Earlier this year, the Unidata Users Committee asked members of the Unidata community to participate in a survey regarding their use of scientific software packages, software training, and community services, and to favor us with their insights into possible future directions for the program. The survey ran through early February, and collected responses from 261 individuals from 25 countries. Both the Users Committee and the Unidata Program Center staff thank those who completed the survey for taking the time to participate and for providing such thoughtful feedback.
The survey comprised 107 questions on a variety of topics (although the survey was structured to allow respondents to skip over sections in which they had little interest or expertise). Those interested in perusing the full results can view them here:
https://www.surveymonkey.com/results/SM-SLT9V5PJ/
The Unidata Users Committee surveys the Unidata community roughly every five years. This article will attempt to provide a high-level summary of the 2016 survey results, along with some comparison with the previous survey and a discussion of the uses to which we intend to put the new information.
Statistical Highlights
The current and previous community surveys did not use the same questions and categories, so a longitudinal analysis is not feasible. It is, however, possible to make rough comparisons of the answers to a number of questions. While the total number of survey respondents did rise slightly to 261 from 240 in the previous survey, the absolute number of responses is too small for rigorous statistical analysis.
- The percentage of of respondents who identified themselves as “faculty” stayed roughly the same as in the previous survey, at just under 20 percent. Similarly, the number who identified themselves as “professional staff” was relatively constant at around 30 percent. The number identifying themselves as students (undergraduate, graduate, or post-doctoral) more than doubled, to roughly 27 percent. It's difficult to discern the factors involved in this increased student involvement with Unidata, but some signs point to interest in Python as a motivator of students and early career scientists.
- As in the previous survey, the large majority of respondents identified their area of study as atmospheric science. The rough distribution of respondents across the fields listed stayed the same, with a small but noticable uptick in the number of hydrologists providing comments (13 indicated involvement in this field, as compared with only 2 respondents in the previous survey).
- Overall, satisfaction with the activities and direction of the Unidata program remains very high with 94 percent of respondents saying they were “Satisfied” or “Highly Satisfied” with the Unidata program. (For comparison, 95 percent said they were “Satisfied” or “Highly Satisfied” with the program in the previous survey.)
Topical Responses
The bulk of the survey dealt with specific questions about software packages, data distribution, and Unidata community activities. The following are high-level summaries of the responses to several of these topical question groups.
AWIPS & GEMPAK
The National Weather Service's AWIPS II software was not really on the Unidata communty's radar at the time of the prevous community survey, but this year 36 respondents reported having used the software, more than half in the past week. Roughly half of the 36 respondents who have used AWIPS II are running their own Environmental Data EXchange (EDEX) servers, others are connecting to National Weather Service EDEX servers or to the experimental cloud-based EDEX server run by Unidata.
GEMPAK remains a popular software package for visualization; more than twice as many respondents identified themselves as users of GEMPAK than of AWIPS, despite long-standing suggestions that AWIPS will replace GEMPAK. However, many GEMPAK users did state that they consider the package to be old technology and said they were considering replacements including AWIPS, Python, and the IDV.
IDV
The Integrated Data Viewer continues to be Unidata's most popular visualization package, although by a smaller margin than in the previous survey. While the IDV has traditionally been popular as a teaching tool, nearly 70 percent of those who said they use the package said they use it as a research tool.
Python
The Python programming language has exploded in popularity among geoscientists since the previous survey (which did not address it specifically). More than 60 percent of those answering questions about Python said they use the language in their scientific work, although only a small percentage had tried Unidata's Python packages MetPy or Siphon. An overwhelming majority of respondents (99 percent) encouraged Unidata to continue creating Python resources geared toward the geoscience community.
netCDF
The Network Common Data Form (netCDF) is far and away Unidata's most widely used techology, both within Unidata's traditional atmospheric science community and beyond. 87 percent of the survey respondents reported using netCDF in some way; more than half said they had written or used software that makes use of the netCDF libraries in the past week.
Data Distribution
Making geoscience data available to universities via push and pull mechanisms is at the core of Unidata's mission. On the push side, in which data are delivered to receiving sites on the internet in near real time, roughly half of the respondents said they are using Unidata's Local Data Manager (LDM) to request data. Roughly one quarter said that their institutions participate in the Internet Data Distribution (IDD) network. Interestingly, a large percentage of respondents (between 20-40 percent) were unsure whether their institutions used the LDM or participated in the IDD. This suggests that for many community members, the mechanisms by which data are made available are less important than the fact that they are available.
On the pull side, in which data are requested from specialized data servers on an as-needed basis, more than 30 percent of respondents said they provide data on a THREDDS Data Server (TDS), and 35 percent said they routinely accessed data on a TDS. Again, a large number of respondents reported being unsure about whether they provided or accessed data using a TDS.
Cloud Computing
Taking advantage of features of the emerging cloud-based computing environment is a major focus of Unidata's current strategic plan. UPC staff have been working on a variety of cloud-computing projects, from helping store NEXRAD archives in the commercial Amazon cloud, to creating containerized versions of Unidata applications, to crafting techniques for running legacy software applications in a container environment for display in a web browser. These projects are in their beginning stages, and only a few of the survey respondents had experience with Unidata's efforts. Roughly half, however, said they perceived a “push” within their organization to use more cloud-computing resources.
Next Steps
Both the Unidata Users and Strategic Advisory Committees have seen and discussed the survey at a high level. Members of the Users Committee have agreed to review the detailed comments more thoroughly, and generate a list of topics or ideas that appear to have strong community support or interest. The Strategic Advisory committee will be able to use the Users Committee analysis as fodder for discussions this summer that will initiate the process of developing Unidata's next five-year strategic plan.
Unidata Program Center staff have also discussed the survey results, and while we'll still be working within the framework of Unidata's current strategic plan, we'll be looking for opportunities to enhance our activities in the areas of Python development and training, cloud computing, and generally working to make these newer technologies easier for Unidata community members to use.
Software developers at the Unidata Program Center will also take the survey comments into account when planning future development of the various Unidata software packages. Individual development groups will provide responses to issues raised by survey respondents in future News@Unidata articles.
