Over the last few decades, it has become increasingly evident that passenger vehicles are by far the most dangerous way to travel. As a result, it has become more and more important to find an efficient and effective method to predict traffic risk. However, predicting traffic accidents and where they are likely to occur is a very complex problem, with large amounts of data being needed for most meaningful predictions.

At Pacific Spatial Solutions, we are currently trying to tackle this problem by training a machine learning model to predict road and intersection risk in Japan nationwide. As we are trying to predict traffic risk on a national level it is only natural that the data we use cover the same area.

High-quality satellite imagery is essential to assess the carbon impact of nature-based forest conservation and restoration projects [1]. However, getting that high quality imagery is uniquely difficult in areas that need carbon financing the most: tropical forests. Tropical forests present a unique challenge for satellite imagery analysis due to persistent cloud cover, which often renders optical imagery unusable and creates data gaps.

Example composites highlight how the HLS-L30 product alone can have gaps when attempting to make a seasonal composite, as fewer cloud-free observations.

Ever tried to make sense of the myriad file types in spatial data science and felt like you've wandered into a linguistic labyrinth? Trust me, you're not alone. As a data scientist who's spent more time wrangling datasets than I care to admit, I thought I'd take a casual stroll down memory lane with an old high school friend: regression models. Just a simple plot of actual vs. predicted, right? But when spatial data's involved, you can't just sit back and relax—you've got to keep one eye on the geometries.

I'm currently working on an agricultural project, and growing up on a farm gives me a personal stake in this. This blog illustrates my solution to the geometry debacle. I'll first take you to the area where I grew up: Lyon County.

Overture is an open data project that publishes interoperable map datasets. It aims to foster an ecosystem of developers creating downstream map services around its data products. Fused emerged as a solution to enrich Overture datasets on the fly and serve them with XYZ Tile endpoints.

This clip shows how coverage expands in (top right) Astoria when I add building heights from the NSI dataset (as num_story * 3) to Overture buildings.

This article explores a User Defined Function (UDF) that utilizes global precipitation data to compare rainfall patterns between different locations worldwide and then creates an interactive app!

Fused is a powerful and versatile tool that can do nearly anything with just Python. Its versatility is its strength, but it is also an obstacle. It's easy to walk about wondering: what, concretely, can Fused do for me?

Here are six concrete ways you can use Fused today.

In this article I show how to create an object detection layer on 50cm imagery in realtime. It explains how to create a Fused User Defined Function (UDF) to load satellite image tiles to call an inference model, then publish it as an interactive map app.

In this video tutorial, I show a complete data app workflow in Fused. Starting with exploring the data in Fused, the tutorial walks through developing a UDF to serve the data, and then a Fused App to share results.

With Fused, this whole workflow takes just minutes from beginning to end. Fused helps me visualize the data at every step, iterate on my analytical logic, and finally publish a dashboard.

In an era where data-driven decisions are vital, accurate and scalable vegetation analysis plays a crucial role across various industries, from environmental monitoring to urban planning. While AI and machine learning have transformed image analysis, they often bring complexities and resource demands that aren't always practical for large-scale, real-time applications.

Pachama is a technology company that harnesses satellite data and AI to empower companies to confidently invest in nature. The engineering team at Pachama created a Land Suitability Tool to help landowners and project developers qualify parcels of land to implement carbon projects. They turned to Fused to simplify their data workflows.

Isaac Brodsky, the CTO of Fused, delved into the power of Fused during a LinkedIn live session with Matt Forrest. They discussed the contrast of Python vs. SQL for data analytics, the advantages of serverless geospatial processing, and showcased a live demo of the UDF Builder. During the demo, Isaac created a User Defined Function visualize Overture building footprints that are within a certain proximity of water.

The combination of Fused serverless operations and DuckDB offers blazing fast data processing. Fused embraced Python to create serverless User Defined Functions (UDFs). Now, with the help of DuckDB, Fused enables developers to leverage the ease and familiarity of SQL in these functions  -  without compromising performance and parallelism.

Fused is a modern geospatial toolkit for companies to code, scale, and ship geospatial workflows of any size.

This week we are unveiling Fused, a toolkit to enable interoperability between all geospatial datasets and tools in the modern data stack. Fused is the glue layer that integrates data platforms with data tools via a managed serverless API.

Fused enables interoperability between datasets and tools in the modern data stack. It's a glue layer to integrate data platforms with data tools via a managed serverless API.