Suggestions
Todd Crawford
Todd Crawford is a highly respected meteorologist and data scientist with extensive experience in the weather industry. He previously served as the Chief Meteorologist, Senior Project Manager, and Data Scientist at The Weather Company, an IBM Business, where he worked for over 20 years from 2000 to 2021.14
During his tenure at The Weather Company, Todd was known for his exceptional skills in analyzing complex data sets and drawing correlating conclusions.3 His expertise in meteorology and data science made him a valuable asset to the company's weather forecasting and analysis capabilities.
Currently, Todd Crawford holds the position of VP Meteorology at Atmospheric G2.26 In this role, he continues to apply his extensive knowledge and experience in meteorology and data science to advance weather forecasting and climate analysis.
Todd's professional background also includes:
- Post-Doctoral Scientist position from 1998 to 2000, prior to joining The Weather Company4
- A reputation for brilliance in his field, particularly in data analysis and interpretation3
As of March 21, 2024, Todd Crawford was actively involved in recruiting for Atmospheric G2, seeking a Meteorological Data Scientist to join their growing Weather, Climate, and Analytics team.5
Todd is based in Hudson, New Hampshire, United States.3 On social media, he can be found on Twitter under the handle @tcrawf_nh, where he shares his views and opinions on meteorology and related topics.6
Highlights
There was a lot of hype entering the 2024 North Atlantic tropical season, and for good reason. Historically warm SSTs, which helped enable Beryl, the earliest cat-5 hurricane on record in the basin, and a trend towards La Nina conditions definitely favor a "hyperactive" season. But, things have been very quiet since Beryl, with only a hint of things picking up as we turn the calendar to August. Maybe this season won't end up as advertised?
Let's define a "hyperactive" season as having at least 10 hurricanes, including at least 5 major hurricanes. By that definition, we've had 4 hyperactive seasons in the last 25 years - 2005, 2010, 2017, and 2020. It is not coincidental that all four of these seasons were characterized by a transition into La Nina conditions, as we are seeing in 2024. It is also no coincidence that North Atlantic SSTs were unusually warm in those four years, as they are now. So, is our relatively slow start unusual for hyperactive seasons? The table below details the activity for those four seasons, along with 2024, through the end of July (assuming we don't get a named storm in the next 8 days), and we can see that two of the big seasons (2005, 2020) were big early and often, while the other two (2010, 2017) were similarly "quiet" to 2024.
CONCLUSION: It is too early to give up on the idea of a hyperactive 2024 North Atlantic tropical season, given that 2010 and 2017 were similarly quiet into early August before erupting the rest of the season, and given the unusually favorable conditions in place.
When it comes to AI and weather forecasting, the last year has been quite a whirlwind, with the introduction of new AI weather models from various private corporations, universities, and the ECMWF. At Atmospheric G2, we've quickly transitioned from "gee, this is interesting, let's see if these things are useful" to "these are really useful, how can we integrate these to serve our clients best".
The ECMWF has certainly been the leader in bringing AI weather models to the masses with the release of their AIFS model. The AIFS has been integrated into the skill-weighted, bias-corrected blend of models that Atmospheric G2 uses since mid-May, with positive results and improved overall accuracy.
Last week, ECMWF announced a probabilistic version of AIFS, adapting a rather mind-blowing diffusion-based technique initially developed by the team at Google in 2023. Not only is the first version of of the AIFS ensemble already slightly better than the legacy ECMWF ensemble (based on probabilistic metrics like CRPS), the diffusion-based technique results in forecasts that do not become overly smoothed with time, like the deterministic AIFS and most of the other recent AI models.
At Atmospheric G2, we are excited to integrate (and, more importantly, try to keep up with) the latest and greatest in the accelerating AI weather modeling world...
#ai #weatherforecasting #ag2 #energy
