:06:15 +1200</pubDate></item></channel></rss>