Data Collection and Quality Control
The standard daily observation of total liquid precipitation (rain or melted snow) is performed at 0700 L. This observation is submitted to the CoCoRaHS Web site or phoned into the CoCoRaHS office. Precipitation maps are automatically created and updated throughout the day to include observations that are reported late. The system also accepts multiple day observations.
In addition to the daily precipitation observation, volunteers can also submit reports of intense rain or hail on the CoCoRaHS Web site. The intense rain report queries the observer for an estimate of the total rainfall over a specified time (e.g., one hour) and any indications of flooding.
The hail and intense rain reports are sent directly from the CoCoRaHS server to the Local Data Acquisition and Dissemination server at the NWS office in Boulder. If criteria such as a certain hail size or rainfall rate are met, the report is reformatted and sent to the appropriate NWS offices. NWS workstations can be set to alarm on receipt of this report, ensuring that the warning forecaster sees the data immediately. These reports are used for nowcasting and for the issuance and verification of flood and severe thunderstorm warnings (described below). As CoCoRaHS continues to expand, the data will be sent to other NWS offices in real-time.
In addition to daily maps, a variety of data reports are produced that allow quick and easy analysis of frequency, magnitudes, and areal extent of rain, snow, and hail. Because data quality is a paramount concern, we’ve taken a number of steps to ensure data accuracy and consistency.
All volunteers are strongly encouraged to attend a two-hour training course. This course covers placement of rain gauges and hail pads for maximizing gauge catch efficiency and practice measuring precipitation from a variety of storm types. Participants are also given instructions on how to report daily and extreme event data, and common errors and how to avoid them. We also provide written instructions to volunteers, including detailed information on snow measuring techniques. Training materials are posted on the CoCoRaHS Web site. We are developing a training video to enhance the training process.
Data Entry Quality Control.
Training is not a sufficient guarantee of data quality. Automated checks during Web-based data entry sometime identify and disallow incorrect station numbers, dates, time, or unrealistic precipitation values. Meanwhile, experienced CoCoRaHS personnel process phone messages submitted by volunteers and correct detectable errors before entering observations into the database.
Currently, there is no automated data quality control performed after data have been entered into the system. One or more CoCoRaHS volunteers check precipitation maps each day to visually identify potential errors. Sometimes errors and their causes are obvious and corrections are submitted immediately. More often, however, observers are asked to confirm their data entry. As CoCoRaHS continues to expand, methods such as “PrecipVal” (Urzen et al. 2004) could be adopted for CoCoRaHS spatial data checks in order to decrease random errors. PrecipVal utilizes independent data sources to increase confidence in a particular gauge estimate (Urzen et al. 2005). Typically, we ask volunteers to check their own station data to verify accuracy with their personal records (usually at the end of each water year – September 30).
To ensure accurate station locations (for analysis, etc.), CoCoRaHS staff use commercial mapping software and hand-held Global Positioning System (GPS) units to determine latitude, longitude, and elevation for each CoCoRaHS station.