Content area
Abstract
There exists a strong, well documented linear relationship between network averaged sustained (or mean) wind and gust speeds within the San Diego Gas and Electric (SDGE) mesonet that is characterized by the slope of the linear regression between the two known as the network average gust factor (GF). The network average GF potentially contains information on the average obstruction of the stations within the network and when paired with numerical model output, may be used for gust prediction. Using the network average GF for gust prediction can also result in the correction of biases due to poorly resolved terrain or unresolved obstructions. This thesis explores the presence of strong network average GFs in a collection of diverse observation networks, different methods for estimating network GF, and the influence of potential measurement and environmental controls on network average GF.
Sustained wind speed and gust data from the Air Research Laboratory Field Research Division (ARLFRD), Dugway Proving Grounds (DPG), Delaware Earth Observing System (DEOS), and SDGE mesonets were taken for the year of 2015 from NCEP MADIS for calculations and comparison of network average GF. Additional data obtained to aide in the assessment of controls on network average GF include wind measurements from select regions of ASOS stations (both METAR filtered and raw one-minute) and the RAWS network, wind measurements from HPWREN, and temperature measurements from the ARLFRD.
Using these data each network was found to have a strong and persistent network average GF each unique from each other. Estimates of these GFs were found to be most accurate when using a zero-intercept linear regression model, arithmetic averaging between stations, and only incorporating observations time when nearly all potential stations within a network report in the calculation. By using these methods when calculating network average GF the physicality of the estimate is ensured, the shape of the GF distribution preserved, and scatter along the regression line reduced.
Two main groups of influences on network average GF were assessed; measurement techniques and environmental variables. The former category consists of wind reporting rules, averaging interval, sampling rate, and anemometer mounting height. It was found that poor or restrictive reporting rules have the potential to compromise estimates of the network average GF; while a specific ratio between averaging interval and sampling rate is needed for proper comparison of GFs between stations or networks. Assessment of anemometer mounting height revealed the vertical structure of network average GF to be logarithmic or exponential in nature.
The latter category of controls on network average GF include stability, sustained wind speed, and surface roughness. Overall, variations of network average GF with sustained wind speed outweighed those induced solely from stability or surface roughness; however, this may be due to the dependency of sustained wind speed itself on stability and surface roughness. Because of this, variations in network average GF due to sustained wind speed was used to develop a framework to expand gust prediction using a GF to include a range of potential gusts.
