Figure 4. Differences between ASOS and HO-83 daily average temperatures (ASOS minus HO-83), in degrees Fahrenheit, for the period Oct 1993 through Mar 1994.
The largest difference in average temperature (delta-T = TASOS - THO83) was -6°F, which occurred on February 27. Overall, ASOS daily average temperatures were .67°F colder than those measured by the standard HO-83 (Table 1). The "largest" monthly average temperature difference, occurred in February, during which delta-T was -1.06°F (Table 1). This month also showed the highest standard deviation of the differences between ASOS and the HO-83.
TABLE 1. Temperature differences (in degrees Fahrenheit) and standard deviations of the differences between ASOS and the standard HO-83 thermometer from Oct 1993 through Mar 1994.
This same tendency was reflected in the daily maximum and minimum temperature measurements as well. The daily maximum temperatures showed the biggest difference, with ASOS averaging .85°F colder than the standard readings (Table 1). Figure 5 shows that the ASOS temperature was equal or below that of the HO-83 on all but one occasion, and this was by only 1°F. The resulting standard deviation of the difference between measurements showed that the maximum temperature had the greatest variance of the three values (Table 1). The greatest difference in maximum temperatures occurred on February 27, 1994, in which the ASOS maximum temperature was 8°F colder than the HO-83. Overall, February also showed the largest difference and the highest standard deviation of the six months.
Figure 5. Differences between ASOS and HO-83 daily maximum temperatures (ASOS minus HO-83), in degrees Fahrenheit, for the period Oct 1993 through Mar 1994.
The differences in minimum temperatures (delta-T) averaged -.49°F through the period studied. Figure 6 shows that the difference in minimum temperatures was zero or negative on most occasions, with positive values on seven days. February was again the month with the largest differences. In this month, ASOS averaged -9°F colder than the HO-83 (Table 1). The standard deviation was also highest in February.
Figure 6. Differences between ASOS and HO-83 daily minimum temperatures (ASOS minus HO-83), in degrees Fahrenheit, for the period Oct 1993 through Mar 1994.
The trend of more significant temperature differences between ASOS and the standard equipment during January and February may be because these were especially cold months. January, with an average temperature of 14.1°F, was 6.8°F colder than normal. February was 4.8°F colder than the monthly normal of 18.1°F. Table 2 shows how the differences in high, low, and average temperatures varied by temperature regimes. The differences (delta-T) in average and minimum temperatures indicated that ASOS tended to deviate the most from the standard HO-83 when temperatures were on the "cold" end of the scale. This was especially true for temperatures colder than 9°F. The 23 occasions during which minimum temperatures were in the ranges from -30°F to 0°F had average delta-T values between -1°F and -2°F. This may partially explain the large negative delta-T values seen in average and minimum temperatures during the coldest months of January and February (Table 1).
TABLE 2. Differences between ASOS and HO-83 temperatures in degrees Fahrenheit, stratified by temperature range.
However, this temperature regime bias was not as well indicated in the differences in maximum temperatures. Table 2 shows that the most significant maximum temperature differences (delta-T) were scattered across the full range of temperatures than were the average and minimum differences. The biggest differences occurred in the temperatures ranges of 20°F to 29°F and from 50°F to 59°F. However, since these differences were not obviously clustered at one end of the temperature range, as the average and minimum temperature differences were, it is difficult to make any conclusions about temperature regime biases in maximum temperature differences. To complicate matters further, the delta-T value for maximum temperatures was somewhat similar during the last three months, that consisted of the colder than normal months of January and February, and relatively warmer month of March. Table 1 shows that January was .90°F lower, while February was 1.06°F lower. March's delta-T of minus 1°F, even though it was not a particularly cold month. This suggests that factors, other than those related to the ambient temperature alone, were at work in influencing the maximum temperature differences between ASOS and the standard HO-83.
Several other factors, besides temperature regime alone, likely affected the differences between the ASOS and HO-83 ambient temperature measurements through the period of study. One factor may be the influence of the proximity of the ASOS ambient air sensor to the drainage ditch. Even before ASOS was installed, the effect of the drainage ditch on local temperatures had been noticed by NWS personnel at both WSO Lansing, and the state forecast office (WSFO Detroit/Pontiac) for several years.
Overnight low temperatures in Lansing are frequently cooler than surrounding reporting stations. This discontinuity can often be seen on an even smaller scale, when the HO-83 temperatures are compared to the backup maximum/minimum thermometers. These backup thermometers are located in an instrument shelter near the universal raingage, about 3500 feet southwest of the field equipment location (Figure 2). These temperature discontinuities are frequently blamed on the nearby drainage ditch and the unrepresentatively low elevation of the field sensors.
As described earlier, the distance between the ASOS and standard field equipment array is only about 75 feet, with ASOS no more than 1 or 2 feet lower in elevation. Conditions most favorable for the influence of the nearby drainage ditch to occur are when the lowest several meters of the atmosphere become stratified, with cool, damp air pooling in the drainage ditch and surrounding valley. This occurs most often under clear skies, especially at night, when a low level inversion develops and winds become light. This condition often leads to ground fog that has been observed to frequently develop in the field equipment area.
To see if the conditions favoring this temperature stratification correlate with the temperature differences, several scatter plots were made. Plots of the daily minutes of sunshine against the maximum, minimum, and average temperature differences between ASOS and standard HO-83 measurements were made to test for a correlation between clear or sunny skies and delta-T. These plots showed no significant correlation. Similar plots were created to examine a potential relationship between the 24-hour average sky cover from MAPSO and delta-T, but these were also inconclusive. A plot of daily average wind speeds against the various temperature differences did show some interesting trends.
The scatter plot in Figure 7, showing the 24-hour daily average wind versus differences in average temperature, indicates that the most significant average temperature differences occurred when the average winds were 11 miles per hour or less. However, on numerous other occasions, the temperature differences were zero or -1°F in the same average wind range. A plot of maximum temperature differences against the daily average wind showed a better correlation (Figure 8). With only two exceptions, all differences in maximum temperature of minus 2°F or less occurred with average wind speeds under 10 miles per hour. As with the average temperature plot, there were also numerous cases with winds in the same range with no temperature departure.
Figure 7. Differences in daily average temperatures (ASOS minus HO-83) in degrees Fahrenheit plotted against daily average wind speed in MPH. Daily average wind speed is the 24-hour average using MAPSO output from standard observing equipment.
Figure 8. Differences in daily maximum temperatures (ASOS minus HO-83) in degrees Fahrenheit plotted against daily average wind speed in MPH. Daily average wind speed is the 24-hour average using MAPSO output from standard observing equipment.
Of the 12 instances during which the differences were -3°F or less, eight of these were in the months of January and February, and two were in March. The other two days were in October and December. Of these cases, 11 occurred on days with average winds of 10 miles per hour or less. Often, it was noted that winds were typically light at the time that the daily maximum temperature was reached. On the days that the ASOS maximum temperature was 5°F or colder than the official HO-83, the wind speed for the three-hour period near the time of maximum temperature was from calm to about 5 miles per hour. On the day that ASOS was 8° colder than the HO-83, the wind was calm. Sky conditions varied on these days from sunny to cloudy and precipitation was recorded on some of these days at the time of the maximum temperature. No specific wind direction was favored, and nearly every direction was represented. Also, there was significant snow cover on the ground on most of these days.
A correlation was not as apparent in a plot of the minimum temperature differences against the daily average wind (Figure 9). In fact, the largest positive differences between the ASOS and standard equipment daily minimum temperatures occurred on a day with an average wind speed from six to eight miles per hour. A check of the three largest negative differences did, however, show that winds were calm around the time of the occurrence of the minimum temperatures on two of these occasions, and around 7 miles an hour on the other.
Figure 9. Differences in daily minimum temperatures (ASOS minus HO-83) in degrees Fahrenheit plotted against daily average wind speed in MPH. Daily average wind speed is the 24-hour average using MAPSO output from standard observing equipment.
The reason that the minimum temperatures were not as obviously associated with lower daily average winds is likely due to the way that daily average winds are computed. On a typical day, the minimum temperature occurs during a relatively short period, such as during the couple of hours near sunrise. The daily average wind value used consists of the average of 24 hourly observations and may not be a fair representation of the wind speeds for the short time in which the diurnal minimum temperature occurred. Many minimum temperature differences were likely associated with light winds at the time of occurrence, but this was likely masked by an unrepresentatively high average wind value for the day. The numerous plots with temperature differences occurring with lower daily average winds likely represent cases with light winds at the time of the minimum temperature occurrence. However, the fact that these differences were not significant for most of these cases indicates that there were other factors that prevented the minimum temperatures from being significantly different. To truly test the effect of wind speed on the temperature differences, the conditions at the times of the maximum and minimum temperatures would have to be noted for each day through the study and compared. However, this is beyond the scope of this study.
The differences in peak wind turned out somewhat more significant. Figure 11 shows that overall, peak winds were lower on ASOS, especially during the last half of the sampling period. However, ASOS did show a couple of peak wind readings that were significantly greater. On December 5, 1993, the ASOS value was 5 miles per hour greater than the standard, and on January 19, 1994, it was 10 miles per hour greater. Overall, the ASOS peak winds were .98 miles per hour less than the MAPSO output for the period of study. This is likely due to the five second smoothing of the wind data that occurs in ASOS. However, the different anemometer heights and equipment problems make it difficult to draw any significant conclusions.
Early in the study, an interesting trend developed that may have been associated with the failure of the wind equipment that occurred during the study period. On November 15, the difference in average winds went from almost exclusively positive, to negative, as seen in Figure 10. While, the differences in peak winds went from almost equally positive and negative to almost exclusively negative (Figure 11). The development of this trend may have signaled the beginning of the problem that ultimately resulted in the outage of the ASOS equipment, or less likely, a problem with the standard anemometer.
Figure 10. Differences in daily wind speed in miles per hour between ASOS and the standard anemometer (ASOS minus standard anemometer).
Figure 11. Differences in daily peak wind speed in miles per hour between ASOS and the standard anemometer (ASOS minus standard anemometer).
The comparison between the daily precipitation output from ASOS and the official measurement showed considerable differences. The ASOS tipping bucket tended to measure less liquid precipitation than the official weighing gage. This was especially true when measuring water equivalent of snowfall. Table 3 shows that through the 6-month study period, ASOS measured 2.55 inches less precipitation than the universal weighing raingage. The most significant difference occurred in February, when ASOS measured less than 1 percent of the total official liquid precipitation. Table 3 also shows that ASOS consistently measured less water equivalent of snowfall. The largest deviations from the weighing gage occurred during months with the greatest percentage of frozen precipitation, which were December, January, and February.
TABLE 3. Comparison of ASOS and Universal recording raingage precipitation measurements in inches for Oct 1993 through Mar 1994. Column three (from the left) indicates MAPSO-ASOS, while the values in parentheses are [(MAPSO-ASOS÷MAPSO]x100. Column four are the snow water equivalents measured by ASOS, and values in parentheses is (ASOS÷MAPSO)x100.
This can also be seen in Figure 12, where the departures for days with mostly frozen precipitation are indicated by wider bars. ASOS measured .01 inches or more precipitation than the standard weighing raingage on 10 of the days with precipitation through the study. These were mostly for rain events.
Figure 12. Differences in daily liquid precipitation amounts in inches between ASOS heated tipping bucket raingage and the standard universal raingage. Values were reported only for days in which either raingage measured a trace or more of liquid precipitation. Wide bars indicate days that most of the precipitation fell as frozen.
Although this study analyzed only six months of ASOS climatological output, some cursory conclusions can be drawn from the data. Of the differences noted, the under-reporting of precipitation during the winter was the most significant. Problems that result from the use of heated tipping buckets to measure snow have been well documented (Groisman and Legates 1994). Some fine adjustments of the thermostats may be needed to catch more of the snow without causing significant evaporation. The tipping bucket did perform reasonably well in measuring rainfall, with a tendency to underreport during peak rain events. This problem may be corrected by future adjustments in the algorithm.
The next most striking difference was the consistently lower temperatures reported by ASOS. This was although the hygrothermometer used by ASOS is essentially the same as the standard one currently in use. The differences were most significant in the daily maximum temperatures. The correlation between these temperature differences and the daily average wind speeds suggests that the slight difference in the environment between the sensors may be a factor. The most apparent difference in environment between the two equipment sites is the proximity of a nearby drainage ditch to the ASOS combined sensor group. This drainage ditch has been suspected to affect local temperatures, measured with the standard equipment, long before ASOS was installed. Wind speeds are likely a significant factor in the differences between ASOS and the standard equipment temperatures, but other weather factors may also be at play.
One possible solution to the consistently lower ASOS temperatures would be to relocate the ASOS combined sensor group to another location in the airfield away from the drainage ditch. More data are needed to confirm the temperature differences, especially to analyze the equipment's performance during the summer months. Closer analysis of weather conditions during the times of the largest temperature differences may also shed further insight on the source of the temperature discrepancies.
The short duration of ASOS wind data, caused by equipment problems and unreliability, made it impossible to come to any significant conclusions. Based on the methods of wind measurement, however, it can be hypothesized that ASOS will likely produce better representations of average wind speed. The ASOS software five second average wind sampling rate will provide much greater temporal resolution than the one minute samples taken each hour and used by MAPSO.
On the other hand, the reliance of five second averages to compute peak winds may result in lower peak wind values than current manual methods. This tendency was seen from the first two months of reliable peak wind output, in which the ASOS value averaged .55 miles an hour lower than that read from the wind recorder charts by observers. An adjustment in the software to scan one second peak wind values would possibly cure this problem and provide a truer representation gusty conditions.
Further study and comparisons of ASOS versus the standard climatological data are needed to confirm the observations of this study, especially through the summer season. If the trends seen in this comparison persist through a longer period, adjustments of the equipment, or sensor location will be required to preserve the integrity of the local climatological record.
The author thanks Dick Wagenmaker, and Mike Evans, both at WSFO Detroit/Pontiac, for their helpful comments and suggestions.
DOC, NOAA, NWS, 1984: Instruction Manual, Hygrothermometer (HO83), Volume 1, Support Volume. Engineering Division, 50 pp.
____________, ____________, OFC MSSR, 1988: Federal Meteorological Handbook No. 1-SURFACE OBSERVATIONS. Government Printing Office, (GSA), A10-4, April.
____________, FAA, and U.S. Navy, 1992: Automated Surface Observing System - USERS GUIDE, 1992. Government Printing Office (GSA), June 1992.
Groisman, P.Y., and D.R. Legates, 1994: The accuracy of United States precipitation data. Bull. Amer. Meteor. Soc., 75, 215-227.