Hydra Home Background of the HYDRA project

1. Introduction
2. Historical records
3. Exposure corrections
4. Geographic interpolation of wind speed data
5. Extreme wind statistics
6. Wind fields

1.  Introduction

In 1983 Wieringa and Rijkoort's assessment of the Dutch wind climate was published. This book has become a standard work for the Dutch wind engineering community. The assessment was mainly based on wind speed records measured in the sixties and seventies. Both land and sea stations were used in the analysis, however, the transition in the coastal zone was excluded.
    The National Institute for Coastal and Marine Management (RIKZ) and the Institute for Inland Water Management and Waste Water Treatment (RIZA) in the Netherlands are legally obliged to redo their risk assessment of the Dutch dike systems regularly. On the IJsselmeer, the Waddenzee and parts of the Zeeland waters wind is the main source of waves. This urged the need for an updated wind climate assessment, including the water-land (and vice versa) transition zones.
    At KNMI a project was started to obtain such an update. The initial focus of this project, called HYDRA, is on extreme wind speeds. However, the project yields important spin-off for the Dutch wind energy community. The project was started in May 1998 and will run for three years.

2.  Historical records

For the 1983-assessment wind speed records with a length of almost 20 years or longer were available from about 17 sites in the Netherlands. In total 400 years of data were available from these stations (~24 years per station on average). Among other factors a major revision of the observation network in 1971 cause these records to be highly heterogeneous in observational methods. For instance, only 10-min. averages once an hour are available before 1971, after 1971 also hourly averages are available.
    For the present assessment, considering only observations done after 1971, 20 stations have records longer than 20 years making up a total of 500 years (~25 years per station). An overview of the available records is given here. Although in the period after 1971 less rigorous changes in observational methods have taken place compared to the period before 1971, changes in the station's environment still necessitate exposure corrections.

3.  Exposure corrections

In their 1983-assessment Wieringa and Rijkoort derived local roughness lengths and exposure corrections from gustiness analysis. From the roughness length the wind speed at a reference level (60 m) can be computed using the logarithmic wind profile. This 60-m wind can in turn be used to compute the wind over a hypothetical measuring site. When for this hypothetical site measuring height 10 m is used, and roughness length 0.03 m (WMO requirements), the resulting wind speed is called the ''potential wind'' speed. The ratio between the potential wind speed and the measured wind speed is the exposure correction factor.
    Gusts have been measured together with the average wind speed from 1971 onwards and environmental changes will be reflected by changes in the measured gustiness. For gustiness analysis information on the inertia of the anemometer and recorder system through the coarse of time is necessary. Wieringa's (1976) gustiness model can only be used for analogue recorders with continuous reading. Therefore in the present assessment the gustiness model of Beljaars (1987) will be used. Possible differences between the two models were assessed by Verkaik (2000a).

4.  Geographic interpolation of wind speed data

To estimate the wind climate over regions where few wind speed stations are available various methods will be used. The internal boundary layer model by Kudryavtsev & Makin (1996) has been tested especially for the lake "IJsselmeer" (Verkaik, 1999; Kudryavtsev et al., 2000). However, the use of internal boundary layer models over extensive regions with many roughness transitions proved cumbersome. Moreover, a slightly modified version of the regionalization methods by Wieringa (1986) performed better and is much easier to apply (Verkaik, 2000b).

5.  Extreme wind statistics

Return periods for extreme wind speeds are evaluated in the context of the Rijkoort-Weibull model (Rijkoort, 1983; Smits, 2001a), but in the past several methods have been studied. The one that is very popular and frequently been used in applications is the Gumbel distribution, which is Type I of the three Fisher-Tippett distributions. These distributions make use of the maxima of windspeeds in predefined periods (mostly years), which can be plotted into a so called "Gumbel plot". See for example, the Gumbel plot of Schiphol Airport for the period 1951-2000. Using the Gumbel distribution results in a straight line (the middle one of the dotted lines in the plot). The top line is curved like the Fisher-Tippett type II (bounded at the lower end) and the bottom line is curved like the Fisher-Tippett type III (bounded at the upper end). The next step is to determine which one of the three types is most likely. In literature, several estimates of the way of curvature are discussed. In this case the Fisher-Tippett type III seems to have the best fit.
    An advantage of these distributions is that the uncertainty in estimations of the parameters and windspeeds can be calculated precisely. An important disadvantage is the lost of information, because only the maximum values are used. However, there are several years with more than one severe storm. Another disadvantage is the aim of samples of independent and identically distributed random variables. When using more information by shortening the length of a period (say one month instead of one year) this is not the case, because storms are mostly season dependent (in the Netherlands, the chance a storm occurs in December is much greater than in June). So, shortening the length of a period will result in biased parameters. Due to the necessarily of throwing away important information, other methods that make use of more information will be studied.
    One of these methods concerns a modification of the Rijkoort-Weibull model, described in Smits (2001b). The most important modifications within this model are the implementation of a threshold when estimating the Weibull parameters and the introduction of a persistence correction based on physical grounds.
    Another method models storm maxima instead of all hourly mean wind speed values, described in Smits (2001c). Two distributions have been used to model the storm maxima. These concern the conditional Weibull distribution and the generalized Pareto distribution.

6.  Wind fields

RIKZ and RIZA both use wave and water levels that are strongly dependent on the wind field as input variable. This means that not only point statistics of wind speed is of importance, but also the statistics of wind fields. At the moment, a simplified approach of wind fields is applied at RIKZ and RIZA with regard to variation of the wind speed and direction in time and space that is not accurate physically. To obtain an overview of the possibilities to improve this approach, a literature study has been performed (Ettema, 2002). The outcomes of this study can be used for future research to develop an approach that fulfils to the demands of RIKZ and RIZA.


  • Beljaars, A. C. M., 1987: The influence of sampling and filtering on measured wind gusts. Journal of Atmospheric and Oceanic Technology 4, 613-626.
  • Ettema, J., 2002: Analysis of wind fields for wind climate assessment of the Netherlands. Royal Netherlands Meteorological Institute, Climatological Services.
  • Kudryavtsev, V. N.,  and V. K. Makin, 1996: Transformation of wind in the coastal zone. KNMI Scientific Report WR 96-04, Royal Netherlands Meteorological Institute.
  • Rijkoort, P. J., 1983: A compound Weibull model for the description of surface wind velocity distributions. KNMI Scientific Report WR 83-13, Royal Netherlands Meteorological Institute.
  • Smits, A., 2001a: Analysis of the Rijkoort-Weibull model. KNMI Technical Report TR-232, Royal Netherlands Meteorological Institute.
  • Smits, A., 2001b: Estimation of extreme return levels of wind speed: a modification of the Rijkoort-Weibull model. Royal Netherlands Meteorological Institute, Climatological Services.
  • Smits, A., 2001c: Estimation of extreme return levels of wind speed: an analysis of storm maxima. Royal Netherlands Meteorological Institute, Climatological Services.
  • Verkaik, J. W., 1999: 1-D model voor interne grenslagen voor water en land. Royal Netherlands Meteorological Institute, Climatological Services.
  • Verkaik, J. W., 2000a: Evaluation of two gustiness models for exposure correction calculations. Journal of Applied Meteorology 39, 1613-1626.
  • Verkaik, J. W., 2000b: Windmodellering in het KNMI-HYDRA project -- opties en knelpunten. Royal Netherlands Meteorological Institute, Climatological Services.
  • Wieringa, J., 1976: An objective exposure correction method for average wind speeds measured at a sheltered location. Quarterly Journal of the Royal Meteorological Society 102, 241-253.
  • Wieringa, J., and P. J. Rijkoort, 1983: Windklimaat van Nederland, Staatsuitgeverij, Den Haag, the Netherlands.
  • Wieringa, J., 1986: Roughness-dependent geographical interpolation of surface wind speed averages. Quarterly Journal of the Royal Meteorological Society 112, 867-889.