There’s an enormous amount of open satellite data that covers Nepal’s glacial lakes (and glaciers). A lot of it is unusable without a week of validating and processing it.

I started an open source project with the aim to change that for at least one lake.

(tl;dr – Open Source Repo here: github.com/andjames/OpenImja)

On the 26th, I sat in on NASA’s webinar on how to stream datasets from Earthdata. It happened to be the same day as the collapsed glacier that led to a landslide and flood in Nepal.

It was impressive to see the Earth Observation community open up datasets in response.

The problem though is that every service is different. Every API has a different structure and login requirements. Some data is optical, some radar or hyperspectral, some comes as raw data that needs significant processing.

There’s no single place — or process — to pull it together and determine what’s actually useful.

And “data available” doesn’t necessarily mean “data usable.” One cloud can ruin a satellite pass.

So I decided to start experimenting pulling disparate datasets into one open-source repo.

I’m starting with one glacial lake in Nepal: Imja Tsho, just below Island Peak (Imja Tse) and above the valley that leads to Everest.

I first was there in 2003 on a volunteer trip with Porters’ Progress, looking at the living conditions, equipment and first-aid access of porters working the Everest route. Imja Tsho was already a concern then.

An early-warning system was eventually installed downstream, but Nepali officials told the BBC in April that it may no longer work.

Meanwhile, there’s an incredible amount of data on the lake:

Almost 40 years of Landsat imagery. Sentinel-1 and Sentinel-2. NASA high-resolution elevation data. Historical sonar/bathymetric surveys. Ground monitoring. Research papers and flood models. And that’s just existing open source data – commmercial providers have higher res data with more recent acquisitions.

But it’s scattered across the internet and buried in different formats.

OpenImja is an attempt to pull that into one reproducible, open dataset — including not just what data is listed as “available”, but what is “usable.”

I’ve started with Google Earth Engine data and a 1985–2026 satellite archive scrape for NDWI-derived lake measurement data.

There’s a lot more to do.

If anyone working in Earth observation, glaciology, Nepal open data or adjacent areas wants to contribute — or tell me what I’m getting wrong — the repo is here: https://github.com/andjames/OpenImja

Timelapse from Google Earth of Imja Tsho’s growth since 1984. (Source: CNES / Airbus (11/13/2021–3/7/2026), Maxar Technologies (older–2/20/2010)