Currently not possible for earthquakes but OK for volcanoes!

Knowledge of volcanic processes is incomplete, but there have been great strides made in forecasting eruptions. Various physical processes can be monitored for changes which can signal an impending eruption. The record of past eruptions is also used to help determine what and where the risks are highest. At the present time, only 20% of volcanoes are being monitored. As might be expected this is mainly in MEDCs such as Japan and the USA which have the researchers, technology and cash to undertake these activities. Even in these areas records are not complete. An example of a closely monitored volcano is Sakurijama, In South Kyushu, Japan (below).

In countries lacking the financial and technological resources for such monitoring, more basic but still useful techniques have been used. In the Philippines, local people are trained to look out for early-warning signs such as sulphur odours, steam releases and crater glow.

Once scientists have detected signs of activity the events must be interpreted to produce a hazard assessment and prediction of what will happen. Only then can government officials and other agencies such as the news media be informed and warnings and evacuation be introduced to the general public. This is still difficult to do accurately, and interpretations may differ among the scientists involved. This was to some extent the case with Nevado del Ruiz and led to the delay of the warning to the Colombian government, although the hazard map eventually produced was very accurate. The ashfalls were confined to the marked areas and the valleys like the Lagunillas (leading to Armero town) were affected by lahars as predicted.

Volcanoes often show signs if impending eruption and monitoring can be increased as the volcano becomes active. Case studies include Pinatubo, Philippines and the Caribbean island of Montserrat with the Soufriere Hills volcano.  http://www.mvo.ms/