Laserfiche WebLink
The following process was used to for fill missing temperature and precipitation climate data: <br />1. Fill missing data using monthly linear regression in TSTooI with historical data from <br />another climate station(s) located in close proximity that produces reasonable regression <br />results. <br />2. Fill the remaining missing data using regression in TSTooI with another climate station <br />for which historical data has already been filled through regression. <br />Task 53 recommends the following process to fill missing annual frost date data: <br />1. Fill missing daily minimum temperature data using monthly linear regression in TSTooI <br />with historical data from another climate station(s) located in close proximity that <br />produce reasonable regression results. <br />2. Fill the remaining missing daily minimum temperature data using regression in TSTooI <br />with another climate station for which historical data has already been filled through <br />regression. <br />3. Calculate the annual frost dates from the regressed daily data using July 19th as the date <br />of separation between spring and fall frost dates. <br />3. Fill Missing Monthly Climate Data <br />Close proximity and the highest correlation coefficients (R) possible were the main criteria for <br />selecting an independent station to fill temperature and precipitation data. Preference was also <br />given to an independent station whose historical data alone provided a complete record for the <br />dependent station (as opposed to first filling the dependent station and then using it to fill the <br />independent station). Regression using an independent station selected on the above criteria was <br />performed first, providing a dataset with at least 95% filled data, although very few complete <br />station datasets. To fill the remaining 5% of data for stations in the Arkansas River Basin, the <br />stations were divided according to elevation. A complete station above 5000' was used as the <br />independent station to fill the remaining dependent stations in the upper basin and a complete <br />station below 5000' was used as the independent station to fill the remaining dependent stations <br />in the lower basin. Only if the complete independent station provided very low correlations <br />would another independent station be selected. Due to smaller elevation differentials between <br />stations in the Republican River Basin, one complete station was used to fill the remaining 5% of <br />data, regardless of elevation. <br />Table 4 and Table 5 provide the temperature and precipitation regression analysis results for the <br />Arkansas River Basin stations, respectively. Table 6 and Table 7 provide the regression <br />analysis results for the Republican River Basin stations. If the dependent station was filled using <br />regression with historical data, the table indicates the independent station used in the regression. <br />If necessary, a comma separates the independent station used in the second regression (listed <br />after the first station in the table) with a "filled" suffix. The Correlation Coefficient (R) listed <br />in the tables reflect the regression results from the first regression analysis. <br />Task 76 8 2.doc 7 of 17 <br />