SYSTEMATIC BIAS IN UTAH SNOW DATA

The United State Department of Agriculture, Natural Resources Conservation Service (NRCS) currently operates two Snow Water Equivalent SWE data collection systems: manual snow courses and a telemetered system called SNOTEL which use distinctly different measurement techniques. All climatological data have biases that may impact their applications outside of the original intent if not considered. Analysis shows SWE data is very useful for their intended purpose predicting current streamflow via hydrologic models by using the most recent SWE data (20 to 40 years), representing current watershed conditions.

This research focuses on two issues which have had significant impact on data stability in Utah: vegetation and physical change at the manual sites and SWE sensor changes in the SNOTEL system. It was recognized very early on that snow accumulation at these sites would systematically change over time. “Forest cover is a significant factor in snow accumulation and melt at a snow course. Therefore, any change in forest cover may gradually affect the readings obtained over a period of years”, NRCS National Engineering Handbook, Section 22, 1972. Snow course analysis in Utah shows dramatically altered conditions at some sites and little to no impact at others. Each site with vegetative or physical change regardless of elevation or latitude has decreased SWE (30 year average) whereas at sites with no vegetative change, SWE remains steady indicating that vegetation impacts, not climate change may be the source of the observed decreasing SWE.

The SNOTEL system initially used stainless steel pillows as a SWE sensor. Later, hypalon pillows became the sensor of choice. Data comparison of dual steel/hypalon pillow sites shows that hypalon pillows begin SWE accumulation later, melt out earlier, and accumulate 8% to 25% less SWE than the steel pillows. Thus this data set has a systematic bias of declining SWE and early melt due to sensor change that could be confused for and/or mask a climate change signal if not properly analyzed.

In making long term comparative analyses from the Snow Course data set, systematic bias should be quantified and removed from each individual site in order to give an accurate depiction of change due to climate. In the SNOTEL data set, sensor change bias should be quantified and removed.