Many people are interested in ways to classify volcanoes. There is probably a natural human instinct to try and give labels to all things. This is not a bad instinct and many times it makes it easier to understand the particular thing that is being classified. For example, you start to identify patterns when you classify things and these patterns may lead to a better understanding of whatever it is you are classifying. However (and that is a big "however"), when you are classifying natural things (they might be fish, plants, birds, oceans, minerals, volcanoes, or whatever), you MUST remember that the classification scheme is made up by human beings and Nature might decide to not follow the rules exactly. There will ALWAYS be exceptions to your classification scheme and there will ALWAYS be things that fall into more than one category. As long as you realize this and it doesn't bother you, you'll be just fine. Certainly there are different ways to classify volcanoes and all of them have particular benefits and drawbacks. These include classifying by lava chemistry, tectonic setting, size, eruptive character, geographic location, present activity, and morphology. As an example of how these can get mixed together, note that there are basaltic strato volcanoes (i.e. Mt. Fuji), big basaltic calderas (i.e. Taal), big gradual-sloped basaltic shields (i.e. Mauna Loa) and big steep-sloped basaltic shields (i.e. Fernandina). Additionally, although most volcanoes associated with subduction zones are steep-sided andesite or dacite cones, there are a few basaltic shields along these zones as well (i.e. Masaya, Westdahl, Tolbachik). These examples highlight the above-mentioned hurdle that any student of the Earth needs to get over - Nature makes exceptions to human rules.
Unfortunately, there is one particular volcano classification system that many people think is the only system. Not only is it not the only system, it is not a very good system. This is the famous "3 types of volcanoes" (shield volcanoes, strato volcanoes, and cinder cones), and it is found in many textbooks from elementary school to college. Why is this 3-types scheme so bad? First, it has no place in it for large caldera complexes (such as Yellowstone), flood basalts, monogenetic fields, or mid-ocean spreading centers. These are important types of volcanoes that you would never hear about if you thought there were only 3 types. Second, although you can occasionally find a cinder cone sitting somewhere all by itself, it is way more common for a cinder cone to either be one of many vents on a large (polygenetic) volcano or a member of a monogenetic field. Finally, if you actually think about the system you run into logical problems, as a teacher from Pittsburgh pointedly complained to VolcanoWorld about: She wanted to know how Pu'u 'O'o could be a cinder cone on Kilauea if cinder cones are a type of volcano and Kilauea is a shield volcano. The answer is that Pu'u 'O'o is one of hundreds of vents on Kilauea, and it happens to be a cinder cone.
Who knows what the origin of this 3-volcano system is, but the sad thing is that many people use it without thinking as far as the Pittsburgh teacher did. The cinder cone part may come from the fact that some cinder cones have names such as "This Volcano" or "Volcan That" even when they are just vents on a larger volcano. In these cases the cinder cone is probably all that has ever erupted in the collective memory of the local folks. They logically consider it to be "the volcano" and may think of the larger structure that hasn't erupted since they've been around (and may in part be highly eroded or vegetated) to be "just" a mountain.
For most volcanological applications a classification based on morphology is probably the most useful. In their excellent book Volcanoes of the World, Tom Simkin and Lee Siebert list 26 morphological "types" of volcanoes. That's certainly thorough but kind of extreme. You can account for probably >90% of all volcanoes with 6 types. Additionally, any system will be more useful if you use modifiers from the other potential classification schemes with the morphological types (i.e. active andesite strato volcano, extinct hotspot shield volcano, etc.).
The following descriptions of 6 morphological volcano types are really brief. They were originally written for an "ask-a-volcanologist" answer - if they tell you things you already know, please don't feel insulted. In most any good volcanology book you should be able to find more details and many more examples.
Shield volcanoes are the largest volcanoes on Earth that actually look like volcanoes (i.e. not counting flood basalt flows). The Hawaiian shield volcanoes are the most famous examples. Shield volcanoes are almost exclusively basalt, a type of lava that is very fluid when erupted. For this reason these volcanoes are not steep (you can't pile up a fluid that easily runs downhill). Eruptions at shield volcanoes are only explosive if water somehow gets into the vent, otherwise they are characterized by low-explosivity fountaining that forms cinder cones and spatter cones at the vent, however, 90% of the volcano is lava rather than pyroclastic material. Shield volcanoes are the result of high magma supply rates; the lava is hot and little-changed since the time it was generated. Shield volcanoes are the common product of hotspot volcanism but they can also be found along subduction-related volcanic arcs or all by themselves. Examples of shield volcanoes are Kilauea and Mauna Loa (and their Hawaiian friends), Fernandina (and its Galápagos friends), Karthala, Erta Ale, Tolbachik, Masaya, and many others.
Here are 4 of the volcanoes that comprise the big island of Hawai'i. They are Mauna Kea (MK), Mauna Loa (ML), Hualalai (H), and Kohala (K). The photo was taken from near the summit of East Maui volcano (EM). These are the largest volcanoes on Earth.
This is a vertical air photo of the summit caldera of Mauna Loa volcano (North is to the left). Notice that the caldera is composed of numerous smaller "cookie-cutter" collapses which have coalesced to form the main caldera. Notice also that many of the lava flows (dark and light are 'a'a and pahoehoe, respectively) have been truncated by the caldera margin. This is an indication that they erupted from the volcano summit when the caldera was full. Collapse since then has produced the present caldera. In this manner of collapsing and filling, calderas come and go throughout the active lifetime of a basaltic volcano.
Strato Volcanoes comprise the largest percentage (~60%) of the Earth's individual volcanoes and most are characterized by eruptions of andesite and dacite - lavas that are cooler and more viscous than basalt. These more viscous lavas allow gas pressures to build up to high levels (they are effective "plugs" in the plumbing), therefore these volcanoes often suffer explosive eruptions.
Strato volcanoes are usually about half-half lava and pyroclastic material, and the layering of these products gives them their other common name of composite volcanoes.
Left: This is a schematic diagram of a strato volcano, intended to illustrate the different layers of different materials that comprise them. The purple colors are meant to represent ash layers, either the products of fall-out from big eruption clouds or the products of pyroclastic flows. Notice that these ash layers tend to be thin but widespread. The orange colors represent lava flows, and note that some of them have cinder cones associated with them at the vent. The green colors are meant to represent lava domes, and notice that they do not flow very far. Each eruption, regardless of what it produces, is fed from the magma chamber by a dike. Most dikes come up through the center of the volcano and therefore most eruptions occur from at or near the summit. However, some dikes head off sideways to feed eruptions on the flanks.
Right: This is a pit that has been dug into the ground at Cotopaxi, a big strato volcano near Quito, the capital city of Ecuador. The pit is about 2 meters deep and in it you can clearly see a number of ash layers exposed. It is also easy to see that the layers are different - some are coarse and others are fine, some are dark-colored and others are light-colored
The lava at strato volcanoes occasionally forms 'a'a, but more commonly it barely flows at all, preferring to pile up in the vent to form volcanic domes. Some strato volcanoes are just a collection of domes piled up on each other. Strato volcanoes are commonly found along subduction-related volcanic arcs, and the magma supply rates to strato volcanoes are lower. This is the cause of the cooler and differentiated magma compositions and the reason for the usually long repose periods between eruptions. Examples of strato volcanoes include Mt. St. Helens, Mt. Rainier, Pinatubo, Mt. Fuji, Merapi, Galeras, Cotopaxi, and super plenty others.
Although they are not as explosive as large silicic caldera complexes, strato volcanoes have caused by far the most casualties of any type of volcano. This is for many reasons. First is that there are so many more strato volcanoes than any of the other types. This means that there will also be lots of people who end up living on the flanks of these volcanoes. Additionally, strato volcanoes are steep piles of ash, lava, and domes that are often rained heavily on, shaken by earthquakes, or oversteepened by intruding blobs of magma (or all of these). This makes the likelihood of landslides, avalanches, and mudflows all very high. Occasionally as well, entire flanks of strato volcanoes collapse, in a process that has been termed "sector collapse". Of course the most famous example of this is Mt. St. Helens, the north flank of which failed during the first stages of the big 1980 eruption. Mt. St. Helens was certainly not the only volcano to have suffered an eruption such as this, however. Two other recent examples are Bezymianny (Kamchatka) in 1956, and Unzen (Japan) in 1792. The 1792 Unzen sector collapse dumped a flank of the volcano into a shallow inland sea, generating devastating tsunami that killed almost 15,000 people along the nearby coastlines.
Left: This is a photo of lahar deposits near Santa Maria volcano (Guatemala). This used to be a wide, deep river valley, and you can see the far wall of the valley where the trees are growing. The lahar deposits extend from that far wall to behind where this photo was taken. You can see that between lahar events the river cuts into the lahar deposits, but every time there is another event, it fills up again. The people give an idea of the size of stones that can be carried by a lahar.
Another very common and deadly hazard at most strato volcanoes is called a Lahar. Lahar is an Indonesian word for a mudflow, and most geologists use the term to mean a mudflow on an active volcano. Sometimes the word is reserved only for mudflows that are directly associated with an ongoing eruption (which are therefore usually hot), but that starts to make things confusing. It is probably simplest to just call any mudflow on a volcano a lahar. Lahars are so dangerous because they move quickly, and often times a small eruption or relatively small rainstorm can generate a huge lahar. The most recent huge volcanic disaster occurred at a Colombian volcano called Nevado del Ruiz in 1985. This disaster has been well-documented by numerous post-eruption studies. Nevado del Ruiz is a very tall volcano, and even though it lies only slightly above the equator it has a permanent snow and ice field on its summit. On November 13, 1985 a relatively small eruption occurred at the summit. Even though only a little bit of ash fell and only small pyroclastic flows were produced, they were able to melt and destabilize a good deal of the summit ice cap. The ice cap had already been weakened and fractured by a few months of pre-cursor seismic activity. The melted snow and ice, along with chunks of ice, surged down gullies that started high on the slopes, picking up water, water-saturated sediments, rocks, and vegetation along the way. The eruption occurred just after 9:00 pm, and about 2 and a half hours later lahars managed to travel the approximately 50 km down river valleys to the town of Armero. The lahar entered Armero at 11:30 pm as a wall of muddy water nearly 40 meters high, and roared into the city, producing an eventual thickness of 2-5 meters of mud. Somewhere around 23,000 people were almost instantly killed. The path of destruction almost exactly matches similar disasters that occurred in 1595 and 1845. It also almost exactly covered the highest lahar-designated area on the volcanic hazard map that had been prepared prior to the 1985 eruption. Unfortunately that map had not yet been distributed by the time of the 1985 eruption.
Another place that is starting to get really tired of lahars is Pinatubo, in the Philippines. The 1991 Pinatubo eruption was the second largest this century (after Katmai in 1912), and deposited a huge volume of relatively loose pyroclastic material on already-steep and gullied slopes. Additionally, the rainfall in the Philippines is very high. The combination of all this unconsolidated material and heavy rainfall has generated probably hundreds of lahars, some of which have been enormous. Timely evacuation meant that only a couple hundred people were killed directly by the 1991 eruption. Many times that many have been killed or injured by lahars since the 1991 eruption. These lahars will continue to be a problem for decades after the big eruption.
Rhyolite caldera complexes are the most explosive of Earth's volcanoes but often don't even look like volcanoes. They are usually so explosive when they erupt that they end up collapsing in on themselves rather than building any tall structure (George Walker has termed such structures "inverse volcanoes"). The collapsed depressions are large calderas, and they indicate that the magma chambers associated with the eruptions are huge. In fact, layers of ash (either ash falls or ash flows) often extend over thousands of square kilometers in all directions from these calderas. Fortunately we haven't had to live through one of these since 83 AD when Taupo erupted. Many rhyolite caldera complexes, however, are the scenes of small-scale eruptions during the long reposes between big explosive events. The vents for these smaller eruptions sometimes follow the ring faults of the main caldera but most often they don't. The origin of these rhyolite complexes is still not well-understood. Many folks think that Yellowstone, for example, is associated with a hotspot. However, a hotspot origin for most other rhyolite calderas doesn't work; they occur in subduction-related arcs. Examples of rhyolite caldera complexes include Yellowstone, La Primavera, Rabaul, Taupo, Toba, and others.
This is an outcrop in the Los Chocoyos ignimbrite, the product of one of the most powerful eruptions known...
Monogenetic fields don't look like "volcanoes", rather they are collections of sometimes hundreds to thousands of separate vents and flows. Monogenetic fields are the result of very low supply rates of magma. In fact, the supply rate is so spread out both temporally and spatially that no preferred "plumbing" ever gets established; the next batch of magma doesn't have a pre-existing pathway to the surface and it makes its own. A monogenetic field is kind of like taking a single volcano and spreading all its separate eruptions over a large area. There are numerous monogenetic fields in the American southwest and in México, including Michoacan-Guanajuato, San Martín Tuxtla, Pinacate, and the San Francisco volcanic field.
Image of the San Francisco volcanic field credit:
Flood basalts are yet another strange type of "volcano." Some parts of the world are covered by thousands of square kilometers of thick basalt lava flows - individual flows may be more than 50 meters thick, and individual flows extend for hundreds of kilometers. The old idea was that these flows went whooshing over the countryside at incredible velocities (e.g., like a flash flood). The new idea is that these flows are emplaced more like flows, namely slow moving with most of the great thickness being accomplished by injecting lava into the interior of an initially thin flow. The most famous US example of a flood basalt province is the Columbia River Basalt province, covering most of SE Washington State and extending all the way to the Pacific and into Oregon. The Deccan Traps of NW India are much larger and the Siberian Traps are even larger than that (but poorly understood). The Ontong Java plateau may be an oceanic example of a flood basalt province.
Click here for more information on the Columbia River Flood Basalts
Click here for more information on the Deccan Traps
This is a map of the major oceanic spreading centers. This is sometimes considered to be one ~70,000 km-long volcano. Here, the plates are pulled apart by convection in the upper mantle, and lava intrudes to the surface to fill in the space. Or, the lava intrudes to the surface and pushes the plates apart. Or, more likely, it is a combination of these two processes. Either way, this is how the oceanic plates are created.
The lava produced at the spreading centers is basalt, and is usually abbreviated MORB (for Mid-Ocean Ridge Basalt). MORB is by far the most common rock type on the Earth's surface, as the entire ocean floor consists of it. We know that spreading occurs along mid-ocean ridges by two main lines of evidence: 1) the MORB right at the ridge crest is very young, and it gets older on either side of the ridge as you move away; and 2) sediments are very thin (or non-existent) right near the ridge crest, and they thicken on either side of the ridge as you move away. Mid-ocean ridges are also the locations of many earthquakes, however, they are shallow and generally of small magnitude.
We have never witnessed an eruption along a mid-ocean ridge, although a few times earthquake swarms have been detected along them (mainly by secret US Navy listening devices). When scientists have investigated soon after, fresh-looking basalt, plumes of hot chemical-laden water, and recently-killed marine organisms have been observed, indicating that an eruption almost certainly had occurred.