Comments on manuscript wes-2022-48     Submitted on 20 May 2022

"Predicting power ramps from joint distributions of future wind speeds"

by

Thomas Muschinski, Moritz N. Lang, Georg J. Mayr, Jakob W. Messner, Achim Zeileis, and Thorsten Simon

The paper describes an improved methodology for the probabilistic

forecast of hourly wind speeds on a forecast horizon of one day. Focus

of the work is the prediction of power ramps, i.e, the strong increase or

decrease of wind power over a time window of one or more hours. Basis

of such predictions is typically an ensemble of physics-based

numerical wheather predictions (NWP), which is the transformed into

wind power predictions using idealized wind turbine power curves.

The authors address the temporal multi-point correlation structure of

wind speeds and their forecasts as one crucial problem to improve

probabilistic forecasts. Together with existing methods they introduce

their own approach to explicitly model the joint multivariate

distributions of hourly wind speeds with respect to their mutual

temporal dependencies. This new approach of Multivariate Gaussian

Regression (MGR) has previously been published in a journal on

statistics and econometry, and is consequently described only shortly.

The main part of the paper performs a detailed comparison of the

various methods at the example of one power ramp event measured at the

German FINO 1 platform in 2019. The proposed method of MGR outperforms

the other appraoches in the comparison.

The paper addresses a highly relevant problem in wind power

forecasting, namely the so-called power ramps. Moreover, with the

temporal multi-point correlation structure of wind speeds the authors

address one of the central and most demanding challenges of the field

and of atmospheric flows in general. Their approach is promising and

the given example is convincing.

Technically, the paper is well written and well readable. The

structure is clear and comprehensive. English language style is

fluent, precise, and, as far as I can say, correct. Results are

presented clearly and with very appropriate graphics. References are

given whereever necessary.

The reviewer is an expert neither in NWP nor in the advanced

mathematical approaches of the paper. However, in my eyes this paper

makes an important contribution, and it is almost ready for

publication.

General remarks:

The demonstration of the proposed method using just one single example

of a power ramp is quite limited. However, given the length of the

paper of already 20 pages, more examples do not seem to make sense.

Could the authors comment on the performance of the method for more

examples, or, elaborate on possibilities of a wider evaluation?

Specific remarks:

P. 3 L. 85: The bi-linear interpolation between grid points is

assumably widely used and probably also accepted. However, it is knwon

to reduce fluctuation amplitudes. It would be helpful to have any

estimate to what extent that effect is present for the given case.

Technical remarks:

P. 3 L. 77: The phrase "but observations generally far from zero" does

not seem to make sense. Please double-check.

P. 17 L. 355: Inserting a "that" after "ensure" would be helpful,

although (to my understanding) not strictly necessary.

The manuscript, “Predicting power ramps from joint distributions of future wind speeds” describes a novel statistical postprocessing method called the multivariate Gaussian regression (MGR) to provide calibrated and sharp hourly ECMWF 100-m wind speed forecasts for lead times +24h to +47h. Observations from three offshore meteorological towers are used for verification. The main advantages to the new method are that it can explicitly model temporal dependencies across multiple lead times and is not limited to the number of members available in the raw ensemble, unlike the various other multivariate postprocessing methods discussed in the paper and compared against in the figures. Skillful modeling of these joint distributions across lead times is essential for reliable power ramp predictions and consequently the balancing of a power grid with wind energy resources. The authors found that the new method outperformed the various other methods commonly used in the literature when assessing the scores and skill scores of wind speed and wind power forecasts. The paper is overall well-written, interesting, and valuable to the wind energy field. Here are my recommendations for further clarity.

Scientific and clarification comments:

L48: I’m not sure what was meant by the phrase, “and not supported by the…..” Can you please reword for clarification?

L59 and throughout: the use of higher resolution may be confusing to readers in this context as higher resolution when referring to NWP forecasts typically means higher temporal or spatial resolution for numerical integration. Here, it seems to mean more ensemble members. Can you replace with another term or distinguish for the reader the meaning in this paper’s context?

L83-87: Please mention somewhere in here that the forecasts are initialized at 00Z. It’s mentioned later, but I found myself asking that question in this section.

Sect 3: Wind speeds can only take on positive values. Why not use the truncated Gaussian distribution truncated at zero rather than the untruncated Gaussian distribution?

Equations and equation explanations in the text: Some of the equations and explanations of variables in the text reuse symbols which I found difficult to keep track of while reading. For example, i and j referred to lead time and next lead time, respectively in some earlier places in the paper while i referred to ensemble member in other parts (e.g., equation 5) and a new variable, m, became the new lead time variable. The definition of N also changed as well. For better readability, please make the variable symbols unique and consistent throughout the paper.

L253: Can you please provide more details about what the DSS score tells the reader about the forecast? Does it inform about the performance of the calibration and sharpness, just one, or some other aspect of the forecast?

L295: Can you please state why the COP(Err) was selected as the reference method for wind speed as opposed to the other methods?

Figure 1: How were the 50 members out of 1000 selected? Would the gray lines show comparable spread or much more spread if the full 1000 members were shown?

Figure 2: The caption says that the data were aggregated by month and year. Is it just showing December data like Figure 1 or all months? If all months, do you expect there to be differences in the scores, and thus optimal method, if all months were not aggregated?

L323: Can you please state why the ECC method was selected as the reference method for power ramps as opposed to the other methods?

Technical comments:

L148: Spelling of seasonally.

L160L Spelling of postprocessed individual

Eq(5): Please define symbol

L 208: I think you can remove ‘on’ before ‘become’

292: Spelling of smooth

Figures 1 & 4: Please put the corresponding forecast hour along the horizontal axis or state the forecast hour range of the figure within the caption.