So at the beginning of the oil industry, wells were drilled blindly near known seeps in the hopes that they would strike it lucky.
With the boom in demand caused by the industrial revolution in the late 19th Century and the start of the 20th, the reservoirs indicated by the surface show in countries like the United States had nearly all been exploited. This was then an era of 'wildcat' drilling - drilling in places where there were no known oil and gas fields. The drillers chose places that they thought looked promising - at the edge of a hill on an otherwise flat plain, for example, with the thinking that maybe it acted as a reservoir trap. Some wildcatters were hugely successful, but many others would have been hugely bankrupt.
Modern day exploration relies mainly on seismic data acquisition, or seismic, for short.
This is when sound waves are sent through the target rock, in which they are partially reflected by changes in rock type (e.g. if the formation changes from siltstone to sandstone, or to volcanics, or granite, or even if the density of the formation changes slightly - maybe one year was wetter and more mud got washed down from a river to form a thicker band of claystone). These reflected waves are picked up by geophones (microphones specially designed to pick up sound waves from the earth) - the waves that are detected by the geophones first will be from shallower rock, so a picture of the shape of the different formation layers (strata) can be obtained. From these you can see if there are any trap structures which have the potential to be reservoirs.
This is performed both offshore and onshore, with the following:
You can shoot two main types of seismic:
Not only does seismic not give consistent indications of hydrocarbons, or tell you if you're looking a viable petroleum bearing structure, but used alone it does not indicate accurate depths of the formations you're looking at:
Reflection is common to all wave types (whether sound, in the case of oil exploration, or light, or x-rays or gamma). Reflection is caused by a change in speed of the wave, and the speed of sound in all materials depends mainly on their density (sound is the transmission of energy through solid particles - if the particles are closer together (denser) then they have to travel less far before they hit the next particle, and the next particle hits the next, and so on, in the form of a sound wave).
Rocks, as they have lots more particles in a small space, have a higher speed of sound (a lower acoustic impedance) than air, which will have a lot fewer particles.
But sound doesn't travel at the same speed through all rock types (hence the reflection at the interfaces between different strata), and if you don't know the type of rocks that you're looking at on your seismic graph, then you won't know their density, and if you don't know their density then you won't know how fast the sound has passed through them. This means that seismic reflections that look very close together could turn out to be a lot further apart when you actually drill through them, or vice versa.
It could also mean that you spend tens of millions of dollars drilling into what looks like a wonderful hydrocarbon trap, but turns out to be a 'pull-up' - when an area of lower seismic velocity surrounds an area with higher seismic velocity (like a salt dome or carbonate reef). It will appear that the reflector is a structural high, though this 'pull-up' is not physically present in the formation - it is only a result of the different transit times of the formations adjacent to each other.
If you have already drilled in the region where you're shooting seismic then the data becomes altogether more useful. You gain knowledge of the rocks as you're drilling through a variety of methods , so it can become a case of just following the seismic reflector line from your existing well (called an 'offset well') to another location in order to find another reservoir, or to get an idea of how big the reservoir you've just found is.
Apart from seismic, another technological method that can be used to detect oil is gravimetry.
This is when small differences in gravitational force are detected by an aeroplane with sensitive purpose built accelerometers. This method can be used to detect formations which have a lower density - therefore lower mass, therefore lower gravity, such as salt domes. Gravimetry would be used in conjunction with seismic as it doesn't give very detailed data on stratigraphy, but can be used to survey very large regions quickly and relatively cheaply. As mentioned, oil is often found on a completely different continent on the basis that it has already been found somewhere across the oceans, and they used to be joined (e.g. they've found it in West Africa from drilling in Brazil, even though there?s thousands of miles of water and seabed rock between them).
As more wells are drilled and more seismic is shot, the understanding of plate tectonics increases significantly, with computer simulations deducing the changes in the Earth's geology over the past hundreds of millions of years. Such modelling can be used to ascertain where the best places to drill are - where an ancient sea was, or the delta of a Jurassic era river.
Despite all the expensive seismic and technological innovation, oil exploration remains a risky business. It is said that only one in ten exploration wells drilled finds oil or gas, and only one in ten of those is an economically viable prospect. That means out of 100 wells drilled, they were only correct in their geological prognoses once (although these exploration companies are never "wrong", they'll always say something like "we increased our understanding of the geology of the region" to try to minimise the damage to their share price). When you consider that it can cost over $100 million to drill a single exploration well, you start to realise what a massive reward there must be if you are successful.