Just Published

Down-to-earth predictions

A colorful satellite image of the eastern United States.

Forecasts of soil temperature tend to improve when a climatological index of leaf area (top, for a typical July 1) is replaced by a more detailed index derived from satellite data (bottom, for the period around July 1, 2006). (Images courtesy Kevin Manning.)

In a potential boon for agriculture, a NASA-funded effort that involves NCAR and the private firm DTN/Meteorlogix has produced one of the world’s most accurate systems for predicting soil temperature up to two days in advance. Since the project began in 2006, the typical 48-hour error in soil temperatures has been reduced by about 10% at a depth of 2 inches (5 centimeters) and by around 50% at a depth of 4 in (10 cm), as compared to prior results from the NCAR-based High-Resolution Land Data Assimilation System (HRLDAS).

Led by Bill Myers and Fei Chen, the NCAR team is combining HRLDAS and the Noah Land Surface Model with several weather prediction models optimized by NCAR’s Dynamic Integrated ForeCast system. Much of the forecast improvement has come from expanding the number of nodes (soil depths tracked by HRLDAS) from four to six, and from enhancing the parameterization of surface-atmosphere exchange processes.

The team has also incorporated new high-resolution data sets derived from NASA’s Moderate Resolution Imaging Spectroradiometer. These data give a more timely and accurate picture of the seasonal ebb and flow of U.S. vegetation, which affects the amount of radiation entering and leaving the ground and the amount of water entering the air from plants.

The soil forecasts are generated each night at NCAR and sent to DTN, the nation’s largest agricultural weather provider. DTN uses the forecasts as background in producing text products and phone briefings for the firm’s 60,000-plus clients. Eventually, a user-friendly version of the forecasts could go directly to clients. After this year’s growing season, the RAL team will work on further improvements to the forecast system—particularly for soil moisture, where the challenge is more daunting than for soil temperature.