🍂 Fall Outlook from Climate Prediction Center

🍂 Fall Outlook from Climate Prediction Center

The Climate Prediction Center recently released its fall outlook for the United States which includes September, October, and November. They note that ENSO-neutral conditions are present and sea surface temperatures are near average across most of the tropical Pacific Ocean. ENSO-neutral conditions are expected through the end of summer and into early fall, followed by a brief period of favored La Nina conditions in the fall and early winter. The fall temperature outlook favors above-normal temperatures for the majority of the contiguous U.S. and Alaska. The fall precipitation outlook favors below-normal precipitation in the Central Great Basin and Southwest into the central and southern Plains, and parts of the Middle Mississippi, Ohio, Tennessee, and Lower Mississippi Valley regions. Above-normal precipitation is favored over western Alaska, the Pacific Northwest, and the coastal Southeast.

Climate Prediction Center

☀️ Summer Heat is Arriving Earlier and Ending Later

Climatologist Brian Brettschneider examined the hottest 90 days of the year from 1965 to 1994 and compared their frequency over 1995 to 2024 and found that summer temperatures in the United States are arriving earlier and ending later in the year. The southern United States has experienced the most warming, gaining two extra weeks of hot summer temperatures. Summer temperatures are lasting about 9 days longer in Washington D.C. and about 6 more days in New York City and Seattle than 30 years ago. The Midwest has seen the least lengthening of the season, with some areas actually shrinking. The cause of the prolonged summer heat is contributed to the overall increase in global temperatures and increases in greenhouse gas emissions. The past ten years have been the hottest on record.

Kasha Patel and Naema Ahmed, The Washington Post

📉 Bigger Climate Models are Not Always Better

A new study by a team of researchers from MIT shows that simpler models can outperform deep learning models for climate prediction. The team used climate emulators that rapidly simulate the effect of human activities onto a future climate to compare traditional physics-based techniques like linear pattern scaling (LPS) models to state-of-the-art deep-learning models. Their results showed that LPS outperformed deep-learning models on predicting nearly all parameters they tested, including temperature and precipitation. The high amount of natural variability in climate model runs can cause deep learning models to perform poorly on unpredictable long-term oscillations, like El Nino and La Nina. When more data that addresses natural climate variability was added to the deep-learning model, the deep-learning model performed slightly better than LPS for local precipitation, but LPS was still more accurate for temperature predictions.

Adam Zewe, MIT News