"}],"short_description":"This blog provides an introduction to location encoders, explains how location embeddings work, and discusses their applications in GIS.","flexible_content":[{"acf_fc_layout":"content","content":"Isolated geographic coordinates (latitude\u00a0and\u00a0longitude) lack descriptive context.\u00a0Unless\u00a0you’re familiar with geographic coordinates, raw numbers like these don’t\u00a0tell you much.\u00a0Is\u00a0this a city, a forest, a desert,\u00a0or\u00a0a mountain? Humans recognize places by their surroundings, not their coordinates.\u00a0A\u00a0satellite image instantly reveals buildings, roads, farmland, snow, water, or forest.\u00a0To make coordinates more meaningful, location embeddings\u00a0represent\u00a0these numbers in a rich description of the surrounding physical environment.<\/p>\n
Before understanding a location encoder, it helps to understand the concept of an embedding.<\/span>\u00a0<\/span>An embedding is a compact numerical representation that captures the important characteristics of something, for example:<\/span>\u00a0<\/span><\/p>\n Similarly,\u00a0a geographic location can be represented the same way: instead of storing only latitude and longitude,\u00a0a\u00a0stored\u00a0learned vector summarizes what makes that place unique. This is a\u00a0<\/span>location embedding<\/span>.<\/span>\u00a0<\/span><\/p>\n A location encoder is a deep learning model that converts coordinates into these embeddings. Rather than treating coordinates as just numbers on a map, it learns the environmental and geographic characteristics tied to\u00a0various\u00a0locations across the globe so places with similar landscapes (forests, farmland, deserts, coastlines, cities, and so on) produce similar embeddings, even across different countries.<\/span>\u00a0<\/span><\/p>\n For example, given only the coordinates 34.0564<\/span>\u00a0<\/span>degrees\u00a0north, 117.1956<\/span>\u00a0<\/span>degrees\u00a0west, a computer has\u00a0no\u00a0context about the physical landscape. A location encoder processes these coordinates into an n-dimensional embedding<\/span>,<\/span>\u00a0something like [0.22, 0.79, 0.41, …, n] that encodes the environmental attributes of that area. Since computers\u00a0can’t\u00a0see\u00a0a map<\/span>\u00a0<\/span>the way a human can, this translation step is what gives raw coordinates structured, usable meaning.<\/span><\/p>\n Esri has developed location encoding models, each designed to capture different aspects of geographic information. These include the\u00a0<\/span>USA Geodemographic embeddings<\/span><\/a>\u00a0created\u00a0using\u00a0<\/span>Geodemographic\u00a0Foundation Model (GDFM)<\/span><\/a>, which is trained on demographic data, and the\u00a0<\/span>Global Location Encoder (Sentinel-2)<\/span><\/a>, which is trained on Sentinel-2 satellite imagery. Because these models are built on different architectures and trained on different data sources, they capture\u00a0different\u00a0representations of geographic locations.<\/span>\u00a0<\/span><\/p>\n The encoder learns a continuous function over geographic space.\u00a0Locations that are environmentally similar but far apart\u2014such as\u00a0two temperate rainforests on opposite sides of\u00a0the world\u2014can end up close together in embedding space, while\u00a0neighbo<\/span>ring\u00a0locations with\u00a0very different\u00a0land cover may end up far apart. Spatial proximity in the real world\u00a0doesn’t\u00a0equal proximity in embedding space; ecological similarity does.<\/span>\u00a0<\/span><\/p>\n By learning the relationship between coordinates and satellite imagery, the location encoder captures the environmental and physical characteristics of a region.\u00a0The Esri\u00a0<\/span>Global Location Encoder (Sentinel-2)<\/span><\/b><\/a>\u00a0model is based on\u00a0<\/span>SatCLIP<\/span><\/b><\/a>\u00a0<\/span>architecture and\u00a0is\u00a0trained on 2024 globally distributed Sentinel-2\u00a0Level-2\u00a0imagery\u00a0using 12\u00a0bands\u00a0of the 13 bands (it does not use\u00a0B10)<\/span>\u00a0<\/span>.\u00a0<\/span>\u00a0<\/span>Learn\u00a0more about<\/span><\/b>\u00a0<\/span><\/b>how to use Global Location Encoder (Sentinel-2) model<\/span><\/b><\/a>.<\/p>\n The workflow typically follows a structured path from raw data to a usable GIS product:<\/span>\u00a0<\/span><\/p>\n Location embeddings support a range of downstream tasks<\/span>\u2014such as\u00a0<\/span>similarity search, clustering, and geospatial data enrichment <\/span>and improve predictive\u00a0<\/span>modelling<\/span>\u00a0tasks like classification and regression.<\/span><\/span>\u00a0<\/span><\/span><\/span>Traditionally,\u00a0<\/span>identifying<\/span> areas with similar geographic characteristics <\/span>whether for mineral exploration, land cover mapping, or environmental monitoring <\/span>has relied on manually engineered indices, threshold-based classification, or time-intensive expert interpretation. The\u00a0<\/span><\/span>Generate Embeddings Using AI Models<\/span><\/span><\/a>\u00a0and\u00a0<\/span><\/span>Find Similar Features Using Embeddings<\/span>\u00a0<\/span>GeoAI<\/span>\u00a0tool<\/span>s<\/span>\u00a0offer a different approach<\/span>.<\/span>\u00a0<\/span>The first uses<\/span>\u00a0deep learning<\/span>\u00a0to derive\u00a0<\/span>location embeddings<\/span>\u00a0that encode the underlying spatial, spectral, and textural characteristics of a landscape,\u00a0<\/span>and the second allows\u00a0<\/span>analysts\u00a0<\/span>to<\/span>\u00a0<\/span>use these location embeddings to\u00a0<\/span>search for areas similar to known examples across a study area, a country, or the entire globe.<\/span><\/span><\/p>\n To identify areas with geographic characteristics similar to known coal mining sites, three representative polygons (mine_query<\/em>, outlined in yellow) were selected as query features. These were run against a raster of pre-computed location embeddings for the full study area using the Find Similar Features Using Embeddings<\/a> tool.<\/p>\n Rather than relying on hand-picked spectral indices or manually engineered features, this approach uses embeddings that encode the underlying spatial and spectral patterns of each location, capturing texture, context, and structure in a way that’s difficult to replicate with traditional band math alone. The tool computes a similarity score between every pixel’s embedding vector and the embeddings of the query polygons, effectively asking: How close is this location, in feature space, to the areas already known to be mines?<\/em><\/p>\n The resulting similarity scores were rendered as a heat map, with the red-to-yellow gradient (sparse to dense) indicating increasing similarity to the query features. To validate the output, high-similarity zones were compared against validation_mines_cils<\/em>, a reference layer of previously mapped mining locations (black points). The strong spatial correspondence between the model’s top similarity clusters and the validated mine sites confirms that the embeddings captured the distinguishing geographic signature of coal mining activity\u2014not just at the three query sites, but across analogous, previously unlabeled areas in the study region.<\/p>\n<\/div>\n"},{"acf_fc_layout":"content","content":" Here, a single polygon located in the high-elevation terrain of the Himalayan range (Query_feature_Himalaya<\/em>, yellow<\/em>) was used as the query feature, run against a candidate set of location embeddings spanning the broader Himalaya\u2013Gangetic transect.<\/p>\n The tool returned the output embeddings feature class (purple points)<\/em> as the candidate locations with the closest embedding vectors in feature space to the query polygon. As expected, the returned matches align closely with the high-relief, snow-covered mountainous terrain running along the Himalayan arc, while candidate points in the flat, cultivated lowlands to the south were not selected. This confirms that the embeddings encoded terrain-level characteristics such as elevation, ruggedness, and land cover typical of high mountain environments, allowing a single representative sample to generalize across an entire mountain range without manually defined rules for slope or elevation thresholds.<\/p>\n<\/div>\n"},{"acf_fc_layout":"content","content":" In this case, two query polygons (Query Features, blue <\/em>) were selected from the flat, agriculturally dominated Gangetic plains, south of the Himalayan foothills. These were compared against a broader set of embedding candidates distributed across both the plains and the mountains. The threshold to get the similar embeddings is set to 0.5.<\/p>\n The resulting output embeddings feature class (red points)<\/em> shows a clear concentration within the plains region, closely tracking the low-lying, densely cultivated and settled landscape characteristic of the Ganges basin, while candidate points in the Himalayan terrain were largely excluded. The embeddings here pick up on the flat topography, land use, and river-plain characteristics that define the Gangetic landscape, again showing that a small number of representative query samples is enough for the model to generalize the target land-cover pattern across a large, geographically diverse study area.<\/p>\n<\/div>\n"},{"acf_fc_layout":"content","content":" This is a regional close-up of the global desert similarity search shown earlier, focused on North and South America. The similiar_features_to_sonoran_desert<\/em> layer uses graduated point symbology across five similarity bins, from low similarity (dark, 0.01\u20130.20<\/em>) to high similarity (bright yellow, 0.78\u20130.98<\/em>).<\/p>\n The highest-similarity points (yellow<\/em>) cluster tightly around the southwestern United States and northern Mexico, precisely the Sonoran Desert region the query feature was drawn from, confirming the model correctly recognizes the query’s own surrounding landscape as most similar to itself. Moving outward, scores fade through orange and brown as the terrain transitions into the Rockies, the Great Plains, and the more humid eastern US and Central America, dropping further (dark points) across the tropical Amazon basin.<\/p>\n Notably, the model performs just as well at picking out arid terrain much farther south: elevated similarity points appear along the Atacama Desert on the Peru\u2013Chile coast and the Patagonian steppe in Argentina, which are both true desert and semi-arid environments, despite being on a different continent from the query feature and separated from it by the Amazon rainforest. This confirms the model is responding to genuine environmental similarity (aridity, sparse vegetation, and surface reflectance) rather than simple geographic proximity to the query point, which is the same behavior that drove the strong global clustering seen in Figure 4, now visible at a finer regional resolution within a single continent-pair view.<\/p>\n<\/div>\n"},{"acf_fc_layout":"content","content":" 3. Clustering<\/p>\n While the previous examples used embeddings for weakly supervised similarity searches, comparing candidate locations against known query feature embeddings can also be used unsupervised, displaying the data’s default groupings.<\/p>\n The workflow starts with the Generate Embeddings Using AI Models<\/a> tool, which produces a location embedding vector for each feature in the study area, encoding its underlying spatial and spectral characteristics. These embeddings are then passed through Extract Embedding to Fields<\/a>,which unpacks each vector into individual numeric attribute fields, converting the embedding into a standard multivariate table that the ArcGIS Pro statistical tools can work with directly.<\/p>\n With the embedding dimensions now available as fields, they’re used as input in the Multivariate Clustering<\/a> tool, which groups features based on statistical similarity across all embedding dimensions simultaneously with no predefined query feature and no training example required. The tool can detect the optimal number of clusters automatically, or the analyst can specify it directly, depending on how much control is needed over the resulting classification.<\/p>\n<\/div>\n"},{"acf_fc_layout":"image","image":{"ID":2976557,"id":2976557,"title":"","filename":"im6-1.png","filesize":783446,"url":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im6-1.png","link":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/geoai\/beyond-latitude-and-longitude-giving-geographic-context-to-coordinates\/im6-2","alt":"Map displaying unsupervised clusters of geographic regions based on location embeddings.","author":"370342","description":"","caption":"Figure 5. Unsupervised clustering of location embeddings using the Multivariate Clustering tool, visualized across Africa, the Middle East, and South Asia.","name":"im6-2","status":"inherit","uploaded_to":2976369,"date":"2026-07-11 23:02:27","modified":"2026-07-11 23:10:17","menu_order":0,"mime_type":"image\/png","type":"image","subtype":"png","icon":"https:\/\/www.esri.com\/arcgis-blog\/wp-includes\/images\/media\/default.png","width":799,"height":525,"sizes":{"thumbnail":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im6-1-213x200.png","thumbnail-width":213,"thumbnail-height":200,"medium":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im6-1.png","medium-width":397,"medium-height":261,"medium_large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im6-1.png","medium_large-width":768,"medium_large-height":505,"large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im6-1.png","large-width":799,"large-height":525,"1536x1536":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im6-1.png","1536x1536-width":799,"1536x1536-height":525,"2048x2048":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im6-1.png","2048x2048-width":799,"2048x2048-height":525,"card_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im6-1-708x465.png","card_image-width":708,"card_image-height":465,"wide_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im6-1.png","wide_image-width":799,"wide_image-height":525}},"image_position":"center","orientation":"horizontal","hyperlink":""},{"acf_fc_layout":"content","content":" The result, shown in Figure 5<\/em>, is a set of distinct clusters\u20147 in above example\u2014each rendered in a different color that emerge directly from the structure of the embedding space. The pattern shows the clusters align closely with recognizable large-scale landscape types rather than administrative or political boundaries. The orange cluster traces the Sahara and Arabian deserts almost precisely. Red highlights the humid tropical belt across Central and West Africa. Green marks vegetated transition zones and highlands, while blue is seen around the Congo Basin’s dense equatorial forest core. Even without a labeled example, the embeddings alone are enough to separate desert, savanna, tropical forest, and montane environments into coherent, spatially contiguous groups.<\/p>\n This makes for a useful complement to the similarity-search workflow shown earlier. The Find Similar Features Using Embeddings tool answers “Where else looks like this place I already know?”, but running the Multivariate Clustering tool on the same embeddings answers a broader question, “How many distinct landscape types exist in this study area, and where are their boundaries?” Used together, the two tools provide both a targeted search capability and a way to discover landscape typologies that analysts may not have anticipated, all built on the same underlying embedding features.<\/p>\n<\/div>\n"},{"acf_fc_layout":"content","content":" 4. Train a predictive model<\/p>\n Beyond similarity search and clustering, location embeddings can also be used to train or improve machine learning models for regression and classification tasks. The Train Using AutoML GeoAI tool includes a Use Location Embeddings<\/em> option that automatically enriches your input dataset on the fly with location embeddings before training, without requiring a separate embedding generation step. This is particularly valuable for datasets in which the target variable has a strong but hard-to-quantify spatial dependency\u2014such as forest density, population distribution, health outcomes, or groundwater condition\u2014where nearby geographic context often carries as much of a predictive signal as the tabular attributes themselves.<\/p>\n<\/div>\n"},{"acf_fc_layout":"image","image":{"ID":2976417,"id":2976417,"title":"im5","filename":"im5.png","filesize":85697,"url":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im5.png","link":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/geoai\/beyond-latitude-and-longitude-giving-geographic-context-to-coordinates\/im5","alt":"tool screenshot","author":"370342","description":"","caption":" \n\nFigure 6. Training an AutoML model using location embeddings. ","name":"im5","status":"inherit","uploaded_to":2976369,"date":"2026-07-10 21:43:20","modified":"2026-07-11 13:13:26","menu_order":0,"mime_type":"image\/png","type":"image","subtype":"png","icon":"https:\/\/www.esri.com\/arcgis-blog\/wp-includes\/images\/media\/default.png","width":330,"height":628,"sizes":{"thumbnail":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im5-213x200.png","thumbnail-width":213,"thumbnail-height":200,"medium":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im5.png","medium-width":137,"medium-height":261,"medium_large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im5.png","medium_large-width":330,"medium_large-height":628,"large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im5.png","large-width":330,"large-height":628,"1536x1536":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im5.png","1536x1536-width":330,"1536x1536-height":628,"2048x2048":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im5.png","2048x2048-width":330,"2048x2048-height":628,"card_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im5-244x465.png","card_image-width":244,"card_image-height":465,"wide_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im5.png","wide_image-width":330,"wide_image-height":628}},"image_position":"center","orientation":"horizontal","hyperlink":""},{"acf_fc_layout":"content","content":" When this option is enabled, Train Using AutoML<\/a> generates location embeddings for each training sample. This process is described in more detail in section 1.f of the AutoML training workflow, See How AutoML Works<\/a> to learn more about the underlying process. View a sample notebook demonstrating location embeddings for enhanced house price prediction in King County using AutoML<\/a> for a worked example.<\/p>\n"},{"acf_fc_layout":"image","image":{"ID":2976553,"id":2976553,"title":"","filename":"im1-1.png","filesize":287511,"url":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im1-1.png","link":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/geoai\/beyond-latitude-and-longitude-giving-geographic-context-to-coordinates\/im1-2","alt":"Side-by-side maps illustrating baseline and location embedding\u2013enhanced groundwater classification across India.","author":"370342","description":"","caption":"Figure 7. Groundwater classification across India administrative blocks, baseline classification, and classification with location embeddings.","name":"im1-2","status":"inherit","uploaded_to":2976369,"date":"2026-07-11 23:02:14","modified":"2026-07-11 23:11:13","menu_order":0,"mime_type":"image\/png","type":"image","subtype":"png","icon":"https:\/\/www.esri.com\/arcgis-blog\/wp-includes\/images\/media\/default.png","width":798,"height":217,"sizes":{"thumbnail":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im1-1-213x200.png","thumbnail-width":213,"thumbnail-height":200,"medium":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im1-1.png","medium-width":464,"medium-height":126,"medium_large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im1-1.png","medium_large-width":768,"medium_large-height":209,"large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im1-1.png","large-width":798,"large-height":217,"1536x1536":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im1-1.png","1536x1536-width":798,"1536x1536-height":217,"2048x2048":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im1-1.png","2048x2048-width":798,"2048x2048-height":217,"card_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im1-1.png","card_image-width":798,"card_image-height":217,"wide_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im1-1.png","wide_image-width":798,"wide_image-height":217}},"image_position":"center","orientation":"horizontal","hyperlink":""},{"acf_fc_layout":"content","content":" In this case, administrative block boundaries across India (as shown in the first image of Figure 7) were symbolized by groundwater class \u2014 Critical, Safe, Saline, Semi-Critical, and Over-Exploited \u2014 serving as the input geography and ground-truth reference for the classification task. Two AutoML models were then trained on this data: one without location embeddings, and one with the Use Location Embeddings option enabled as show in Figure 6<\/em>.<\/p>\n The baseline model, trained without location embeddings, misclassified administrative blocks concentrated notably along parts of western and central India, totaling 112 misclassified areas (red shown in the center image of Figure 7<\/em>). The model trained with location embeddings enabled, by contrast, brought this number down sharply to just 30 misclassified blocks, with the vast majority of the country now correctly classified (green shown in the third image of Figure 7<\/em>). This reduction underscores the importance of location embeddings in spatial groundwater classification . Neighboring blocks tend to share hydrogeological conditions, and location embeddings refine the model with spatial context, rather than recording each block as an isolated observation.<\/p>\n"},{"acf_fc_layout":"content","content":" The following are limitations of the model:<\/p>\n Location embeddings offer a new lens for GIS, turning raw coordinates into more accurate representations of place that require minimal training data, do not require manual feature engineering, and can be used in a variety of scenarios, from study-area mineral exploration to regional terrain mapping to global desert detection.<\/p>\n<\/div>\n"},{"acf_fc_layout":"content","content":" References<\/strong><\/p>\n\n
What a\u00a0<\/span>l<\/span>ocation\u00a0<\/span>e<\/span>ncoder\u00a0<\/span>d<\/span>o<\/span>es<\/span><\/span><\/strong>\u00a0<\/span><\/h2>\n
Why\u00a0<\/span>l<\/span>ocation\u00a0<\/span>e<\/span>ncoding\u00a0<\/span>m<\/span>atters in GIS<\/span><\/span>\u00a0<\/span><\/strong><\/h2>\n
From\u00a0<\/span>c<\/span>oordinates to\u00a0<\/span>i<\/span>nsights<\/span><\/strong><\/h2>\n
\n
Us<\/span>e<\/span>\u00a0and\u00a0<\/span>appl<\/span>y the\u00a0<\/span>model with tools<\/span><\/strong><\/h2>\n
\n
\n<\/span>The following\u00a0examples span\u00a0various\u00a0scales and terrain types\u2014from a regional mineral exploration use case to a worldwide desert similarity search\u2014to illustrate how this workflow performs across diverse geographic contexts.<\/li>\n<\/ol>\n"},{"acf_fc_layout":"content","content":"Find an open-pit coal mine<\/strong><\/h3>\n"},{"acf_fc_layout":"image","image":{"ID":2976554,"id":2976554,"title":"","filename":"im2-1.png","filesize":284875,"url":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im2-1.png","link":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/geoai\/beyond-latitude-and-longitude-giving-geographic-context-to-coordinates\/im2-3","alt":"Similarity heat map of an open-pit coal mining study area showing areas with high and low similarity to the query mine polygons.","author":"370342","description":"","caption":"Figure 1. Similarity heat map generated for an open-pit coal mining study area using GeoAI location embeddings.","name":"im2-3","status":"inherit","uploaded_to":2976369,"date":"2026-07-11 23:02:16","modified":"2026-07-11 23:06:21","menu_order":0,"mime_type":"image\/png","type":"image","subtype":"png","icon":"https:\/\/www.esri.com\/arcgis-blog\/wp-includes\/images\/media\/default.png","width":800,"height":500,"sizes":{"thumbnail":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im2-1-213x200.png","thumbnail-width":213,"thumbnail-height":200,"medium":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im2-1.png","medium-width":418,"medium-height":261,"medium_large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im2-1.png","medium_large-width":768,"medium_large-height":480,"large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im2-1.png","large-width":800,"large-height":500,"1536x1536":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im2-1.png","1536x1536-width":800,"1536x1536-height":500,"2048x2048":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im2-1.png","2048x2048-width":800,"2048x2048-height":500,"card_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im2-1-744x465.png","card_image-width":744,"card_image-height":465,"wide_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im2-1.png","wide_image-width":800,"wide_image-height":500}},"image_position":"center","orientation":"horizontal","hyperlink":""},{"acf_fc_layout":"content","content":"
Identify Similar Himalayan Mountain Terrain<\/strong><\/h3>\n"},{"acf_fc_layout":"image","image":{"ID":2976555,"id":2976555,"title":"","filename":"im4-1.png","filesize":895031,"url":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im4-1.png","link":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/geoai\/beyond-latitude-and-longitude-giving-geographic-context-to-coordinates\/im4-2","alt":"Map showing locations identified as most similar to a Himalayan mountain query using location embeddings","author":"370342","description":"","caption":"Figure 2. Similar feature identification for a Himalayan mountain query using GeoAI location embeddings.","name":"im4-2","status":"inherit","uploaded_to":2976369,"date":"2026-07-11 23:02:21","modified":"2026-07-11 23:07:36","menu_order":0,"mime_type":"image\/png","type":"image","subtype":"png","icon":"https:\/\/www.esri.com\/arcgis-blog\/wp-includes\/images\/media\/default.png","width":800,"height":515,"sizes":{"thumbnail":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im4-1-213x200.png","thumbnail-width":213,"thumbnail-height":200,"medium":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im4-1.png","medium-width":405,"medium-height":261,"medium_large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im4-1.png","medium_large-width":768,"medium_large-height":494,"large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im4-1.png","large-width":800,"large-height":515,"1536x1536":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im4-1.png","1536x1536-width":800,"1536x1536-height":515,"2048x2048":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im4-1.png","2048x2048-width":800,"2048x2048-height":515,"card_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im4-1-722x465.png","card_image-width":722,"card_image-height":465,"wide_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im4-1.png","wide_image-width":800,"wide_image-height":515}},"image_position":"center","orientation":"horizontal","hyperlink":""},{"acf_fc_layout":"content","content":"
Identify similar features on the Gangetic plains<\/strong><\/h3>\n"},{"acf_fc_layout":"image","image":{"ID":2976556,"id":2976556,"title":"","filename":"im5-1.png","filesize":808838,"url":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im5-1.png","link":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/geoai\/beyond-latitude-and-longitude-giving-geographic-context-to-coordinates\/im5-2","alt":"Visualization of locations with embedding similarity to a Gangetic Plains reference area.","author":"370342","description":"","caption":"Figure 3. Similar feature identification for a Gangetic plains query using GeoAI location embeddings.","name":"im5-2","status":"inherit","uploaded_to":2976369,"date":"2026-07-11 23:02:24","modified":"2026-07-11 23:08:38","menu_order":0,"mime_type":"image\/png","type":"image","subtype":"png","icon":"https:\/\/www.esri.com\/arcgis-blog\/wp-includes\/images\/media\/default.png","width":799,"height":525,"sizes":{"thumbnail":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im5-1-213x200.png","thumbnail-width":213,"thumbnail-height":200,"medium":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im5-1.png","medium-width":397,"medium-height":261,"medium_large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im5-1.png","medium_large-width":768,"medium_large-height":505,"large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im5-1.png","large-width":799,"large-height":525,"1536x1536":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im5-1.png","1536x1536-width":799,"1536x1536-height":525,"2048x2048":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im5-1.png","2048x2048-width":799,"2048x2048-height":525,"card_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im5-1-708x465.png","card_image-width":708,"card_image-height":465,"wide_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im5-1.png","wide_image-width":799,"wide_image-height":525}},"image_position":"center","orientation":"horizontal","hyperlink":""},{"acf_fc_layout":"content","content":"
Search for global desert similarity<\/strong><\/h3>\n"},{"acf_fc_layout":"image","image":{"ID":2976558,"id":2976558,"title":"","filename":"im7-1.png","filesize":490982,"url":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im7-1.png","link":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/geoai\/beyond-latitude-and-longitude-giving-geographic-context-to-coordinates\/im7-2","alt":"similarity map showing how closely landscapes across the Americas match the Sonoran Desert.","author":"370342","description":"","caption":"Figure 4. Similarity scores for the Sonoran Desert query, viewed across the Americas.","name":"im7-2","status":"inherit","uploaded_to":2976369,"date":"2026-07-11 23:02:30","modified":"2026-07-11 23:09:35","menu_order":0,"mime_type":"image\/png","type":"image","subtype":"png","icon":"https:\/\/www.esri.com\/arcgis-blog\/wp-includes\/images\/media\/default.png","width":800,"height":516,"sizes":{"thumbnail":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im7-1-213x200.png","thumbnail-width":213,"thumbnail-height":200,"medium":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im7-1.png","medium-width":405,"medium-height":261,"medium_large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im7-1.png","medium_large-width":768,"medium_large-height":495,"large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im7-1.png","large-width":800,"large-height":516,"1536x1536":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im7-1.png","1536x1536-width":800,"1536x1536-height":516,"2048x2048":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im7-1.png","2048x2048-width":800,"2048x2048-height":516,"card_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im7-1-721x465.png","card_image-width":721,"card_image-height":465,"wide_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/07\/im7-1.png","wide_image-width":800,"wide_image-height":516}},"image_position":"center","orientation":"horizontal","hyperlink":""},{"acf_fc_layout":"content","content":"
Limitations<\/strong><\/h2>\n
\n
Conclusion<\/strong><\/h2>\n
\n