Utilization of real-time vineyard observations to produce downscaled temperature forecasts for frost protection operations

Abstract

Continued drought in California, along with a threatened and endangered salmon population, has required the state of California to control Russian River water usage to protect young, budding winegrape vines during spring frosts. Improved forecasts of the location and duration of frost with increased lead time at individual vineyards could reduce the number of hours of spraying required. In addition, better forecasts could assure adequate supplies of water in off-stream storage systems, reducing demands on tributaries and the threat of fish kills. An automated digital forecast system has been developed and is running routinely to provide vineyard-specific temperature forecasts to growers, commercial frost forecasters, and water managers. The system is based on the use of a high-resolution terrain grid and focuses on real-time hourly observations from 72 vineyards located within the Russian River Basin. The National Weather Service’s (NWS) Graphical Forecast Editor software is configured with a 0.32 km  0.74 km grid spacing, and an objective analysis tool is used to generate a high-resolution observation grid of temperature, dewpoint, and relative humidity—along with computed wet-bulb temperature. The minimum values of each of these parameters also are computed. These observed grids are used, in turn, to bias correct the gridded numerical model and model output statistics guidance grids, as well as the NWS official forecast grids, using a 30-day regression of forecast versus observed values. Verification shows that for the overnight lows (≤2°C), the bias correction improves skill in overnight minimum temperature forecasts by a factor of 3–4 for most forecast guidance used.