It was discussed in prospect of PhD research proposal previously, but this question is for all ages of researchers.
- What factors should one take into consideration before selecting a research field?
Are you passionate about it?
Are you good at it?
Will people pay you to do it?
Everything else is noise.
Some points for STEM:
What are worldwide current emerging technologies? Which topics become attention by scholars, industry and funding institutions in journals.
Funding: What topics (like green energy in Germany) are on the local research agenda of your country for the next decades?
Fundamental research: where are contradictions of widely-accepted but incomplete theories in a field. Are the major paradigms consistent? (e.g. bringing together general relativity and particle physics). See my answer here on open problems
Meterology improvement/measurement methodology. Often very tricky, time- and cost-intensive engineering problems, but often making falsification/discovering of new phenomenons possible at all, opportunity to develop patents
Are there enough possible and qualified students, scholars for collaboration in a special (new) research field you can hire/attract? Of course, someone has to make the first step, but in STEM money and manpower is often the crucial factor to compete with other research groups, otherwise it's easy to waste a lot of money and time while someone already published the results you are still trying to measure. Without good students and a well-equipped lab it will be hard to attract money, students and scholars. In Germany you see this in form of "clustering" of several research institutions interested in semiconductors and chip-technology in German silicon valley near the city of Dresden.
Can you and your group dive easily into a new or related research topic? Do you have the infrastructure, is the gap of knowledge too big? How big is the risk, if you fail in this topic or it becomes suddenly un-trendy/unimportant. That's why research groups often do some diversification of research interests. Don't put all your money on a single stock.
a really good nose, overview and knowledge of current state of the art and knowledge.