Frederik Kratzert
@kratzert.bsky.social
Research Scientist @ Google Research. Working on the intersection of machine learning and hydrology. Located in Vienna, Austria.
Bzgl Lebensmittelautomaten: Hier die wienerische / bowa Variante die einem 24/7 Zugang zu frischen Bohnenglück garantiert. ☕
June 2, 2025 at 4:47 PM
Bzgl Lebensmittelautomaten: Hier die wienerische / bowa Variante die einem 24/7 Zugang zu frischen Bohnenglück garantiert. ☕
Found something from the past. Must be from my Bachelor thesis time so maybe around 2012/2013?!
May 24, 2025 at 2:16 PM
Found something from the past. Must be from my Bachelor thesis time so maybe around 2012/2013?!
Amazing, we have been looking for a global river network with bifurcations for a long time. I just checked my hometown on the website, any idea what is happening here with the Donaukanal? IIUC, it is not correctly identified as a bifurcating sidearm of the Danube but rather a headwater catchment?
May 19, 2025 at 8:05 AM
Amazing, we have been looking for a global river network with bifurcations for a long time. I just checked my hometown on the website, any idea what is happening here with the Donaukanal? IIUC, it is not correctly identified as a bifurcating sidearm of the Danube but rather a headwater catchment?
Arrived in Denver and was welcomed with a huge downpour+rainbow over the airport area, as seen from the bus towards Boulder
May 11, 2025 at 11:30 PM
Arrived in Denver and was welcomed with a huge downpour+rainbow over the airport area, as seen from the bus towards Boulder
Looking at my heart rate it seems like I did serious sports during my talk.
May 1, 2025 at 1:39 PM
Looking at my heart rate it seems like I did serious sports during my talk.
Today in the afternoon, we have the oral blocks of our session "Deep learning in hydrology" at #EGU25. Looking forward to all the talks and also very much to our invited keynote speaker Andy Wood.
April 30, 2025 at 7:23 AM
Today in the afternoon, we have the oral blocks of our session "Deep learning in hydrology" at #EGU25. Looking forward to all the talks and also very much to our invited keynote speaker Andy Wood.
Currently it's raining, but the forecast for next week looks great. The best season of the year in Vienna is starting, just in time for #EGU25
April 25, 2025 at 6:33 AM
Currently it's raining, but the forecast for next week looks great. The best season of the year in Vienna is starting, just in time for #EGU25
Some Caravan swag arrived just in time for #EGU25. If you are interested in Caravan stickers or coasters, just hit me up during the conference.
April 22, 2025 at 12:43 PM
Some Caravan swag arrived just in time for #EGU25. If you are interested in Caravan stickers or coasters, just hit me up during the conference.
Here is another one: This is from a presentation in late 2017 (before my first LSTM related paper). Already here, I was very curious on looking at the learned understanding of the model. Here, the "snow cell", that made it even into hess.copernicus.org/articles/22/...
April 14, 2025 at 9:48 AM
Here is another one: This is from a presentation in late 2017 (before my first LSTM related paper). Already here, I was very curious on looking at the learned understanding of the model. Here, the "snow cell", that made it even into hess.copernicus.org/articles/22/...
Guess the "joke" we were trying to make here was not as obvious than we thought.
April 7, 2025 at 2:08 PM
Guess the "joke" we were trying to make here was not as obvious than we thought.
Is this an April's fool joke? I really don't need another community platform and to curate a feed there with "topics of interest". Also, why would I want to see only research papers from a single publisher?
I think we are all better off if we simply share our papers on BlueSky?
I think we are all better off if we simply share our papers on BlueSky?
April 1, 2025 at 2:56 PM
Is this an April's fool joke? I really don't need another community platform and to curate a feed there with "topics of interest". Also, why would I want to see only research papers from a single publisher?
I think we are all better off if we simply share our papers on BlueSky?
I think we are all better off if we simply share our papers on BlueSky?
Furthermore:
- bug fixes
- documentation improvements around forecast models.
I took that opportunity to look at the GitHub stars over time. Almost scary how linear that trend is 😅
- bug fixes
- documentation improvements around forecast models.
I took that opportunity to look at the GitHub stars over time. Almost scary how linear that trend is 😅
March 17, 2025 at 10:15 PM
Furthermore:
- bug fixes
- documentation improvements around forecast models.
I took that opportunity to look at the GitHub stars over time. Almost scary how linear that trend is 😅
- bug fixes
- documentation improvements around forecast models.
I took that opportunity to look at the GitHub stars over time. Almost scary how linear that trend is 😅
Large-Sample Hydrology at EGU25 is looking packed. A lot of new members for the CAMELS family. Hopefully also joining the Caravan soon. 🐪
I'll be presenting for my colleague Guy, our Caravan extension that adds various different weather products as well as forecast archives to all Caravan gauges.
I'll be presenting for my colleague Guy, our Caravan extension that adds various different weather products as well as forecast archives to all Caravan gauges.
March 14, 2025 at 8:21 PM
Large-Sample Hydrology at EGU25 is looking packed. A lot of new members for the CAMELS family. Hopefully also joining the Caravan soon. 🐪
I'll be presenting for my colleague Guy, our Caravan extension that adds various different weather products as well as forecast archives to all Caravan gauges.
I'll be presenting for my colleague Guy, our Caravan extension that adds various different weather products as well as forecast archives to all Caravan gauges.
Closest station I could quickly find on our national met agency zamg.ac.at. Looks veeeery normal to me, no? 🫣
March 10, 2025 at 6:36 PM
Closest station I could quickly find on our national met agency zamg.ac.at. Looks veeeery normal to me, no? 🫣
Pretty interesting to see the development of NeuralHydrology starts over time. It is almost perfectly linear.
github.com/neuralhydrol...
github.com/neuralhydrol...
January 30, 2025 at 8:40 AM
Pretty interesting to see the development of NeuralHydrology starts over time. It is almost perfectly linear.
github.com/neuralhydrol...
github.com/neuralhydrol...
Seems like the #EGU25 submission period closed, so here is a fun little graph:
Abstract submission count over time relative to the submission deadline for our "Deep Learning in Hydrology" session at #EGU25. At this point I might be able to predict next years submission count 😅
Abstract submission count over time relative to the submission deadline for our "Deep Learning in Hydrology" session at #EGU25. At this point I might be able to predict next years submission count 😅
January 15, 2025 at 12:45 PM
Results for my own abstract.
I actually plan to submit to our own session (Deep Learning in Hydrology) and since we decided many years ago to never give an oral to an abstract of a co-author, finally return to a poster presentation.
I actually plan to submit to our own session (Deep Learning in Hydrology) and since we decided many years ago to never give an oral to an abstract of a co-author, finally return to a poster presentation.
January 8, 2025 at 7:59 AM
Results for my own abstract.
I actually plan to submit to our own session (Deep Learning in Hydrology) and since we decided many years ago to never give an oral to an abstract of a co-author, finally return to a poster presentation.
I actually plan to submit to our own session (Deep Learning in Hydrology) and since we decided many years ago to never give an oral to an abstract of a co-author, finally return to a poster presentation.
Since only the new GPU fits into my tower/motherboard, I guess it is finally time to retire my old friend. This beautiful 970 powered all of my early research. You will be remembered my friend!
January 3, 2025 at 3:18 PM
Since only the new GPU fits into my tower/motherboard, I guess it is finally time to retire my old friend. This beautiful 970 powered all of my early research. You will be remembered my friend!
Found some time today to install my Christmas present. A nice little upgrade from my old 970+2060 dual setup to a 4070 TI Super. But damn these GPUs are large.
January 3, 2025 at 3:16 PM
Found some time today to install my Christmas present. A nice little upgrade from my old 970+2060 dual setup to a 4070 TI Super. But damn these GPUs are large.
Here is what I came up with yesterday: A mini-game on a liquid crystal display, where you control a spaceship (>) and need to avoid comets (*) that fly in from right to left. You can only move up and down (2 buttons left) and if you hit a comet, the game ends. It can be restarted from the 3rd button
January 3, 2025 at 12:10 PM
Here is what I came up with yesterday: A mini-game on a liquid crystal display, where you control a spaceship (>) and need to avoid comets (*) that fly in from right to left. You can only move up and down (2 buttons left) and if you hit a comet, the game ends. It can be restarted from the 3rd button
Chocolate+ Pistachio from before the "Dubai Schokolade"-Trend.
December 30, 2024 at 4:51 PM
Chocolate+ Pistachio from before the "Dubai Schokolade"-Trend.
My colleague Martin Gauch will present our ongoing research on end-to-end trained, deep learning based distributed hydrology models today in the morning session at poster 983 at #AGU24.
December 13, 2024 at 1:24 PM
My colleague Martin Gauch will present our ongoing research on end-to-end trained, deep learning based distributed hydrology models today in the morning session at poster 983 at #AGU24.
Martin Gauch, one of my colleagues, is currently giving a talk about Google's Flood Forecasting Initiative at the Google booth at #AGU24
December 11, 2024 at 7:03 PM
Martin Gauch, one of my colleagues, is currently giving a talk about Google's Flood Forecasting Initiative at the Google booth at #AGU24