Comparing Wind Data at Local Meteorological Stations and in Forested Areas Using Roughness Length and Topographic Exposure Indices

This study evaluates whether wind data at local meteorological stations can be used for studying wind damage in forested areas. Data from four Automated Meteorological Data Acquisition System (AMeDAS) stations and one weather station, located in the northern and southern portions of Hokkaido Island, was compared with wind data collected at flux towers located in the Teshio and Tomakomai regional forests. Anemometers were affixed to the flux towers at a height of 32 m in Teshio and 40 m in Tomakomai; thus, the data was assumed to adequately represent wind conditions in a forested area without vegetation effects. Mean wind speed at the AMeDAS stations and weather station sites was converted to extreme wind speed using roughness length on eight directions. This comparison was based on the storm (October to May) and typhoon (August and September) seasons. Subsequently, a distance-limited topographic exposure index (TOPEX) was calculated for all anemometer positions using 1, 2, 5, and 10 km distances from the position. We found that wind at sites with low TOPEX scores and small roughness length could significantly represent wind conditions in nearby forested areas. Wind speeds with high TOPEX scores but small roughness length occasionally showed correlation with conditions at the flux tower. Although additional study is required, Received September 28, 2009; Accepted December 12, 2009 136 Kamimura, K. & Saito, S. using TOPEX with roughness length could be beneficial for deriving suitable wind data sets that correlate with wind conditions in nearby forested areas.


Abstract:
This study evaluates whether wind data at local meteorological stations can be used for studying wind damage in forested areas. Data from four Automated Meteorological Data Acquisition System (AMeDAS) stations and one weather station, located in the northern and southern portions of Hokkaido Island, was compared with wind data collected at flux towers located in the Teshio and Tomakomai regional forests. Anemometers were affixed to the flux towers at a height of 32 m in Teshio and 40 m in Tomakomai; thus, the data was assumed to adequately represent wind conditions in a forested area without vegetation effects. Mean wind speed at the AMeDAS stations and weather station sites was converted to extreme wind speed using roughness length on eight directions. This comparison was based on the storm (October to May) and typhoon (August and September) seasons. Subsequently, a distance-limited topographic exposure index (TOPEX) was calculated for all anemometer positions using 1, 2, 5, and 10 km distances from the position. We found that wind at sites with low TOPEX scores and small roughness length could significantly represent wind conditions in nearby forested areas. Wind speeds with high TOPEX scores but small roughness length occasionally showed correlation with conditions at the flux tower. Although additional study is required, using TOPEX with roughness length could be beneficial for deriving suitable wind data sets that correlate with wind conditions in nearby forested areas.

Introduction
Wind damage is critical for sustainable forest management due to an increasing number of mature forests and climate condition change (Kuboyama andOka, 2000, Kamimura et al., 2008). There are ongoing efforts to reduce wind damage risk using statistical, empirical, or mechanistic procedures including terrain and/or wind information (Gardiner et al., 2000, Mitchell et al., 2001 (Suárez et al., 1999). TOPEX is a simple indicator to show how much the target point is exposed by wind, which is the most critical factor to influence windthrow (Ruel et al., 2002). These models are useful for estimating the average wind climate on flat or complex terrain, but not for particular wind conditions such as typhoons and subtropical cyclones.
Wind data in Japan is generally available at local meteorological stations including Automated Meteorological Data Acquisition System (AMeDAS) and weather stations managed by the Japan Meteorological Agency. Specifically, AMeDAS is located approximately every 17 km on Japanese land (Kondo et al., 1991) in or near urban areas.
There are several advantages to use these data including easily accessible though internet. In addition, the data at local meteorological stations located near forests are often used in wind damage studies to find out the wind condition at occurring wind damage. However, it remains unclear whether this data accurately represents wind climate in nearby forests. It is also difficult to certify whether it is certainly representing wind climate in the forests. This is because wind varies depending on several factors including terrain, neighbouring conditions, and anemometer height. Therefore, this study aims to find a simple index or procedure to identify whether wind data at local meteorological stations suitably represents the wind conditions in nearby forested areas.

Methods
Two study sites were selected, one in the Teshio region located in  Table 1 shows the description of the towers, AMeDAS, and weather station.
In this study, wind data at the flux tower was assumed to correctly observe wind climate in the forests. To avoid vegetation and canopy effects, wind speed and direction measured at 32 m height on the Teshio tower and those at 42 m height on the Tomakomai tower were used.
Since height and periods of data collection varied among the anemometers, wind data were calibrated using the following steps. First, the data were divided to two seasons. In Japan, strong wind is generally caused by typhoons or extratropical storms. The highest frequency of strong wind is often observed in winter or in spring, while catastrophic damage is mainly due to typhoon although the frequency is low (Hatakeyama, 1966). Therefore, the data were classified into two seasonal groups; storm (October to May) and typhoon (August to September). All wind data were calculated as the hourly mean speed and degree azimuth, which were then classified to eight directions (N, NE, E, SE, S, SW, W, NW). Subsequently, standard deviation (s.d.) of the mean wind speed was calculated for each season and direction. Using s.d.
extreme wind speed (U extreme ; m/s) was converted as: where U mean is the mean wind speed (m/s) and "3" is a constant, which can be determined on average due to empirical studies (Kondo, 2000).
Wind speed data at the AMeDAS and the meteorological station were calibrated to compare the data at the towers because the anemometer height of the meteorological station and the AMeDAS varies depending on terrain conditions and/ building height located near stations (Kondo et al., 1991). The speed data were calculated to be the wind speed at the same height as the anemometer heights of the flux tower (32 and 42 m) using the following equation (Kondo, 2000).
where U t is the wind speed at the anemometer height on the tower (m/s), U a is the wind speed at the anemometer of AMeDAS or the weather station (m/s), z t is the anemometer height of the tower (m), z a is the anemometer height of AMeDAS or weather station (m), and z 0 is the roughness length (m). Unique values of z 0 in eight directions provided by Kondo et al. (1991) were applied for this calculation (Tab.
2).  Next, distance-limited TOPEX of eight directions was calculated for 1, 2, 5, and 10 km from the anemometer position using 50 m × 50 m of digital elevation map. The TOPEX score is a parameter representing exposure from wind by measuring the angle from the skyline including the target point to the point on the highest elevation in the eight directions within the limited distance (Wilson, 1984, Quine and White, 1994, Mitchell et al., 2008. The negative scores are counted as zero. The calculation was mainly based on ArcGIS ver. 9.2 (ESRI Co.).
Finally, Pearson's correlation coefficient was obtained for the wind scores were compared to find relationship between wind speed measurements and terrain exposure by wind. Table 3 shows the correlation of the extreme wind speed between the flux towers and the AMeDAS and weather station. In storm season, U extreme at Nakagawa AMeDAS, in particular from the SW, was significantly correlated to that at the flux Tower (Teshio). U extreme from the E and NW showed a relatively significant relationship. In typhoon season U extreme from the SW and W at Nakagawa AMeDAS seemed to be correlated to that at the tower. In terms of Nakatonbetsu AMeDAS, U extreme from the N was most significant during storm season, while U extreme from the SE showed the best correlation during typhoon season. At both AMeDAS, U extreme from the SW was often observed.

North region
Total TOPEX scores at all anemometer positions increased more than twice between 1 and 10 km distance (Tab. 4).
More specifically, the TOPEX scores of the Nakagawa AMeDAS were approximately twice and those of the Nakatonbetsu AMeDAS two to four times greater than those of the flux tower. For the Nakagawa AMeDAS, there were no shelters in the southern side (E to W) within 2 km from the anemometer (Fig. 2). In addition, strong wind was observed particularly from the SW partly due to low shelter effects. At the Nakatonbetsu AMeDAS, almost all directions were sheltered within    wind speed from NE (see Fig. 2 and Tab. 3). Wind from the N shows the best correlation in storm season despite of a shelter within 1 km from the AMeDAS. At the Teshio tower, the TOPEX scores also increasing, and they were close to those of the Nakatonbetsu AMeDAS within 5km. Wind from the SE also showed better correlation than the wind from oth er directions. Although the TOPEX score of the SE was always high, the roughness length was the smallest among those in other directions. This intended that wind turbulence was stable compared to the wind from other direction (Kondo et al., 1991).

South region
At the Shikotsukohan AMeDAS, a significant relationship with the tower was found in the N during two seasons. On the other hand,  Figure 3 shows more detailed terrain condition using the TOPEX scores of eight directions. Few differences were found between the Tomakomai weather station and the tower in all directions. There is also a lake in the NW of the Shikotsukohan AMeDAS. This exposed area could lead to better correlation of wind from the W, NW, and N in both seasons except wind from the NW during typhoon season.
To sum up, wind with low TOPEX scores and low roughness length tended to suitably represent U extreme in forested areas located within 25 km from the local meteorological sites. Wind with high TOPEX scores but low roughness length also showed better correlation to the wind in the forests, but only if there were no high shelters located near anemometer positions, i.e. low TOPEX scores within 1 to 2 km distance were required (See Tab. 2, Fig. 2 and 3).

Discussion and Conclusion
Wind data at the local meteorological stations were compared with that at the flux towers located in the forested areas to determine whether the data were statistically related to the wind condition in the nearby forests. We found that wind data at the sites with low TOPEX scores and small roughness length correlated to the wind in the nearby forested areas. This finding allows wind speed data at different anemometers to be integrated in order to obtain better data-sets of wind conditions in forested areas. In addition, our study clarifies that distance from forested areas is not always the key to finding anemometers associated with wind conditions in forested areas. However, several disagreements were also found using the relationship between TOPEX and wind climate, perhaps caused by the neighbouring environment of the anemometer and limitations of TOPEX.
There are several problems with measuring wind, especially at AMeDAS. For instance, the AMeDAS location or anemometer height occasionally changes due to changes in neighbouring environment. For instance, AMeDAS tends to be moved to the other place if it suffers from any interruption due to construction of new buildings or change of tree height. In addition, since most weather stations and AMeDAS were generally located in the centre of town with various anemometer heights, it is difficult to observe surface wind flow (Mitsuda, 1997).
In this study, we calculated wind speed at the same height of the tower anemometer using roughness length provided by Kondo et al. (1991). However, roughness length might change due to new buildings constructed around the anemometers. Some of the anemometer heights have been also changed. Thus, it is necessary to find information on neighbouring environments before using wind climate data for nearby forests.
Distance-limited TOPEX provides a useful overview of wind exposure for site terrain conditions, which one of the keys of wind condition.
TOPEX is also easily calculated using GIS. With these advantages, TOPEX can be applied to predict local wind estimation by using airflow models such as the mesoscale numerical weather prediction model (Mitchell et al., 2008). However, our study shows that TOPEX cannot be the only indicator to select wind data for nearby forested area. This is because TOPEX itself does not account for environmental conditions such as roughness length. Sites with less exposure are more likely to receive less extreme wind. On the other hand, if the site has large roughness length but low mean wind speed, then extreme wind speed tends to be accelerated caused by strong turbulence during unstable airflow condition (Kondo and Kuwagata, 1984). In addition, Turnipseed et al. (2003) points out that forest canopy is more critical than terrain complexity. Airflow above canopy is also strongly associated with stem density (Kondo and Yamazawa, 1983). Therefore, additional information, in particular, roughness length is required to improve wind data selection using TOPEX.
It is challenging to find appropriate wind data for estimating wind conditions in forested area. Nevertheless, using a simple procedure or index would be beneficial. Further studies are required to develop procedures for finding suitable wind data sets at local meteorological stations associated with wind climate in forested areas.