The Earth's climate is changing, but the rate of change and scale/timing of its impacts are far from understood and remain the subject of much debate. Variances of up to 5 °C can be seen between climate forecast models making it difficult for policy makers to take the necessary measures to mitigate and adapt to a warming planet. Global observations of the Earth by satellites are the only way to provide the date necessary to improve the fidelity of the predictions and to test and constrain the models. These data must be collected over decades to allow small trends to be sufficiently aggregated so that they can be reliably detected from a background of natural variability. This places very high demands on the performance of satellite sensors since no single satellite will have the lifetime necessary to attempt to monitor change and thus the accuracy needs to be sufficient to remove any prospect of bias from instrument drift propagating into the long-time base measurements which will necessarily be derived from a combination of satellites. At present few satellites are designed specifically for climate and none with the ability to demonstrate robust SI traceability in orbit at the uncertainty levels necessary (close to those of primary standards in the laboratory). This paper describes some of the scientific challenges and culminates in a proposed satellite mission designed explicitly to address them, and in so doing also has the ability to upgrade the rest of the Earth Observing system through reference cross-calibration.