IEEE Power & Energy Magazine - November/December 2015 - 59

Finally, the need for an ability to evaluate forecasts on the
basis of the value they provide is driving several research and
development efforts. As shown earlier, there are a variety of
metrics that can be used, and currently there is typically not
a direct link between forecast performance metrics and the
value a forecast provides. To this end, the DOE efforts mentioned earlier are developing and demonstrating a set of useful metrics for assessing performance of deterministic and
probabilistic forecasts against both error-based metrics and
metrics that link to the economic value of the forecast via
production-cost modeling and reserve analysis.

Summary and Conclusions
Solar forecasting is one of the lowest cost methods of efficiently
integrating solar energy. In this article, we have focused on
the current state of solar forecasting and identified key issues
related to its development and application. The process of solar
forecasting for various time horizons, methods, and applications has many similarities to wind forecasting, but as solar
output is strongly linked with cloud cover, there are other considerations and possibilities.
Most NWP models do not run at spatial resolutions that are
high enough to explicitly model clouds and do not use detailed
data about cloud cover and cloud formation when initializing.
This means the first three to six hours of most current NWP
models are not particularly useful for solar forecasting. As a
result, methods such as sky imaging and satellite data have
been used to predict near-term solar output, with some success. Additionally, increases in model resolution and more
frequently updated models are helping advance forecasting of
solar irradiance in NWP. Even with improved accuracy, it will
still be important to quantify the uncertainty in the forecast
to ensure efficient integration, and so we described various
methods for quantifying uncertainty.
As performance is a key parameter to any new development, we demonstrated how accuracy can be measured and
evaluated, with a focus on how the metrics can influence the
end result. We showed some of the key factors driving uncertainty such as the look-ahead horizon, forecasting interval
width, tracking systems, and system size. By considering
these factors in plant design and electric system design we
can reduce uncertainty and thus lower the cost to integrate
solar energy.
BTM PV is likely to provide a large part of society's
energy needs in the future. Both detailed bottom-up forecasting methods and aggregated top-down forecasting methods
can be used to manage BTM solar generation. While detailed
methods are likely to be more accurate, data is not always
available, and certain applications may not require such a
fine level of detail. With increasing penetration of BTM
PV, data provision and availability will become increasingly important for successful forecasting and integration.
This includes improved representation of static site data, but
also, where possible, increased access to telemetered output
data or historical meter data. Ongoing research focuses on
november/december 2015

all aspects of the value chain of forecasting, including not
just improving the underlying forecasting methods and combining different forecasting techniques, but also on the ways
that solar forecasts are valued and used in operations.

Further Reading
E. Lorenz, J. Hurka, D. Heinemann, and H. Beyer, "Irradiance forecasting for the power prediction of grid-connected
photovoltaic systems," IEEE J. Select. Topics Appl. Earth
Observ. Remote Sensing, vol. 2, no. 1, pp. 2-10, 2009.
M. Ahlstrom, D. Bartlett, C. Collier, J. Duchesne,
D. Edelson, A. Gesino, M. Keyser, D. Maggio, M. Milligan,
C. Mohrlen, J. O'Sullivan, J. Sharp, P. Storck, and M. Rodriguez, "Knowledge is power: Efficiently integrating wind
energy and wind forecasts," IEEE Power Energy Mag., vol. 11,
no. 6, pp. 45-52, 2013.
J. Kleissl, "Current state of the art in solar forecasting,"
Final Report California Renewable Energy Forecasting, Resource Data and Mapping, Appendix A. California Renewable Energy Collaborative, 2010.
S. Pelland, J. Remund, J. Kleissl, T. Oozeki, and K. De Brabandere, "Photovoltaic and solar forecasting: State of the art,"
IEA PVPS Task 14, Subtask 3.1 Rep. IEA-PVPS T14-01: 2013.
Solar Electric Power Association. Predicting Solar Power Production: Irradiance Forecasting Models, Applications and Future Prospects, Washington, D.C.: Solar Electric Power Assoc., 2014.
K. Orwig, M. Ahlstrom, V. Banunarayanan, J. Sharp,
J. Wilczak, J. Freedman, S. Haupt, J. Cline, O. Bartholomy, H. Hamann, B. Hodge, C. Finley, D. Nakafuji, J. Peterson, D. Maggio, and M. Marquis, "Recent trends in
variable generation forecasting and its value to the power
system," IEEE Trans. Sustainable Energy, forthcoming.

Biographies
Aidan Tuohy is with the Electric Power Research Institute,
Chicago, Illinois.
John Zack is with AWS Truepower, LLC, Albany, New York.
Sue Ellen Haupt is with the National Center for Atmospheric Research, Boulder, Colorado.
Justin Sharp is with Sharply Focused, LLC, Portland,
Oregon.
Mark Ahlstrom is with WindLogics, St. Paul, Minnesota.
Skip Dise is with Clean Power Research, Napa, California.
Eric Grimit is with Vaisala, Seattle, Washington.
Corinna Möhrlen is with WEPROG, Ebberup, Denmark.
Matthias Lange is with energy & meteo systems, Oldenburg, Germany.
Mayte Garcia Casado is with Red Eléctrica de España,
Madrid, Spain.
Jon Black is with ISO New England, Holyoke, Massachusetts.
Melinda Marquis is with NOAA, Boulder, Colorado.
Craig Collier is with DNV-GL, San Diego, California.
p&e

ieee power & energy magazine

Table of Contents for the Digital Edition of IEEE Power & Energy Magazine - November/December 2015