10B.3
WIND CLIMATOLOGY ISSUES, AND THE DEVELOPMENT OF A
COMPREHENSIVE WIND DATA BASE FOR WIND EROSION ESTIMATES
Greg Johnson *
USDA-NRCS
National Water and Climate Center
Portland, Oregon
1. INTRODUCTION
Wind erosion is a significant issue affecting many agricultural regions of the United States. The USDA-NRCS is charged with assessing the potential wind erosion impacts on private lands arising from the adoption of a variety of conservation measures. There is a need for these assessments to be more quantified than in the past, and the tool that will be used to make these assessments is the Wind Erosion Prediction System (WEPS; Hagen, 1991). This modeling system uses either actual or generated weather time series as input. The WINDGEN stochastic wind model, based on a Weibull distribution fit, delivers daily (with disaggregation capability to hourly or finer resolution) values of wind speed and direction in WEPS (Skidmore and Tatarko, 1990). The current WINDGEN parameter database was developed from the 1986 Wind Energy Resource Information System (WERIS; Elliott et al. 1986). Significant spatial variabilities in WEPS-computed soil erosion have been speculated to be caused by large, and perhaps erroneous, differences between various WERIS wind station parameters.
The are several purposes for this preliminary study. First, an examination of the performance of WINDGEN compared to historical observations was sought in a small region where wind erosion estimates were spatially inconsistent. Second, greater insight into spatial and temporal variabilities of observed winds, in a region of high potential erosivity, was sought. Third, relationships between hourly, daily and monthly winds were investigated for potential large-scale mapping of soil erosion by wind. Finally, lessons learned from this study would hopefully guide future efforts to map wind and wind parameters over the entire United States, providing for the accurate generation of wind and weather time series at a fairly fine (2-4 km) horizontal resolution.
2. APPROACH
Observed and generated wind speeds were assessed at three sites of interest in the wind-erosion-prone region of the Columbia Basin in eastern Oregon
______
* Corresponding author address: Greg Johnson, USDA-NRCS, National Water and Climate 101 SW Main St., Suite 1600, Portland OR 97204-3224; email:
and Washington—Pendleton, OR, and Spokane and Yakima, WA. The data included observed hourly, daily and monthly values from the NCDC TD-3280 (Surface Airways) data set, and archived on compact disc by EarthInfo, Inc. Data were available for the period of record, but were examined for years 1964 to 1998. Generated data were obtained from the WINDGEN model structure and parameter sets for these three sites. For internal comparison, data also were extracted from the National Climatic Data Center’s (NCDC) WERIS database, from which WINDGEN parameters were derived (Elliott et al. 1986). For some additional analyses, shown in the appendices, more locations in this study region were examined.
WINDGEN values were derived using the WINDGEN parameters for these sites, and the published procedures used for computing mean daily, and then disaggregated hourly values. In addition, 30 years of daily data were generated for each location using the WINDGEN parameters, and for some analyses these 30 year datasets were used for comparison to WINDGEN values derived from monthly parameter sets.
WERIS-derived values were obtained by examining published WERIS summarized statistics for these sites.
To reduce the volume of analyses and reported statistics, only the primary erosion months of February to June, and September to November, were examined in this study.
3. OBSERVED AND GENERATED WIND CHARACTERISTICS
3.1 Observed Monthly Wind Comparisons
Observed, average monthly wind speeds (over all years) were very similar between the three sites (Figure 1). In all months, highest average wind speeds were observed at Spokane, followed closely by Pendleton and Yakima. Differences were largest in the most winter-like months (Feb. and Nov.). In those months, the average wind speed was approximately 1.5 m/s higher at Spokane than at Yakima. In May, June and September the difference was less than 1 m/s. Wind speed averages ranged from 2.6 m/s at Yakima in November to 4.6 m/s at Spokane in April.
For comparing erosive potential, the third power of these average wind speeds in April (the windiest month) was 55.3 m3/s3 at Yakima, 73.6 m3/s3 at Pendleton, and 96.7 m3/s3 at Spokane. Thus, erosive power from the monthly mean wind speed was approximately 1.75 times greater at Spokane than Yakima.
3.2Hourly Erosive Winds: Generated and Observed Winds Exceeding 8 m/s
A thorough study of winds, both observed and generated, was conducted using a critical erosive threshold of 8 m/s. Although mean daily winds were slightly higher at Spokane throughout the year, observed hours with winds over 8 m/s were much greater at Pendleton in all months, except for the winter months of February and November (Figure 2). For instance, in April, nearly 16% of all hours have winds over 8 m/s at Pendleton, while less than 10 % are this windy at Spokane, and just over 7% at Yakima. Similar differences were noted for March, May and June.
Differences between WINDGEN hourly percentages and those reported in the WERIS database were generally less than 2% at Pendleton, but were as much as 3.5% in some months at Yakima. There may have been errors in deriving WINDGEN parameters from the WERIS database, or the internal structure of WINDGEN may not always accurately replicate the source statistics.
WINDGEN, in general, faithfully reproduced the frequency of winds exceeding 8 m/s. At Pendleton (Figure 3), differences were almost all less than 0.5%. Differences were slightly greater at Yakima in the fall, but still less than 1% in most months. It thus appears that WINDGEN can be used with some assurance to estimate hours with high, erosive winds.
A 30 year time series realization was generated using WINDGEN for each location. Monthly percentages of hours exceeding 8 m/s based on these 30 year realizations were not significantly different from those derived from WINDGEN theoretical equations.
3.3 Hourly Erosive Winds: Generated and Observed Winds Exceeding 15 m/s
It was of interest to see if the results found analyzing 8 m/s threshold wind speeds were consistent with even stronger winds. 15 m/s was chosen as the threshold wind speed. Observed hours with sustained winds above this level were rare, with a maximum of 0.6% of hours in March at Pendleton (Figure 4). In some months, stations never observed even one hour with winds above 15 m/s. Pendleton had the most hours with winds above this threshold in all months except November, and had 2 to 3 times as many hours with these strong winds as the other locations. WINDGEN replicated these rare events quite accurately, indicating that the tails of the wind distribution seemed adequately represented (WINDGEN specifies separate Weibull scale and shape parameters for each location for each month and cardinal wind direction).
4. ESTIMATING HOURLY EROSIVE WINDS FROM DAILY AND MONTHLY STATISTICS
For mapping and other purposes it was of interest to study the relationship between hourly erosive winds and daily and monthly mean winds. A threshold of 8 m/s was used. Examining daily and hourly data for ten Februarys at Pendleton, a significant nonlinear relationship was found (Figure 5). Greatest variability was found at intermediate wind speeds (4.5 to 6.5 m/s). However, mean daily winds do appear to be useful guides for estimating hours of erosive winds. Consistency of this relationship between stations and regions should be examined.
With even greater implications for producing maps and other products of total hours of erosivity, significant relationships were found between mean monthly wind speeds and total hours with winds >= 8 m/s at all locations. An example of all months combined at Pendleton (Figure 6) reveals this relationship, with an r2 = 0.90. Similar graphs were produced for Spokane and Yakima, and the coefficients were quite similar between stations. Thus, there was preliminary support for mapping hours of erosive winds from mean monthly winds, and applying regional relationships or gradients of coefficients to produce very useful spatial products.
- CONCLUSIONS
Erosive winds can be estimated by daily and monthly winds, and spatial patterns can be identified to aid in wind climate mapping. In a small region, locations with highest mean winds do not always have the greatest number of hours of erosive winds. The WINDGEN wind generator, in general, faithfully reproduces the mean and extreme nature of winds.
6. REFERENCES
Elliott, D.L., C.G. Holladay, W. R. Barchet, H.P. Foote and W.F. Sandusky, 1986: Wind energy resource atlas of the United States. DOE Report CH/10093-4. 210 pp.
Hagen, L.J., 1991: A wind erosion prediction system to meet user needs. J. Soil and Water Conservation, 46:105-111.
Skidmore, E.L. and J. Tatarko, 1990: Stochastic wind simulation for wind erosion modeling. Trans. ASAE, 33:1893-1988.
Figure 1. Monthly mean wind speeds (m/s) at Pendleton, Spokane and Yakima.
Figure 2. Observed percentage of hours with wind speeds >= 8 m/s at the three study locations.
Figure 3. Percentage of time wind speed >= 8 m/s at Pendleton, by month, observed vs. WINDGEN.
Figure 4. Observed percentage of hours with wind speeds >= 15 m/s at the three study locations. Note difference between these percentages and the mean monthly wind speeds (Figure 1).
Figure 5. Mean daily wind speed (m/s) vs. no. of hours with winds >= 8 m/s at Pendleton, 10 Februarys
Figure 6. Mean monthly wind speed vs. no. of hours with winds >= 8 m/s at Pendleton, all study months for 16 years.