<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Krishna Karra on Medium]]></title>
        <description><![CDATA[Stories by Krishna Karra on Medium]]></description>
        <link>https://medium.com/@krishna.karra?source=rss-254f3c1f9978------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*zhapy55PJiavosHz</url>
            <title>Stories by Krishna Karra on Medium</title>
            <link>https://medium.com/@krishna.karra?source=rss-254f3c1f9978------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 03:13:05 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@krishna.karra/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[How We Built a Machine Learning Model to Locate and Map Cranberry Bogs for National Geographic]]></title>
            <link>https://medium.com/descarteslabs-team/how-we-built-a-machine-learning-model-to-locate-and-map-cranberry-bogs-for-national-geographic-4a85686c456b?source=rss-254f3c1f9978------2</link>
            <guid isPermaLink="false">https://medium.com/p/4a85686c456b</guid>
            <category><![CDATA[geography]]></category>
            <category><![CDATA[remote-sensing]]></category>
            <category><![CDATA[thanksgiving]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[algorithms]]></category>
            <dc:creator><![CDATA[Krishna Karra]]></dc:creator>
            <pubDate>Thu, 28 Nov 2019 11:01:01 GMT</pubDate>
            <atom:updated>2019-11-28T11:01:01.752Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*sW8NEIatGy-sFDI_w-Q2Rw.jpeg" /><figcaption><em>We delineated cranberry bogs across the United States by building a machine learning classifier using synthetic aperture radar (SAR) data from the European Space Agency’s Sentinel-1 satellite. This raw probability map over Massachusetts (with a cranberry colormap!) reveals the most likely locations of cranberry bogs (shown in bright white) as well as signal artifacts in radar data, scene boundaries and very bog-like coastal wetlands.</em></figcaption></figure><p><a href="https://www.nationalgeographic.com/history/2019/11/united-states-cranberry-harvest-explained-charts/">For a fun piece</a> to chew on during the Thanksgiving holiday, <em>National Geographic</em> asked <a href="https://descarteslabs.com/">Descartes Labs</a> if we could figure out a way to map every cranberry bog in the United States. We had less than a week to complete the project end-to-end, but we were fortunate to have some pre-existing algorithmic work and data products to consult, which gave us a huge head start.</p><p>In the end, the project proved to be a fun and unique problem to solve from a remote sensing perspective. Here, we’ll unpack how we actually did it!</p><h3><strong>Why was this project hard?</strong></h3><p>While it sounds simple, it turns out that identifying cranberry bogs from satellite imagery is actually quite tricky. Depending on the time of year, they might look like ponds, bare earth, or lush, green agriculture (like corn)!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*T6l0Qu3BPYMfTCI5l5yrhg.png" /><figcaption><em>The life cycle of a cranberry bog is best observed over time. In the four images above, which depict bogs in Massachusetts, we see snow, healthy green vegetation, flooded bogs, and even flashes of red during harvest!</em></figcaption></figure><p>Unless they’re <a href="https://www.cranberries.org/how-cranberries-grow/dry-harvest">dry harvested</a>, most cranberry bogs in North America are manually flooded before harvest. Unfortunately, the timing of the flooding has enough variability from region to region that we couldn’t get away with simply selecting a single moment in time that would allow us to identify the unique optical imagery signature that indicates a flooded bog.</p><p>However, what we <em>could </em>do was take advantage of the manual, pre-harvest floods that inundate these bogs by analyzing their signature in synthetic aperture radar (SAR) imagery, which is particularly sensitive to water content.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/553/1*h85XOEatzqXOa8IDxx8nzQ.png" /><figcaption><em>The difference between vegetated and flooded bogs (top to bottom) is evident in optical imagery, shown on the left. To the right, this difference is also evident in synthetic aperture radar imagery, with a sharp decline in backscatter once the bogs are flooded.</em></figcaption></figure><p>After analyzing the flood signature in radar imagery, we leveraged techniques we’d used in a previous Descartes Labs project that involved using radar data to map rice paddies — which are also periodically flooded — in Southeast Asia, and combined those with information available to us through the Descartes Labs platform that provides an annual composite of Sentinel-1 radar data, which — critically — includes annual statistics computed from raw amplitude backscatter data. These changes in backscatter represent multiple states in the lifecycle of a cranberry bog. Without this groundwork in place, we most likely would not have been able to accommodate the very short turn-around time for this project.</p><h3><strong>How did we do it?</strong></h3><p>Since temporally aggregated radar statistics are a somewhat abstract way to represent bogs, we applied machine learning to these data sets to find a decision surface that accurately separates the positive (cranberry bog) and negative (not a cranberry bog) class. For the positive class, we were able to locate several active cranberry bogs by scraping data from the GIS departments of Wisconsin, New Jersey and Massachusetts, which, as the top producers of cranberries in the U.S., thankfully make this information freely available.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gjVf8-jivYluieKxFhU-aA.png" /><figcaption><em>We sampled locations of cranberry bogs obtained from state GIS data from WI, NJ and MA (left to right). The data, while incomplete, is an excellent proxy for locations that are likely to be active cranberry bogs, and accurate enough to build a machine learning classifier.</em></figcaption></figure><p>Picking samples for the negative class ended up being deceptively challenging because sampling from the most ideal regions also risked confusing the classifier. However, sampling a representative set of locations for the negative class is quite important, particularly when dealing with geospatial data, because the machine learning algorithm will otherwise not generalize very effectively. Put another way, teaching the algorithm what <em>isn’t</em> a cranberry bog is just as important as teaching it what <em>is</em> a cranberry bog.</p><p>We ended up sampling seven classes from the <a href="https://www.mrlc.gov/national-land-cover-database-nlcd-2016">National Landcover Database</a> (NLCD), ranging from urban areas, to cropland, to wetlands. Finally, we were careful to also sample this data in states with <em>no </em>cranberry bogs to ensure we didn’t incorrectly sample bogs in the negative class.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8CoEcTs08Lpo-hjduBs-PA.jpeg" /><figcaption><em>We sampled negative examples from seven NLCD classes across multiple states to create a diverse dataset (from left to right: grasslands, wetlands, dense urban areas, medium-dense urban areas, agriculture, evergreen forest, and open water).</em></figcaption></figure><p>Next, we used our Sentinel-1 annual composite to retrieve the temporal statistics (e.g., min, max, mean, standard deviation) for each location. These statistics were formed into a feature vector on a per-sample basis, and fed into a <a href="https://en.wikipedia.org/wiki/Random_forest">random forest</a> machine learning classifier. Once trained, given an arbitrary input image of SAR statistics, the model computed the probability of each pixel belonging to a cranberry bog. We leveraged the Descartes Labs platform cloud infrastructure to run this model across all cranberry-producing states and Canadian provinces in a matter of hours.</p><h3><strong>How well did we do?</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*kX-cKYQrnEa498VN-19h_Q.gif" /><figcaption><em>This raw detector output is a probability image in which each pixel represents the likelihood that it belongs to a cranberry bog. Since the raw output is so noisy, we’ve animated it in a way that shows a progressive reduction of probability ranges starting with 0–100% (very noisy), and ending with 100% (clean, crisp edges). We then fade the 100% probability map into an aerial image for comparison.</em></figcaption></figure><p>The detector performs quite well! After manually inspecting the probability output, we determined that the detector consistently located every cranberry bog that can be identified by the human eye.</p><p>The following set of visuals show a few more interesting discoveries that we made while analyzing the model’s output.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wP5HuQOIEQalQzsk7ZrZ6w.png" /><figcaption><em>The detector output (shown in white) strongly corresponds to actual cranberry bogs. It doesn’t show bogs that are no longer active and/or incorrectly labeled. It also finds bogs that were missing from the training data. Cranberry bog polygons, shown in yellow, were obtained from GIS data from the state of Massachusetts.</em></figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*F_pYbEzS70VWHx7x_9D64A.png" /><figcaption><em>We ran the detector over Connecticut and found the </em><a href="http://www.newenglandhistoricalsociety.com/six-cranberry-bogs-throughout-new-england/"><em>last active cranberry bog in the state!</em></a></figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*byAoLc5FGisaj3LeJKKnGQ.png" /><figcaption><em>Old, overgrown bogs like this one in Wisconsin do not appear to register in the model’s output.</em></figcaption></figure><iframe src="https://cdn.embedly.com/widgets/media.html?url=https%3A%2F%2Fjsfiddle.net%2F97mkwz2h%2Fembedded%2Fresult&amp;src=https%3A%2F%2Fjsfiddle.net%2F97mkwz2h%2Fembedded%2Fresult&amp;type=text%2Fhtml&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;schema=jsfiddle" width="600" height="400" frameborder="0" scrolling="no"><a href="https://medium.com/media/2a52f37c454d2435712fe0d2cf8f6a82/href">https://medium.com/media/2a52f37c454d2435712fe0d2cf8f6a82/href</a></iframe><figure><img alt="" src="https://cdn-images-1.medium.com/max/620/1*JC1orbDxp9-RE-CjD5iuFQ.png" /><figcaption><em>A “peak color” composite of cranberry bogs in New Jersey over a few harvest seasons reveals stunning color visible from space.</em></figcaption></figure><blockquote>We’re grateful to all the hardworking cranberry farmers in North America, to <em>National Geographic </em>for challenging us with this project, and to our colleagues at Descartes Labs who built the platform that allowed us to rise to the occasion. Who knew that rapid access to petabytes of analysis-ready data and integration, coupled with a vast array of open-source Python packages, would allow us to visualize a holiday staple in a whole new way?</blockquote><blockquote>If you’d like to learn more about the Descartes Labs platform and how it could help your work, please contact us <a href="https://www.descarteslabs.com/contact.html">here</a>.</blockquote><blockquote>Happy Thanksgiving!</blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4a85686c456b" width="1" height="1" alt=""><hr><p><a href="https://medium.com/descarteslabs-team/how-we-built-a-machine-learning-model-to-locate-and-map-cranberry-bogs-for-national-geographic-4a85686c456b">How We Built a Machine Learning Model to Locate and Map Cranberry Bogs for National Geographic</a> was originally published in <a href="https://medium.com/descarteslabs-team">descarteslabs-meditations</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Texas Reigns Supreme in U.S. Energy Production (and McCamey Shows Us How)]]></title>
            <link>https://medium.com/descarteslabs-team/texas-reigns-supreme-in-u-s-energy-production-and-mccamey-shows-us-how-72fac43b599e?source=rss-254f3c1f9978------2</link>
            <guid isPermaLink="false">https://medium.com/p/72fac43b599e</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[oil-and-gas]]></category>
            <category><![CDATA[renewable-energy]]></category>
            <category><![CDATA[energy]]></category>
            <category><![CDATA[science]]></category>
            <dc:creator><![CDATA[Krishna Karra]]></dc:creator>
            <pubDate>Wed, 01 May 2019 16:43:30 GMT</pubDate>
            <atom:updated>2019-05-01T17:57:32.918Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*4bhuADmhaT4HYghluGLQ6w.gif" /><figcaption>This eight-mile-across image of McCamey, Texas sits in a part of the state with a seemingly endless network of oil and gas infrastructure, but it also features the giant Upton 2 solar farm (which powers <a href="https://www.mrt.com/business/oil/article/McCamey-Texas-largest-solar-project-is-a-12512737.php">56,700 residences</a>) and part of the King Mountain Wind Energy Center (which powers <a href="http://www.nexteraenergyresources.com/content/where/portfolio/pdf/kingmountain%20.pdf">86,100 residences</a>).</figcaption></figure><p>If Texas leads the nation in energy production (and it does, by <a href="https://www.eia.gov/state/rankings/#/series/101">quite a lot</a>), McCamey could be its microcosm. This community of roughly 2,000 people is located at the cross section of three enormous energy sources:</p><p><strong>The </strong><a href="https://en.wikipedia.org/wiki/Permian_Basin_(North_America)"><strong>Permian Basin</strong></a>, a massive oil field experiencing its biggest <a href="http://time.com/5492648/permian-oil-boom-west-texas/">boom</a> in decades.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Mr63lqkGprLqvgXoAxMAzA.jpeg" /><figcaption>Approximate area of the Permian Basin.</figcaption></figure><p><strong>The </strong><a href="https://en.wikipedia.org/wiki/Great_Plains"><strong>Great Plains</strong></a>, an <a href="https://www.nrel.gov/gis/images/100m_wind/awstwspd100onoff3-1.jpg">intensely windy</a> region of the U.S.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wDaqMkG_hwXtzDF6WlbqCw.jpeg" /><figcaption>Apart from blustery mountaintops nationwide, the <a href="https://en.wikipedia.org/wiki/Great_Plains">Great Plains</a> is the windiest contiguous region of the continental United States. The brightest blues indicate areas with highest winds. | Source: <a href="https://globalwindatlas.info/">Global Wind Atlas</a></figcaption></figure><p><strong>The </strong><a href="https://en.wikipedia.org/wiki/Sun_Belt"><strong>Sun Belt</strong></a>, the sunniest region of the U.S. (and West Texas is in a really <a href="https://www.nrel.gov/gis/images/map_csp_us_10km_annual_feb2009.jpg">sunny part</a> of it!).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ja6wb8uaHCMuw2EO_0Q0YA.jpeg" /><figcaption>The Southwest, including West Texas, is the sunniest region of the continental United States. The brightest yellows indicate areas with the most usable sunlight. | Source: <a href="https://maps.nrel.gov/nsrdb-viewer/?aL=UdPEX9%255Bv%255D%3Dt%26f69KzE%255Bv%255D%3Dt%26f69KzE%255Bd%255D%3D1&amp;bL=clight&amp;cE=0&amp;lR=0&amp;mC=37.24782120155428%2C-95.1416015625&amp;zL=5">National Renewable Energy Laboratory</a></figcaption></figure><p>The image of McCamey, TX at the top shows a patch of earth being tapped for every last ounce of energy it has to offer. A new, several-square-mile solar farm that powers more than 56,000 residences — and is <a href="https://www.mrt.com/business/oil/article/McCamey-Texas-largest-solar-project-is-a-12512737.php">still under construction</a> — is ringed by dozens of wind turbines (powering an additional 86,000+ homes) strategically placed on ridges nearby in a landscape strewn with oil and gas infrastructure. The area is overwhelmingly industrial and remote — Lubbock, the next closest city, is nearly 200 miles away.</p><p>Leveraging information both visible and invisible to the naked eye, <a href="https://descarteslabs.com/">Descartes Labs</a> is using an array of earth observation satellites to track energy infrastructure development. Since the U.S. as a whole relies so heavily on Texas energy, developments in the state’s infrastructure have an impact on millions of people. In McCamey and throughout the state (particularly the western half), wind, solar, and oil and gas together make up the story of modern Texas energy production.</p><p>We mapped these three energy sources by applying deep learning computer vision models to satellite imagery. These models, trained with a diverse set of satellite image data, perform sophisticated pattern matching using neural networks to detect and delineate energy infrastructure. Using the Descartes Labs platform, which natively enables scalable cloud computing, we can train, iterate on, validate and deploy these models quickly and at a massive scale.</p><p>As an example, our oil and gas model can be run over Texas, a state approximately the size of Spain, from <a href="https://www.fsa.usda.gov/programs-and-services/aerial-photography/imagery-programs/naip-imagery/index">National Agricultural Imagery Program</a> (NAIP) data at 1 meter resolution, in a matter of hours. What follows is a rundown of the culmination of multiple projects at Descartes Labs. Individually, each reveals the state’s success in a particular energy source, but together they illuminate a truly enormous and diverse energy powerhouse.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QhP4cYeIunvLP3JhfFd_ww.jpeg" /><figcaption>Image tiles of wind turbine detections in Texas. Each image is about as wide as the <a href="https://en.wikipedia.org/wiki/Washington_Monument">Washington Monument</a> is tall. | Source: NAIP 2016</figcaption></figure><h3>Wind</h3><p>Wind power is booming in Texas. It accounts for about <a href="https://windexchange.energy.gov/states/tx">15%</a> of the electric grid; 12,000-plus turbines across the state produce <a href="https://www.eia.gov/electricity/monthly/epm_table_grapher.php?t=epmt_1_14_b">more wind energy than any other state</a> and since <a href="https://www.eia.gov/state/?sid=TX">2014, wind energy</a> has outstripped output from the state’s <a href="https://www.houstonchronicle.com/business/article/Nuclear-power-woes-extend-to-Texas-13709604.php">two nuclear power plants</a>. Only four countries create more wind energy than the state of Texas, and the development of new wind farms there isn’t slowing. Texas has more wind energy projects <a href="https://www.awea.org/resources/fact-sheets/state-facts-sheets">under construction</a> than any other state.</p><p>Many of the wind turbines in Texas are concentrated in the northern and western parts of the state and on ridges or other areas of locally high elevation. But there are other sizeable wind farms scattered throughout the state, including the Los Vientos wind farm near McAllen and the Javelina wind farm near Loredo. As you can see in the image tiles above, wind turbines have been installed in a variety of landscapes. Wherever strong winds intersect with the power grid, there’s a chance a turbine will pop up.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zwARc61u4LxveCjUyCOEFQ.jpeg" /><figcaption>Image tiles of utility-scale solar farms in Texas. Each image is about a mile across. | Source: © Airbus DS 2016–2018</figcaption></figure><h3>Solar</h3><p>Unlike wind and oil and gas, which are comparatively better-established energy sources in Texas, solar development is more recent. Even so, while solar power currently accounts for less than 1% of total energy production in Texas, the state ranked <a href="https://txses.org/solar-energy-data-2018/">#1 in solar energy potential and #7 in installed solar capacity</a> in 2018. Energy production from solar has increased by <a href="https://www.chron.com/business/energy/article/Texas-solar-growth-only-behind-California-report-13685983.php#photo-15188650">more than five times</a> since only four years ago, and is forecasted to <a href="https://www.bloomberg.com/news/articles/2019-04-04/solar-is-going-to-keep-oil-flowing-in-the-texas-shale-patch">quadruple by 2021</a>.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fplayer.vimeo.com%2Fvideo%2F333468523%3Fapp_id%3D122963&amp;dntp=1&amp;url=https%3A%2F%2Fvimeo.com%2F333468523&amp;image=https%3A%2F%2Fi.vimeocdn.com%2Fvideo%2F779461345_640.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=vimeo" width="640" height="360" frameborder="0" scrolling="no"><a href="https://medium.com/media/5bf35e24fb866541441d98c6489a4466/href">https://medium.com/media/5bf35e24fb866541441d98c6489a4466/href</a></iframe><p>In fact, researchers at Rice University found that <a href="https://www.vox.com/science-and-health/2019/1/7/18167351/texas-energy-solar-wind-renewables-ercot">patterns of wind and sunshine complement each other quite well in Texas</a>. Wind energy there peaks at night, while the sun, of course, shines brightest during the day. This phenomenon is important because energy integration across multiple sources in Texas has historically been a challenge. As the state’s<a href="https://blogs.scientificamerican.com/plugged-in/texas-got-18-percent-of-its-energy-from-wind-and-solar-last-year/"> electric grid is largely isolated from others</a>, nearly all transmission and load balancing must happen in-state. However, <a href="https://www.eia.gov/state/?sid=TX">Texas leads the nation in energy consumption as well as production</a>, and that demand is bolstering continued rapid growth in solar.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Z4lkoo8J9fVF07OnpobI5Q.jpeg" /><figcaption>Image tiles of oil and gas well pad detections. Each image is about as far across as the <a href="https://en.wikipedia.org/wiki/Joey_Meyer_(baseball)">longest home run</a> in baseball history. | Source: NAIP 2016</figcaption></figure><h3>Oil and Gas</h3><p>In 2017, Texas produced <a href="https://www.eia.gov/state/rankings/?sid=US#/series/47">far more</a> natural gas than any other state. In 2018, Texas set a <a href="https://www.forbes.com/sites/davidblackmon/2019/02/13/new-report-details-record-texas-oil-boom-in-2018/#7ed0e9767012">new record</a> for oil production, and more than doubled the amount produced in North Dakota, the next most productive state.</p><p>The Permian Basin, which is located largely in West Texas and may become the <a href="https://www.forbes.com/sites/rrapier/2018/12/27/why-the-permian-basin-may-become-the-worlds-most-productive-oil-field/#4ce1fa1b5ccb">world’s most prolific oil field</a>, rightfully gets a lot of attention when it comes to Texas oil and gas production. To some, the Permian might look like an industrial and fiery wasteland, with nothing in sight but oil and gas well pads (and infrastructure to support their operations) for hundreds of miles. But to landowners who are reaping rewards from this region, the infrastructure indicates booming industry.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2hXxiG0wIUc3W90d2AAY_A.jpeg" /><figcaption><a href="https://go.nasa.gov/2DEU5uM">VIIRS night-time lights</a> (white) and <a href="https://earthdata.nasa.gov/earth-observation-data/near-real-time/firms">FIRMS fire detections</a> (yellow) on April 28, 2019 | Source: NASA</figcaption></figure><p>One big reason why Texas, and the Permian Basin in particular, has become an oil and gas juggernaut on the world stage is because of<a href="https://en.wikipedia.org/wiki/Hydraulic_fracturing"> hydraulic fracturing</a>, commonly called fracking.</p><p>The life cycle of a frack site is quite rapid once construction begins. Crews arrive on site bringing water tanks, sand kings, drilling equipment, fracturing pumps and more. Drilling (first vertically, then usually horizontally) and hydraulic fracturing (when<a href="https://www.fractracker.org/resources/oil-and-gas-101/fracking-fluid/"> fracking fluid</a> is injected into the ground) occur quickly, usually within weeks. Once the well has been stimulated, hydrochloric acid is pumped into the formations to facilitate the flow of fluid and gas. Finally, gas processing units and pipelines are installed, and the well pad comes online and begins operating. Eventually, the pipelines run empty, the well pad is abandoned, but the inoperative site remains a part of the vast, visible, and ever-changing Texan energyscape.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LHu1CktPWiSOJ5AuapLXNw.gif" /><figcaption>Monitoring fracking activity in the Delaware Basin from 2017–2019. Left: <a href="https://sentinel.esa.int/web/sentinel/missions/sentinel-2">Sentinel-2</a> true color (Optical) | Center: <a href="https://sentinel.esa.int/web/sentinel/missions/sentinel-2">Sentinel-2</a> infrared (Flares) | Right: <a href="https://sentinel.esa.int/web/sentinel/missions/sentinel-1">Sentinel-1</a> radar</figcaption></figure><p>We can monitor many stages of the life cycle of oil and gas well pads by fusing information from optical information (construction, storage tanks, equipment), infrared (flaring), and radar data (metallic objects). Each data source can provide a unique look at a well pad, but when merged together, they reveal a more nuanced view of the well pad’s life span and production.</p><p>Descartes Labs is monitoring all of the energy production in Texas (and around the world). For the moment, we see that oil and gas still reigns supreme in the state (just have a look below!), but wind and solar are coming up fast. It will be interesting to monitor how the landscapes of traditional oil and gas towns evolve, as McCamey has, as wind and solar continue their rapid growth in the region.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/750/1*Gg-HyEHMkHSBtQxJfvuMZQ.gif" /><figcaption>Descartes Labs’s detections of wind turbines (<strong>tens of thousands</strong>), utility-scale solar farms (<strong>tens</strong>) and oil and gas well pads (<strong>over one hundred thousand</strong>).</figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=72fac43b599e" width="1" height="1" alt=""><hr><p><a href="https://medium.com/descarteslabs-team/texas-reigns-supreme-in-u-s-energy-production-and-mccamey-shows-us-how-72fac43b599e">Texas Reigns Supreme in U.S. Energy Production (and McCamey Shows Us How)</a> was originally published in <a href="https://medium.com/descarteslabs-team">descarteslabs-meditations</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>