Central to solving the climate change problem is the design of sophisticated observational means and model systems. The integration of reliable measurements over long periods, at high resolution, including a wide range of variables and on a global scale, is a huge worldwide effort. It is paramount to the assessment of climate change. It is also the key for our deeper understanding, for model validation and for harmonizing the evaluation process from data analysis, to simulations, all the way to reliable assessments for decision-makers and socioeconomic benefits to humanity. In this talk, we will summarize key climate processes and address the approach to Earth-Atmosphere Observing System and its application to verify and reduce uncertainties in climate models. Adequacy of climate simulations for specific purposes in the emergent field of climate services will also be discussed, thus bringing the opportunity to illustrate how tools from philosophy of science could help climate scientists perform certain methodological choices.