Three months into the 2023-24 water year, we have our first early look at what sort of runoff to expect on the Rio Grande in the coming year, and it doesn’t look great. The January NRCS median forecast for March-July runoff is 42 percent of “normal” at Otowi, the critical forecast point where the Rio Grande enters New Mexico’s Middle Rio Grande. It’s still early in the snow season, with a wide range of possible outcomes depending on the storm patterns over the next few months. But the best possible outcome (statistically a one chance in 20 of this much water) is still below the 30-year median.

In other words, we’re pretty clearly on track for a below-average runoff year.

The forecast uses the NRCS’s new Multi-Model Machine-learning Metasystem (M4) forecasting tool, part of an effort to develop improved statistical tools using machine learning approaches to the big snowpack datasets rather than the principal components analysis used in the past. The peer-reviewed paper laying out the testing done over the last half decade suggests significant improvement in the tricky task of forecasting runoff.

The biggest uncertainty is always the weather, but I’m excited to see the new, improved statistical models shifting from the research world to operations.