Link to USGS home page.
Scientific Investigations Report 2005-5291

Evaluation of Precipitation Estimates from PRISM for the 1961-90 and 1971-2000 Data Sets, Nevada

By Anne E. Jeton, Sharon A. Watkins, Thomas J. Lopes, and Justin Huntington

Version 2.0

ABSTRACT

The Oregon Climate Service estimated precipitation in Nevada for two 30-year periods, 1961-90 and 1971-2000, using the computer program, Parameter-elevation Regressions on Independent Slopes Model, or PRISM. The accuracy of PRISM estimates for Nevada has not been evaluated statewide relative to known observation points. The intent of the PRISM model is not to specifically match simulated precipitation amounts to recorded station data, but rather to model large-scale orographic processes. This allows for interpolation to locations beyond the recorded data set assuming those locations have similar physiographic and climatic properties. PRISM provides spatially-estimated averages of long-term precipitation that sometimes are not well correlated to known measured points. The known measuring points do not have associated accuracy ratings, and the PRISM model does not include prediction error relative to the station record. Consequently, for this report a difference of ± 15 percent relative to the recorded data was selected as an acceptable range of error or difference. Two data sets were compiled and used in this study to compare recorded precipitation to PRISM precipitation estimates for the 1961-90 and the 1971-2000 periods, and to the National Weather Service normals and the Western Regional Climate Center averages. The National Weather Service normals, however, are considered to be the most accurate reporting of the 30-year means for both the 1961-90 and the 1971-2000 periods. For the earlier and latter periods respectively, 89 and 85 percent of PRISM estimates were within the acceptable range of error for the National Weather Service data set and 63 and 70 percent were within the acceptable range of error for the Western Regional Climate Center data. However, the percent differences are higher when comparing PRISM estimates to recorded data from some of the high-elevation SNOpack TELemtry (SNOTEL) sites and to sites in the 5,500 to 7,500 feet elevation zone.

A statistical difference does not exist at the 95 percent confidence level between the two 30-year National Weather Service means, nor between the 1961-90 National Weather Service and the Western Regional Climate Center averages, hence these two data sets are considered interchangeable. There are, however, statistically significant differences between all combinations of the recorded data sets and the PRISM estimates. In general, the percent differences between the PRISM estimates and the National Weather Service 30-year normals have the narrowest interquartile ranges and the narrowest overall ranges.

Statistical tests using three elevation data sets indicate no significant difference between the U.S. Geological Survey National Elevation Data 30-meter digital elevation model and the PRISM 4-km digital elevation model; however, there are differences between the actual station elevations and the digital elevation model data sets. In general, the larger the elevation difference, the greater the precipitation difference. The largest absolute differences for both the National Weather Service and Western Regional Climate Center data sets coincide with elevation differences of 1,000 feet or more. The spread in differences increases noticeably between the 4,500 feet-8,000 feet range for both National Weather Service and Western Regional Climate Center data. Some of the individual sites with a difference exceeding 25 percent have location or elevation input errors.

There appears to be little correlation between precipitation estimate differences and latitude and longitude.

Mean precipitation volumes using the Hardman precipitation zone map for each of the 232 hydrographic areas in Nevada were compared to the PRISM precipitation volumes for both of the 30-year periods. PRISM estimates of the amount of precipitation for hydrographic areas in the mid-section of the State were as much as 37 percent lower than estimates made using the Hardman map. For the northeastern and some areas in the southeastern part of the State, PRISM estimates are within 5 percent of the Hardman estimates. For much of the State outside the central region, PRISM estimates are from 6 to 155 percent greater than precipitation estimates made using the Hardman map. Typically, PRISM estimates correlate best with Hardman volumes for those hydrographic areas with elevations in the 4,000?9,000 feet range.

In comparing the two 30-year periods climatologically, the 1961?90 period has a mix of negative and positive Pacific Decadal Oscillations, a decadal-scale pattern of climate variability that may be more representative for long-term analyses. The distribution of differences between the PRISM data and the recorded data may be the result of overinterpolation between the limited data points, input error in location and station elevation, and possibly the weight given the hypothesis that precipitation is linearly related to elevation. The precipitation-elevation relations used in PRISM may not be appropriate for all mountainous areas, particularly for leeward slopes. The coarseness of the PRISM grid cells, coupled to the sparseness of the long-term precipitation data and the broad range of differences between PRISM estimates and the recorded data, suggest the optimum use for this data set is regional.

Contents

Abstract

Introduction

Purpose and Scope

Acknowledgments

Precipitation Patterns in Nevada

The PRISM Model

Methods for Comparing the PRISM Model to Recorded Data

Climate Station Averages

PRISM and Climate Station Averages

Station Elevations and PRISM Digital Elevation Model Elevation Grid Values

PRISM and Hardman Precipitation Estimates by Hydrographic Areas

Results

1961–90 Comparisons

1971–2000 Comparisons

Comparison of Hardman and PRISM Precipitation Estimates by Hydrographic Area

Discussion of Results

Summary and Conclusions

References Cited

Figures

Figure 1. Precipitation estimated using PRISM model, location of PRISM control points and
climate stations in Nevada

Figure 2. National Weather Service long-term annual precipitation for selected climate stations in Nevada: (A) Austin, (B) Boulder City, (C) Elko, (D) Fallon, (E) Mina, (F) McGill, (G) Lovelock, (H)
Searchlight, and (I) Winnemucca

Figure 3. Percent difference and absolute differences between PRISM estimated
precipitation and National Weather Service and Western Regional Climate Center
recorded data for 1961-90 and 1971-2000 periods

Figure 4. Scatter plot of relation between recorded precipitation for National Weather
Service and Western Regional Climate Center sites and percent difference (A)
and absolute difference (B) between PRISM estimated precipitation and the
recorded precipitation for the 1961-90 period

Figure 5. Scatter plot of relation between actual station elevation and percent difference
(A) and absolute difference (B) between PRISM estimated precipitation and the
recorded precipitation for the 1961-90 period

Figure 6. Percent difference between estimated PRISM precipitation values and National
Weather Service recorded data for the 1961-90 period

Figure 7. Percent difference between estimated PRISM precipitation values and the
Western Regional Climate Center recorded data for the 1961-90 period

Figure 8. Scatter plot of relation between recorded precipitation for National Weather
Service, Western Regional Climate Center, and SNOTEL monitoring
sites and percent difference (A) and absolute difference (B) between PRISM
precipitation and the recorded precipitation for the 1971-2000 period

Figure 9. Scatter plot of relation between actual station elevation and percent difference (A)
and absolute difference (B) between PRISM estimated precipitation and the
recorded precipitation for National Weather Service, Western Regional Climate
Center, and SNOTEL monitoring sites for the 1971-2000 period

Figure 10. Percent difference between estimated PRISM precipitation values and National
Weather Service recorded data for the 1971-2000 period

Figure 11. Percent difference between estimated PRISM precipitation values and the
Western Regional Climate Center and Natural Resources Conservation Service
SNOTEL recorded data for the 1971-2000 period

Figure 12. Percent difference between estimates of average annual precipitation in hydrographic areas using PRISM precipitation values and Hardman estimates for the 1961-90 period

Figure 13. Scatter plots showing relation between Hardman estimates of annual
precipitation calculated for hydrographic areas in Nevada and PRISM estimates
for 1961–90 and 1971–2000

Table

Table 1. Summary statistics of percent difference between the PRISM precipitation data
and the recorded data sets for the 1961-90 and 1971-2000 periods

Appendixes

Appendixes are included as separate PDF files.

Appendix 1. Climate stations used to evaluate PRISM precipitation estimates for the State of Nevada. Data obtained the Western Regional Climate Center (includes the National Weather Service data) and the Natural Resources Conservation Service SNOTEL network

Appendix 2. Recorded and PRISM estimated precipitation at climate stations in Nevada. Data obtained from WRCC and NRCS

Appendix 3. Actual and estimated elevations for climate stations and associated percent differences

Appendix 4. Percent differences1 between PRISM estimates and National Weather Service normals, and Western Regional Climate Center, and Natural Resources Conservation Service averages for individual climate stations

Appendix 5. Estimates of annual precipitation for hydrographic areas in Nevada based on PRISM and Hardman maps, and associated percent differences

This report is contained in the following files:

For viewing and printing upon download.
(This version of the report is accessible as defined in Section 508 of the Rehabilitation Act of 1973, Amendments of 1998.)


Adobe Reader To view files in PDF format, download free copy of Adobe Reader.

*Downloading Suggestion: It is best to download and save a large PDF file to your hard drive rather than to open it inside your browser.

   A standard click may automatically open the PDF file inside the browser but doing so will result in a very slow load.
   Downloading the PDF file may take several moments but will be worth the wait.
   Once it is downloaded, open the PDF from your hard drive using Adobe Acrobat—it will open in a fraction of the time it would take to open the PDF over the Internet.

Assistance for PC and Mac Operating Systems

Microsoft Windows and Explorer or Netscape:

1. Right-click the link to a file, and then choose 'Save (Link) Target As' from the pop-up menu.
2. In the Save As dialog box, select a location on your hard drive, and then click Save.

Apple OS and Firefox, Explorer or Safari:

1. Right-click (or Control-click) the link to a file, and then choose 'Download Link to Disk' from the pop-up menu.
2. In the Save dialog box, select a location on your hard drive, and then click Save.



Send questions or comments about this report to the Public Information Assistant, GS-W-NVpublic-info@usgs.gov, 775-887-7649.
For more information about USGS activities in Nevada, visit the Nevada Water Science Center home page.


Accessibility FOIA Privacy Policies and Notices

Take Pride in America home page. FirstGov button U.S. Department of the Interior | U.S. Geological Survey
Persistent URL: https://pubs.water.usgs.gov/sir20055291
Page Contact Information: USGS Publishing Network
Page Last Modified: Thursday, 01-Dec-2016 18:53:10 EST