SNOTEL (“Snow Telemetry”) is a major continuing program of the US Dept. of Agriculture (NRCS.USDA.GOV) which is a vital resource for the entire US west. By collecting snowfall data and other meteorological data from its network of over 800 automated sensing station, SNOTEL enables administrators in each state to allocate water in advance. It was a privilege for me to assist the renewal of the system in the 1990’s. This is from my resume:

SNOTEL: 2/1987 - 4/1995


This is a critically-important continuing project of the US Dept. of Agriculture for water supply forecasting in the Western USA.

He was initially involved in the upgrade of the SNOTEL (SNOwpack TELemetry) System. This system collects snowpack data and other climate data from a network of remote site stations scattered throughout the majority of the Western United States (nearly 700). These sites communicate their data via meteor burst technology using a low power radio transmitter/receiver. The data is first received by one of two unmanned radio stations or ÒMaster StationsÓ before being sent via X.25 to the Central Computer Facility in Portland, OR. Here it is made available to help in water supply forecasting for a number of agencies and other end users.

One of the most critical areas of this system is the reliable operation of the Master Station environment. 

“The successful implementation of this upgrade can be directly attributed to the many accomplishments and contributions of John Saunders.

In the early design phases of the project, John could always be counted to produce an accurate assessment of the requirements, followed by a thoughtful, technical approach to meeting those requirements. Most of the early requirements centered on the hardware interfaces for computer control of many of the components at the master station (i.e. selection and operation of the radio power amplifiers and control and monitoring of building power). John called upon his hardware engineering experience and expertise to design and build the prototypes for this control, but his experience is not limited to his thorough knowledge of hardware alone.

John was also very adept at writing the software required to acquire data from the many sensors installed at the master station. His grasp of object oriented C++ programming, in a sophisticated multi-threaded DOS compatible operating system environment proved formidable. In my fifteen years of experience in this field, I have never worked with anyone who has as complete a grasp of technical issues involving both hardware and software as John. His understanding is complete, his dedication to the project was exemplary, and his participation as a team member was unsurpassed.”

act of the two Master Stations has four receivers, one for each quadrant and two communication buffering computers. No available off-the-shelf equipment was available to connect all four receivers to both computers in such a manner that no single failure would disable more than on of these 8 links, including power. 

Here are my sketches for interface equipment to solve this problem. They were professionally made and used.

I got the contract to write the program for the computers on the assertion that object-oriented C++ was the way to go. This was new at the time. 

The messages come at unpredictable times, depending on meteors. The program creates an object for each message, and deletes it after it has been successfully sent to Portland and verified. 

This was a problem since at this time PCs had less than 1 Mb of RAM and had rudimentary memory management. Moreover, the messages varied in length. Inevitably, the PC’s memory became fragmented, and the program crashes.  A program function to give available free memory was not accurate. Therefore I cached frequently on the hard drive, and used a watchdog circuit to detect a crash and reboot the computer.

On reboot, the cashe would be read.