To improve solar and wind renewable energy forecasts by improving the representation of boundary layer and other processes in numerical weather prediction models
Developing forecasts, observations of wind and solar resources, and tools to improve the efficiency and sustainability of the energy system through better understanding and modeling.
NOAA provides the national-scale meteorological observations and numerical weather prediction forecast models used by the renewable energy (RE) industry. As the nation’s wind and solar industries grow, NOAA faces increased demands for better products and services, including improved meteorological observations and more accurate wind and cloud forecasts over a range of timescales. NOAA’s Earth System Research Laboratory (ESRL) is uniquely qualified to provide the improved weather forecasts, observations, and climate information needed to support the effective planning for and efficient operation of a national renewable energy system.
With highly accurate observations, forecasts, and understanding of how wind and solar resources vary and co-vary across time and space, the electric grid will be better able to accommodate the variable nature of wind and solar energy. This will yield greater production of carbon-free renewable energy while also reducing air pollutant emissions.
ESRL is currently working with the wind and solar energy industries and the Department of Energy to improve existing meteorological observing networks and weather forecast models for RE applications. ESRL is working to improve NOAA’s numerical weather prediction guidance, which is used as input to the private sector’s tailored forecast products.
The U.S. has agreed to cut its greenhouse gas emissions by 26-28 percent by 2025 and by 80 percent by 2050, compared to 2005 levels. To meet these goals, a large proportion of electricity that otherwise would have been produced from fossil fuels will need to be generated instead by low-carbon sources—most likely wind and solar power.
Because wind and solar power production depend on the weather, they are variable. This variability of wind and solar power introduces unique challenges to those who must maintain the constant balance between energy supply and demand required for a stable electric power grid. Unless and until energy storage is economical, “flexibility” in the power grid is key to its efficient operation. Improved forecasting across a range of time scales for wind and solar resources will provide critical flexibility and facilitate integration of weather-dependent renewable energy.
There are several ways forecast skill can be improved. One way is to better model atmospheric phenomena, by improving various parts of the weather models known as “schemes” and the mathematical coupling of these “schemes” to other schemes. Another way of improving forecast skill is to improve the data assimilation methods, and another approach is to improve our observations of relevant phenomena. The Earth System Research Laboratory (ESRL) is working on all of these.
ESRL has recently begun to optimize two of its numerical weather prediction (NWP) models—the Rapid Refresh and the High Resolution Rapid Refresh—for wind and solar applications. Specifically, research being done to improve forecast skill is targeting the intersection of wind and power with the atmosphere, including processes such as: turbulence, low level jets, shear, and formation and movement of clouds and aerosols.
ESRL is also performing research to determine the optimal suite of sensors in a national observation network to support integration of wind and solar into the power system. More vertical profiles of winds and more and higher-quality observations of solar irradiance (both total and direct) would support improvements in forecast skill. In the first Wind Forecast Improvement Project, ESRL and its partners were able to collect additional vertical profiles of winds with instruments that were available for the duration of the twelve-month field campaign. Several utility companies and Independent System Operators (ISOs) are helping fill the gap in observations by sharing with NOAA the meteorological measurements they collect at wind and solar plants.
Further research needs include an improved understanding of the co-variability of wind and solar resources, together with energy demand, on broader spatial and temporal scales; research to identify whether and how large-scale climate drivers, such as the El Nino Southern Oscillation and the Pacific Decadal Oscillation, affect wind and solar resources; and improved predictions at two-week, seasonal, annual, and decadal time scales.
The Rapid Refresh (RAP) is an hourly updated weather forecast model/assimilation system, which went into operation on May 1, 2012, at the National Centers for Environmental Prediction (NCEP) as NOAA's hourly updated model. RAP version 2, a major upgrade, was implemented at NCEP on February 25, 2014.
Scientists from the Earth System Research Laboratory's, Global Sciences Division work with colleagues from NCEP, the National Center for Atmospheric Research, and other labs on RAP development.
The High-Resolution Rapid Refresh (HRRR) is a NOAA real-time 3-km resolution, hourly updated, cloud-resolving atmospheric model. The NCEP HRRR has been operational since September 30, 2014.
One of the challenges of integrating large amounts of wind and solar power onto the electric grid is the high temporal variability of these power sources. That is, one gusty day can be followed by a calm day. That means wind power generation can change by large amounts very rapidly, an occurrence called a wind ramp event. NOAA ESRL researchers have developed a Ramp Tool and Metric to identify these wind ramps, and quantify model skill at forecasting them.
This ramp tool has three components: the first is a process to identify ramp events in the time series of power. The second component is a method for matching in time each forecast ramp event with the most appropriate observed ramp event. The third and last component of the ramp tool is a process through which a skill score of the forecast model is determined.