The Leizhou Peninsula: red soil, old volcanoes and the strait between two peoples

Most of us met this place as a word. It sits on a gravestone, or in the jiguan (籍贯) line on an old form, the one that asks for your ancestral hometown, the place a family counts as its origin and passes down the father’s line. Leizhou. The thunder coast. Somewhere south, near Hainan but not Hainan.

This page is about the ground itself. Not the gods on it or the people who left it, which the rest of this site covers, but the rock, the soil and the water. They explain more than you would expect, and one of the things they explain is why the Lui Chew and the Hainanese are two peoples rather than one.

Where it actually is

The peninsula hangs off the southwestern corner of Guangdong province and points south at Hainan Island. The South China Sea is on its east side; the Gulf of Tonkin, which Chinese sources call the Beibu Gulf, is on its west. Chinese Wikipedia calls it one of China’s three great peninsulas; English Wikipedia says the third largest. Our own facts page names the other two, Shandong and Liaodong.

Its parent city is Zhanjiang, and the old prefecture gathered the three counties our homeland page already introduces: Haikang, now Leizhou city itself, then Suixi, then Xuwen at the very tip. English Wikipedia’s article on Xuwen County says plainly that it “is located in the southernmost part of the mainland of China”. Go further south and you are in the water.

Two things are worth knowing about the numbers before you meet them. The first is that reference works give different sizes for this peninsula, and we do not know why:

The last row is ours, and it is the odd one out by a wide margin. We have flagged it for checking rather than quietly writing over it, and this page works from the two encyclopaedias, which put the peninsula at roughly 7,800 to 8,500 square kilometres. Even that is not quite a range: the Chinese figure is written 7800餘, “over 7,800”, so it is a floor rather than a measurement.

If that number means nothing to you, here is one that might: Selangor is 8,104 square kilometres. The homeland is about the size of Selangor. Its length is less contested but not settled either: Chinese Wikipedia and our own page both say about 140 km top to bottom, Britannica says 120 km, and we have printed both rather than picking.

The second thing is the weather, and one of the two mainstream accounts of the name points straight at it. Chinese Wikipedia says the peninsula has long been known for frequent thunderstorms, and gives it 80 to 112 thunderstorm days a year. Averaged out, that is a day with thunder every three to five days. (Its wording is Chinese, 素以雷暴频繁著称; the English here is ours, as every translation on this page is.) The same article is careful about the name: it says there are many accounts of where Leizhou came from and that two are mainstream, one pointing to the frequent thunder, the other to a mountain and a river both written 擎雷, which we would render “raising thunder”. Both the reading of that name and the reading of Qinglei shui (擎雷水) as a river are ours, not the article’s. Our homeland page tells the mountain-and-river version, which is the one the Tang record supports. Both can be true at once, and a coast with eighty thunder days a year is a reasonable place for either.

Why the ground is red

The first thing photographs of the place show is the colour. Not brown, not grey. Red, on the roads and in the fields. There is a reason, and it comes in three steps.

Step one, the rock underneath is old lava. Britannica says the peninsula is “mostly formed of basalt and geologically recent sedimentary rocks”. Basalt is what runny lava turns into when it cools: a dark, heavy, fine-grained rock. English Wikipedia notes that more than 90 per cent of all volcanic rock on Earth is basalt, so this is the ordinary stuff of lava flows rather than anything exotic. English Wikipedia also puts a number on how much of the surface here it makes up, saying basalt terraces account for 43 per cent of the peninsula’s area, with sea-cut terraces at 27 per cent and river-laid flats at 17 per cent. Treat that 43 per cent as an indication rather than a measurement: the article gives no source for it, and we could not trace one.

Step two, basalt is full of iron. English Wikipedia’s article on basalt describes it, in its own one-line summary, as a magnesium- and iron-rich rock. Its opening sentence has basalt forming from lava “rich in magnesium and iron”, and further down it says basalt’s physical properties come from “relatively low silica content and typically high iron and magnesium content”. While that iron is bound up inside solid rock, the rock is dark, not red.

Step three, heat and rain break the rock down. In hot, wet places with a wet season and a dry season, rain works down through rock for a very long time and dissolves away most of what will dissolve. What it cannot easily carry off, in English Wikipedia’s words, is “the more insoluble ions”, and the two it names as the main ones are iron and aluminium. Those stay behind and build up.

The peninsula has that climate, and our sources are surer of the pattern than of what to call it: English Wikipedia files it as humid subtropical, Chinese Wikipedia as tropical monsoon. English Wikipedia puts annual rainfall at 1,400 to 1,700 millimetres under monsoons from three directions, and Leizhou city’s municipal introduction, writing about the city rather than the whole peninsula, describes plentiful rain and a long wet season with drought common in spring and autumn. Heavy rain, then a long dry stretch, then heavy rain again is the alternation the soil needs. The soil that results is called laterite, and the same article says of it: “Nearly all laterites are of rusty-red coloration due to high iron oxide content.” Iron oxide is rust.

So: iron-rich lava, then a hot wet climate, then rust. That is the standard explanation for red tropical soil, and every ingredient of it is documented here, with one scope to declare. English Wikipedia calls the repetition of wet and dry seasons the essential feature of the process, and the source we have for that alternation writes about Leizhou city, which sits on this peninsula, rather than about the whole of it. One honest note about the word itself: none of our sources actually calls the Leizhou Peninsula’s soil “laterite”. UNESCO’s geopark page has plenty to say about the soil, including that volcanic processes here “have created fertile red soils”, but it never names the soil type. The chain fits, and we are naming it rather than quoting it.

Now the part the travel pages leave out. This soil is not simply rich. Two descriptions of it sit in the record and they do not match.

UNESCO’s geopark page says volcanic processes “have created fertile red soils that support dense tropical forest”, and the Global Geoparks Network entry uses almost the same words. Britannica’s entry on the same peninsula is harder. It says: “The soil layer, always thin, has been completely washed away in places, often after grassland fires or overgrazing have destroyed the protective vegetation cover.” It adds that almost all the original forest cover was destroyed long ago, except on the northern hills, and that the whole area “is much drier than the neighbouring mainland or Hainan”.

Both can be true, on different parts of the same peninsula. Britannica itself says the area “is now under development as a national centre for tropical crops and aquaculture”, which is not something you write about ground that will not grow anything, and UNESCO’s page names a village sitting inside what locals call the pineapple sea. Thin soil on a windy, drier-than-average coast still blows and washes away. The park page is written to bring visitors. The encyclopaedia entry is written to describe land. We print both because we cannot tell you which square kilometre you would be standing on.

One detail sits on the encyclopaedia side of that line: Britannica says forest belts have been planted across the peninsula since 1955, and gives their purpose as slowing the wind and so reducing evaporation. The same entry, a few lines earlier, is the one describing soil washed away where the vegetation cover went.

The volcanoes, and the craters

The peninsula is not one volcano. It is a scatter of low ones. Chinese Wikipedia describes the land as mostly tableland, meaning flat-topped ground raised above the country around it, with little rise and fall. It counts the volcanic hills standing on that tableland: seven on the basalt tableland around Suixi, Chengyue and Huguangyan in the north, which it puts at 45 to 55 metres above sea level, and ten on the southern basalt tableland, which it calls flatter without giving it a figure. It gives heights only for the southern ten, and gives them as haiba (海拔), elevation above sea level: generally 25 to 80 metres, with the taller ones over 200 metres. Britannica measures differently again, describing “the cones of numerous extinct volcanoes about 825 feet (250 metres) high in the northern and southern sections”. English Wikipedia names two of them, Yingfengling 鹰峰岭 and Tianyang 田洋, 15 km apart in the middle of the peninsula.

Read those numbers carefully, because they are not all the same kind of number. Elevation above sea level is not height above the ground you would be standing on, and on tableland the difference is most of the hill: the seven northern hills sit on ground already 45 to 55 metres up, and Chinese Wikipedia gives them no heights at all. So a southern hill at 25 metres elevation is a low rise, not a 25-metre climb, and we cannot convert any of these figures into what you would actually walk up. Britannica’s 250 metres is a decent hill. Nothing here is a mountain.

This page also carries two volcano counts, and they do not reconcile. Seventeen is Chinese Wikipedia’s tally of volcanic hills on the peninsula’s tablelands. UNESCO’s “more than 100 volcanoes”, further down, counts volcanic structures of every kind across a geopark that takes in northern Hainan too. Different objects, different ground.

Where did they come from? English Wikipedia says the volcanoes “derive from the east-to-west tectonic extension and thinning of the lithosphere connected with the creation of the South China Sea’s basin”. The lithosphere is the earth’s rigid outer shell, the crust and the stiff top of the layer beneath it. Put plainly: as the South China Sea was opening, that shell was being pulled sideways, the way dough thins when you stretch it, and where it thinned, magma found its way up. The source puts it no more strongly than “connected with”, and neither will we; but the sea and the volcanoes are connected.

A maar is a hole, not a hill

Most people picture a volcano as a cone. Several of the ones here are closer to a bowl, and the reason is water.

Chinese Wikipedia sets out the sequence, describing Huguangyan. Groundwater or surface water meets rising magma directly. The water flashes to vapour on the spot, producing an extremely high-pressure steam explosion. The fragments thrown out fall back and pile up into a ring, leaving a crater hollow inside it, and over a long time the hollow fills with water and becomes a lake. So something is built, but it is a rim rather than a cone.

A crater made this way is called a maar. UNESCO’s description is short enough to quote: maars are “broad, low-relief volcanic depressions formed when rising magma interacts explosively with groundwater”. Several of the sea cliffs in the geopark, UNESCO adds, came from the same kind of blast.

Huguangyan (湖光岩), in Zhanjiang’s Mazhang District, is the famous one: a lake sitting in its own crater, ringed by its own cliffs. Chinese Wikipedia dates the eruption that made it to roughly 140,000 to 200,000 years ago and gives the lake as about 2.23 square kilometres, averaging 14.5 metres deep and reaching 23 metres.

Extinct, dormant, or neither

This page was planned under the title “dead volcanoes”. We are not going to use that word, and the reason is worth setting out, because it is the clearest example on this page of what checking sources actually does.

The two English reference works disagree, mildly. Britannica calls them “extinct”, meaning finished. English Wikipedia says the peninsula is “dotted with a few dormant volcanoes”, meaning sleeping. Those are not the same claim.

Then the geological dating pulls further the other way. English Wikipedia dates one of the peninsula’s two volcanic fields, Leibei (雷北火山群) in the north, to the Pleistocene-Holocene. The Pleistocene is the ice-age epoch that ended about 11,700 years ago; the Holocene is the one we are living in now. So that label reaches to the present day. UNESCO says the site “is regarded as one of the most significant Quaternary basaltic volcanic regions in China”; the Global Geoparks Network says it “is considered among the top of the Quaternary volcanic belts in China”, both of them reporting a judgement rather than making one; and Chinese Wikipedia gives Huguangyan’s geological age as Quaternary.

And our own facts page already says it outright: the peninsula saw eruptions within the last 10,000 years.

So “dead” is not what the record says. What we can honestly tell you is this. No source we can reach gives a calendar date or a firm age for the most recent eruption anywhere on the peninsula. The Smithsonian Global Volcanism Program’s entry and the geological papers that would settle it did not open for us. Nothing has erupted here in any historical record we have seen, and the hills are low: Britannica describes undulating upland of generally low relief, meaning land that does not rise and fall much. But the epoch labels on this volcanism run right up to the present, and ground whose youngest activity may fall within the last ten thousand years is better described as old and quiet than as dead.

One land, then a strait

Here is the fact this page was written for, and it is better than the one we set out to find.

The peninsula and Hainan were once a single landmass. Chinese Wikipedia’s article on the peninsula gives the sequence. A very large basin formed here in the rock, called the Qiong-Lei depression, 琼雷凹陷. Volcanic activity continued, and the Leizhou Peninsula and Hainan Island rose together to become land, with basalt laid over the earlier layers. Then, at the end of the Middle Pleistocene and the start of the Late Pleistocene, roughly the second half of the last ice age, the ground where the Qiongzhou Strait now runs broke along a fault and dropped. A fault is a crack in the rock where one side can move against the other. The sea came into the hollow that left. The peninsula and the island were separated, and the coastline you can see today is what remained.

Chinese Wikipedia’s article on the strait tells the same story from the water’s side: the strait area, Hainan Island and the peninsula were once complete land; crustal movement during the Quaternary glacial period formed a down-faulted trough where the strait now is, and seawater covered it. That article also notes that volcanic rock is spread along both shores, that the northern shore is a basalt structure and the southern shore a lava platform.

The rock on both sides is volcanic rock. The water is the newcomer.

You can see the same thing in how the land is protected today. Leiqiong UNESCO Global Geopark straddles the Qiongzhou Strait. Read the name slowly: 雷 Lei stands for Leizhou and 琼 Qiong for Hainan, the pairing our comparison of the two peoples already uses for the language family. A geopark is an area UNESCO recognises and protects for its geology, the way a heritage site is protected for its history. This one spans Zhanjiang City on the mainland and Haikou City on the island, and covers 3,050 square kilometres, roughly a third of the peninsula’s own area but spread across both shores. UNESCO records 78 geosites in it, meaning individual places of geological interest, along with more than 100 volcanoes and 38 protected volcanic structures. The geology it exists to protect has one name across both shores: the Leiqiong Rift Volcanic Belt.

So the ground says one place. About 30 kilometres of water says two peoples, and the water is what the record shows arriving later.

Elsewhere on this site, on the homeland page and in our comparison of the two peoples, you are told the Lui Chew and the Hainanese are cousins: sister languages, and families that came south in the same movement out of Fujian, both speaking a Min language, the branch of Chinese rooted in Fujian to which Hokkien and Hainanese also belong. Those pages make the case from language and history. The geology puts a floor under it. Two branches of one migration settled two halves of one volcanic province, on the same red dirt, and the strait was already lying between them long before either arrived.

Nobody would say a channel of water alone makes a people. Separate administrations, different neighbours and several centuries of separate history all did their work. But if you want the one physical fact behind the split, it is a fault that dropped and filled with sea.

Our word-by-word test of the two languages reads differently in this light. It is a measurement of what that channel did to speech. The words mostly still cross it. The tune does not.

The strait was also a road

It would be easy to leave the strait as a wall. It was never only that.

Start with its size, because this is where our own pages and the reference works have been talking past each other. They are mostly not in conflict. They are measuring different things, and Chinese Wikipedia is the source that gives the full set, each width labelled:

Narrowest

About 18 km in a straight line, which is the figure our own facts page already gives, and it is right

Average width

29.5 km north to south, close to the "on average 30 km" English Wikipedia gives, though that same article's summary box labels 30 km as the maximum, which its own text does not

Widest

33.5 km in a straight line, close to the 33 km English Wikipedia's Xuwen article gives. Watch that one: Xuwen expresses it as 18 nautical miles, so the number 18 sits in the record as both the narrowest in kilometres and the widest in nautical miles

Length

About 80 km east to west

Depth

44 m on average, 114 m at the deepest by Chinese Wikipedia; English Wikipedia gives about 120 m as the maximum

What it joins

The Gulf of Tonkin, west, to the South China Sea, east

Britannica’s 16 km is the one figure left over, and it is close to the narrowest-point measurement rather than to any of the others. So the picture is simple once the labels are attached: a channel that pinches to about 18 km at its tightest and opens to about 33 km at its widest.

It is not always crossable. English Wikipedia notes that the strait “is susceptible to closure during strong typhoon activity”, and Chinese Wikipedia records that a great earthquake in July 1605 dropped parts of the land on both shores and drowned 72 villages under the sea. A ferry carries road vehicles and rail carriages across today. A bridge was planned in the early 2000s and, in English Wikipedia’s words, “never came to fruition”; a bridge or tunnel was still being discussed as of 2018.

Now the older use. Xuwen, at the tip, was a departure point on the sea route out of Han-dynasty China. A China Daily feature on the county dates the opening of that maritime route to 111 BCE, after Xuwen County was set up, and quotes what it calls the “Geography Records of The Han History”, which is the geography treatise of the Hanshu, the Book of Han; the feature attributes it to Ban Gu, a historian of the Eastern Han. The passage describes ships sailing about five months to reach a kingdom it calls Duyuan.

One correction belongs with that. The Hanshu passage does not name Xuwen alone. It names Rinan, Xuwen and Hepu together, three departure points, of which Xuwen is one. The feature’s headline calls Xuwen “the ancient departure port”; its own body text says Xuwen “is one of the ports of origin in the Han Dynasty”, which is the careful version. Our homeland page gets this right and says “one of China’s earliest Maritime Silk Road departure ports”. Other pages on this site claim more than that, and we have flagged them for correction.

Whatever the ranking, the shape of the thing holds. Two thousand years before any Lui Chew boarded a ship for Melaka, the tip of this peninsula was a place people sailed south from. Britannica gives Xuwen’s port as Hai’an, at the southern tip, and records that in 2000 a railway was completed south from Zhanjiang to Hai’an, from where railway carriages are carried across to Hainan by ferry.

What the land does not explain

It is tempting, at this point, to close the loop: thin soil, typhoon coast, sea road pointing south, so of course they left. Be careful with that one. The land is part of the answer. It is not the answer, and it is not the trigger.

Our page on why they left sets out what the record supports, and it has a section headed “Why the peninsula pushed”. It describes the homeland as “a sea-girt, typhoon-exposed agrarian frontier, fertile in parts, but crowded, remote, and far from the centres of Qing prosperity”, and the first of the five documented drivers it lists is poverty, meaning hardship and scarcity in the home villages. So a hard place to stay is genuinely in the account, and it is first in the list.

What is not in the account is the geology. Nothing in our sources connects the volcanoes, the basalt or the soil type to anyone’s decision to go. The other four drivers are a labour market, a war, an invasion and a law: colonial demand for labour in British Malaya and Singapore; the Taiping Rebellion and the wider late-Qing collapse; the Japanese assault on the Qiong-Lei homeland in 1937; and the Treaty of Nanjing of 1842, which made working abroad lawful. That same page is careful about the treaty in a way worth repeating here: it says the Treaty of Nanjing “did not cause the migration, but it removed the lock on the door the push and pull were already straining against.”

Two cautions ride along. The five drivers come from a study of the Melaka Hainanese, and our page applies them to the Lui Chew by analogy, because Leizhou-specific records of why individual families left are thin, thinner than for the Hokkien or Cantonese. And the picture of the homeland as farmland is younger than it looks: our homeland page records that within living memory the peninsula was scrubland battered by typhoons and wind erosion, and that the 1960s canal and the coastal windbreak forest turned it into the fields it grows today.

What the geography honestly explains is narrower, and still worth having. The homeland is a farming and fishing coast rather than a rice bowl. Britannica records both halves of that and offers a reason for neither: less rice is grown here than in other parts of Guangdong, and the area is under development as a national centre for tropical crops and aquaculture, meaning farmed fish and seafood. A coast with a sea road running off its tip produced people who knew how to use one. And the ground explains, better than anything else on this page, why they are not Hainanese.

What we do not know yet

When the volcanoes last erupted. Epochs, not dates, as the section above sets out. The Smithsonian Global Volcanism Program’s entry and the geological papers that would settle it did not open for us.

How big the peninsula is. Between 7,800 and 12,500 square kilometres, depending on which page you read, including one of our own. We have used the encyclopaedias’ range and flagged our own figure for checking, but we have not established which is right.

How good the soil really is. The park description and the encyclopaedia entry describe two different soils. We suspect both are right in different places, but that is our reading rather than something a source says.

Which villages sent people to Malaysia. Still open, as it is on the homeland page. We do not know whether the Malaysia-bound families came mainly from Haikang, Suixi or Xuwen, and no village-level emigration record for the peninsula has turned up in public sources.

What the red soil looks like to someone standing on it now. We are working from encyclopaedias, two geopark listings, a news feature, a municipal introduction and our own pages. If you have been, and you have photographs of the red roads, the pineapple fields, Huguangyan, or a volcanic hill with a name on it, a contribution would improve this page more than another source would.

Sources for this page: UNESCO's International Geoscience and Geoparks Programme entry for Leiqiong UNESCO Global Geopark, and the Global Geoparks Network entry for the same park; Encyclopaedia Britannica's entry on the Leizhou Peninsula; English Wikipedia's articles on the Leizhou Peninsula, the Qiongzhou Strait, Xuwen County, basalt, laterite and Selangor; Chinese Wikipedia's articles on the Leizhou Peninsula (雷州半岛), the Qiongzhou Strait (琼州海峡) and Huguangyan (湖光岩); Leizhou city's own municipal introduction; a China Daily feature on Xuwen county, which quotes the geography treatise of the Hanshu; and our own homeland, facts and why they left pages. The Britannica and UNESCO entries are quoted from Internet Archive copies, of 22 May 2025 and 25 April 2026, because the live pages would not open for us. Every translation from a Chinese source on this page is ours. Most of these are new to this site: of the sources named above, English Wikipedia's Xuwen County article and the Leizhou city government's own pages are the only ones already on our sources page. The rest, including Britannica, both geopark listings and all three Chinese Wikipedia articles, should be added there with their reliability tier when this page goes up. To see where all of this sits, use the map; if it has made you want to go, read visiting Leizhou first.