U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations
|This report is an archived publication and may contain dated technical, contact, and link information|
Publication Number: FHWA-RD-03-092
Date: May 2006
Verification of LTPP Virtual Weather Stations Phase I Report: Accuracy and Reliability of Virtual Weather Stations
Task 3: Determine Accuracy of Vws Estimates and Verify The Vws Algorithm (Cont.)
The same approach used for SMP and AWS sites was used here to determine the precision and bias of estimates for 5347 locations throughout the United States. For this purpose, the climatic data for all cooperative weather stations from the NCDC between 1994 and 1996 (1096 days) were used. For every weather station, the five closest weather stations were determined and the estimated daily weather data were calculated by simply averaging data for the five closest weather stations, without using distance as a weight or eliminating weather stations with a high elevation difference. The estimated data were then compared with the measured data at the site.
Table 32 shows the mean and standard deviation error for minimum and maximum temperatures by elevation. The results in table 32 indicate that the mean error (estimate bias) increases with elevation. This is because elevation differences between the weather stations are more probable at higher elevations. The lowest mean error was for elevations of less than 250 m; however, the standard deviation did not significantly differ. Below this elevation, the overall standard deviation was 2.1 °C for maximum and 1.9 °C for minimum temperatures. Almost 40 percent of the weather stations in the United States are located below a 250-m elevation. Figures 34 through 37 show the distribution of the mean and standard deviation of the errors for maximum and minimum temperatures.
The overall precision and bias of maximum (2.2 and 0.2 °C, respectively) and minimum (2.0 and 0.1 °C, respectively) temperatures are comparable to the results from the AWS and SMP sections. Since the estimated bias increases with the elevation difference, the bias can be remedied by correcting the temperature for the elevation difference.
Figure 34. Bar chart. Distribution of mean error for maximum temperature (NCDC data).
Figure 35. Bar chart. Distribution of mean error for minimum temperature (NCDC data).
Figure 36. Bar chart. Distribution of standard deviation of error for maximum temperature (NCDC data).
Figure 37. Bar chart. Distribution of standard deviation of error for minimum temperature (NCDC data).
In this part of the study, maximum temperatures were corrected for elevation differences and the results of the corrected and uncorrected errors were compared. The LTPP AWS and SMP databases were used for the study and the maximum temperature was corrected using the following algorithm derived from the analysis of the elevation differences in table 26:
Table 33 includes the mean and standard deviation of error for maximum temperature estimates using a simple average (corrected and uncorrected). Table 34 includes the same data for SMP sections. In both cases, correcting the maximum temperature dramatically reduced the bias, while the standard deviation did not change significantly.
Estimated climatic data were developed for NCDC weather station sites and were then compared to the measured NCDC data. Climatic estimates were developed by averaging data for the five closest weather stations for each NCDC site without using a weight. Table 35 includes the overall average and standard deviation of error (measured minus estimated) for the daily, monthly, and yearly climatic estimates of 5347 NCDC sites throughout the United States. The data period covers 1994 through 1996 and included 1096 days of data for each site. The following are some observations from the results:
These statistics were calculated using all observations and are overall values. The overall values are generally higher than the average per section values, especially for accumulated data such as precipitation and FI. This is caused by the effect of bias on the standard deviation (locations with a high bias will contribute more to the standard deviation).
|Previous | Table of Contents | Next|
Topics: research, infrastructure, pavements and materials
Keywords: research, infrastructure, pavements and materials,Pavements, weather stations, LTPP, climatic data, temperature, precipitation.
TRT Terms: research, facilities, transportation, highway facilities, roads, parts of roads, pavements