Posts

Large language models are here to stay, and I thought it’d be a good idea to write a post about how to download one and run it locally on a personal computer. Of course, this same explanation applies if you’re using a virtual machine in the cloud (AWS, GCP)—the key difference lies in hardware limitations and the size of the models you’re able to run. A personal computer will rarely match the capabilities of a cutting-edge virtual machine....

This post will showcase some applications of one of NASA’s most well-known satellite products: the Black Marble nightlights suite. This set of remotely sensed products allows us to observe how the Earth looks at night, with countless applications in areas such as disaster response, economic activity monitoring, energy infrastructure tracking, and more. Let’s dive in! Figure 1: Nighttime Lights Over Europe, 2023. Product Overview The NASA’s Black Marble product suite provides insights into nighttime lights around the globe....

After spending some time working with GIS files on a daily basis, I realized there’s no comfortable and practical tool for quickly inspecting or transforming files with georeferenced vector data. Having to write three lines of code in a script or in the Python interpreter every time I want to transform geometries feels like a lot, and the sluggishness of desktop applications like ArcGIS and QGIS has always felt like a waste of my life—the cooling fans on my computer say it’s a waste of its life too....

A few weeks ago, I had the opportunity to work with NASA’s MODIS NDVI data. Utilizing an example script from the HDF - EOS Tools and Information Center as reference, I developed a program that takes an area of interest, a start date, and an end date, then downloads and processes the corresponding NDVI data from NASA archives. The program leverages parallel processing to efficiently convert NDVI raster data into vectorized polygons....

One of the most common data requests in geospatial data science is obtaining the road network within a given administrative boundary. This includes identifying road locations, classifying road types, and calculating the total road length within a region—an essential analysis for various applications. OpenStreetMap (OSM) is widely recognized as the go-to source for road network data, thanks to its crowdsourced contributions that provide global coverage. To step outside my usual Python development, I decided to create an R function to extract OSM road network data for a given administrative boundary....

The next article is an ArcGIS StoryMap embedded in an iframe. Some features may not work correctly.

The Problem If you work with Amazon Web Services (AWS) Athena SQL, chances are that you’ve encountered the following error: Query exhausted resources at this scale factor This generally indicates that you are hitting memory limits on the Athena cluster that runs your query (reference 1, reference 2, reference 3). AWS provides several performance optimization strategies, including data optimization, query optimization, and data partitioning. This error can often be the result of a cartesian product when performing spatial operations involving a large number of geometries on both sides of the tables being joined....

A few months ago I had the opportunity to work with the PM2.5 data provided by the Washington University in St. Louis. The initiative comprises a collection of estimated global and regional annual and monthly satellite-derived ground-level fine particulate matter (PM2.5) gridded datasets (Shen et al., 2024), spanning the years 1998 to 2022. This collection combines Aerosol Optical Depth (AOD) retrievals from NASA MODIS, MISR, and SeaWIFS, and is available at resolutions 0....