Figure 1.
Figure 2.
The analysis performed here proved to be a useful exercise. The comparison of ARBCON and National Weather Service (NWS) precipitation datasets over the 23-year period from 1977 through 1999 showed good agreement between the two networks. There were significant differences at the daily scale, due primarily to a variation in reporting times and procedures. These problems were mitigated significantly, however, at the monthly level, while the best agreement was found at the annual and multi-year levels.
There are four variables that appear to have had the largest effect on the overall comparison of the networks. In what we feel are descending order of importance they are: observer reporting errors, gauge separation distance, different network reporting procedures, and rain gauge collection efficiency differences. While gauge separation and network reporting procedures are relatively easy to identify, the others are not. This makes the comparison more difficult, as our suspected largest source of error is not quantifiable with the data in hand. While human error is no doubt a significant issue, the benefits outweigh the detriments as fully automated networks on these scales would be financially prohibitive. In addition, there was no indication that ARBCON contained more reporting errors than that of the NWS dataset. On the contrary, it appears that the primary reason for ARBCON totals being higher than NWS reports is that some NWS reports showed zero amounts when ARBCON observers reported significant precipitation. We believe that this may be a result of "missing" days being reported as zero values in some NWS reports.
Monthly and annual comparisons demonstrated good correlation with R = 0.994 and R=0.996 respectively. ARBCON volunteers, on average, reported approximately five percent more precipitation than those of the NWS with an overall ratio of 1.048. Mean multi-year single observer reports are 12.46 inches for ARBCON sites and 11.89 inches for NWS locations, thus yielding a mean difference per pair of 0.57 inches per year with a standard deviation of 0.44 inches. Based on the results of our study we conclude that the ARBCON compares favorably to the "NWS standard" and should be viewed as a reliable dataset.
