Laserfiche WebLink
156 <br />152 <br />148 <br />144 <br />140 <br />+' 136 <br />R <br />D <br />~ 132 <br />'i 128 <br />a <br />`a~ 124 <br />0 <br />M 120 <br />,__ 116 <br />c <br />H 112 <br />`w <br />108 <br />00 104 <br />100 <br />96 <br />92 <br />88 <br />100 110 120 130 140 150 160 <br />Longmont 2 ESE Spring 32 Degree Frost Date <br />Figure 5 -Boulder vs. Longmont 2 ESE Scatterplot -Frost Dates <br />Based on this assessment, an alternate approach was developed using monthly linear regression <br />with nearby climate stations to fill daily minimum temperature data. Annual frost dates were <br />then calculated from the filled minimum daily temperature data using June 30 as the last spring <br />frost date and July 1 as the first fall frost date. As described further in the Recommendations <br />section, we recommend that an algorithm be added to TSTooI to allow calculation of frost dates <br />from daily data. <br />3. Fill Missing Monthly Climate Data <br />3.1 Temperature and Precipitation <br />The first step in filling missing monthly temperature and precipitation data was to select an <br />independent station to fill the missing data. Close proximity and the highest correlation <br />coefficients (R) using monthly linear regression were the main criteria for selecting an <br />independent station. Preference was also given to an independent station with historical data that <br />provided a complete record for the dependent station (as opposed to first filling the independent <br />station and then using it to fill the dependent station). Table 3 and Table 4 provide the <br />temperature and precipitation regression analysis results, respectively. If the dependent station <br />was filled with an independent station that had already been filled, then the independent station <br />name is followed by the word "filled". <br />Monthly linear regression generally produced high correlations for most of the dependent <br />stations being filled with temperature data. Correlations were not as high for precipitation data <br />filling. It is important to note that if a dependent station had only a few months to fill, the <br />independent station was selected based on the R-value for only the months being filled. This <br />r' <br /> <br />i <br />i <br />'. <br />~. <br />' <br />• ~i <br />i ~~ <br />. ~~' . <br />~` <br />i~ <br />less <br /> <br />. . <br />. <br />i <br />,s <br />•~ <br />•~ . <br />i <br />i <br />i . <br />~~ . <br /> - - Estimated trendline excluding outlying data <br />® Trendline including all data <br />Task 53_2.doc 10 of 17 <br />