Observation is the only way for us to know the real characteristics of the environment. Only after we have obtained certain observations can theoretical study and numerical simulation begin to deepening our understanding and produce reliable results. So observation provides the physical basis for theoretical study and numerical simulation. At the same time, because of the complexity of space environment, there are many phenomena and processes in it. In many cases, these phenomena and processes are unpredictable before we observe them. So observation is the best way to find new phenomena and physical processes in space.
Because of the excellent work of observational scientists, we have been able to use the limited available observations to obtain what we believe to be quite a good view of the space environment, at least for the near earth space region. But here it is better for us to recall the difficulties that observationalist have met because of the limitations of observation. In many cases, we need in situ observations(ground observations are another very important way for us to observe the space, but here, for simplicity, we only consider the limitation of in situ observation), but the vastness of the space environment makes it impossible for our satellites to have a coverage of the whole space. Temporal coverage becomes another problem because of the limitation of the spacecraft life. So, generally we have to constrain our observation to some small regions and within certain periods of time. Even if we were able to provide more and more satellites with very high quality, one problem is insolvable. There are some regions in space that are totally unreachable by in situ observations, such as low corona region, the interior of the Earth and inside of the Sun. Though observations from ground and space outside of these regions can give us very valuable information about these regions, the lack of in situ observations is a serious problem for our understanding. In these cases, theoretical study and numerical simulation become extremely important. Further, we can not suppose all our observations are correct. In fact sampling limitations and measurement errors can result in ambiguous interpretation of the observations, which leads to the misunderstanding of the real physical process. Another major restriction of observations is that high expense and complex operations restrict in situ observations that can be made, though with time this limitation is becoming weaker.