For the analyses, the data and information from the WOKAM digital map and database (Chen et al. 2017b) were combined with a wide range of publicly available global online datasets. All analyses were done using Arc GIS Desktop by ESRI (version 10.4.1).
As WOKAM is part of the World-wide Hydrogeological Mapping and Assessment Programme (WHYMAP), the same data sets/shapefiles concerning continent boundaries, state boundaries, ice sheets, and lakes were used than for all other products of the WHYMAP series. These data were obtained from the WHYMAP team at BGR and are essentially based on data from the online collection of data and maps for ArcGIS (ArcGIS online, ESRI) and from the UN Geospatial Information Section. National population estimations come from the United Nations Population Prospects 2017 Revision (UN 2017). Global landforms were obtained from Sayre et al. (2014). The global climatic zones come from an updated version of the conventional Köppen-Geiger climate classification at 1-km resolution (Beck et al. 2018). Global population densities were obtained from “Center for International Earth Science Information Network” (CIESIN) and “Centro Internacional de Agricultura Tropical” (CIAT) (CIESIN and CIAT 2005a, b). The coastline was obtained from Natural Earth Data. Global precipitation data were taken from Meyer-Christopher et al. (2015). The analyses focus on the continental ice-free land surface. This excludes Antarctica, as it is mostly covered by glacier ice, but includes the ice-free parts of Greenland, as a part of North America.
Essential concepts and principal contents of the World Karst Aquifer Map
Figure 1 shows a strongly generalized and downscaled version of WOKAM, while Fig. 2 shows an enlarged section of the map in order to better illustrate the concept and degree of detail of the original full-scale map. WOKAM is based on the observation that karst aquifers typically develop in carbonate rocks containing more than 75% of carbonate minerals, and in evaporite rocks (Ford and Williams 2007). Although the actual degree of karstification and the aquifer properties depend on several geological, geomorphological, geochemical and hydro-climatological aspects, it is a safe and pragmatic approach to assume that all outcrops of such rocks are karstified at least to some degree. Therefore, all exposed karstifiable carbonate and evaporite rocks were mapped as potential karst aquifers; however, the analyses in the present paper focus entirely on carbonate rocks. Generalization was done at a working scale of 1:10 million. An inherent problem in the generalization of maps is the existence of polygons that are too small or complex in shape to be displayed. To overcome this problem, carbonate (and evaporite) areas were subdivided into “continuous” and “discontinuous”, based on the area’s share of the respective rock type (Chen et al. 2017a). Wherever possible, the mapping unit continuous was applied, because it is readily understood. Polygons classified as continuous often include small patches of nonkarst surfaces. By comparing the original with the generalized polygons, it turned out that the percentage of carbonate rocks was generally >65%. For areas that contain highly scattered or ramified outcrops, the mapping unit discontinuous was introduced. This approach was validated in several regions, and it turned out that the limits of 15 and 65% result in a hydrogeologically meaningful generalization. Therefore, areas with more than 65% of carbonate (or evaporite) rock were classified as continuous on WOKAM, whereas areas between 15 and 65% were classified as discontinuous. Areas containing more than 15% of each rock type (carbonate and evaporite) were mapped as mixed karst. The example map in Fig. 2 shows a region where all of these mapping units are present. Zones where exposed karstifiable rocks plunge under adjacent overlying formations—typically stratigraphically higher clastic sedimentary formations—are highlighted by a line of triangles that points in the direction of nonexposed carbonate or evaporite rocks. These lines indicate the location of nonexposed, deep and confined aquifers which are potential resources for freshwater or, with increasing depth, thermal and mineral water. The detailed map in Fig. 2 shows several of such contacts, e.g. the southern margin of the Edwards Aquifer (Texas, USA) where exposed carbonate rocks (“recharge zone”) plunge under overlying clastic sedimentary formations (“artesian zone”). However, these nonexposed potential karst aquifers are not spatially delineated in WOKAM, because this would require three-dimensional (3D) geological information that was not available at a global scale. Therefore, all following statistical analyses referring to the spatial distribution of karst are restricted to exposed carbonate rocks. Groundwater resources in exposed carbonate rocks are more vulnerable to contamination resulting from intense agriculture and other human activities than those in nonexposed carbonate rocks (Goldscheider 2005). Therefore, a map of exposed carbonate rocks is particularly useful and relevant for land-use planning in terms of groundwater protection.
The Global Lithological Map (GLiM) was an important data source for WOKAM (Hartmann and Moosdorf 2012), but many other regional maps at different scales were also used, according to a well-defined and consistent workflow that was described by Chen et al. (2017a). Evaluation by an international network of regional experts was an important part of this process.
The full-scale WOKAM map and database also include 201 selected karst water sources, including 162 continental freshwater springs, 16 submarine springs, 8 thermal springs and 15 water abstraction structures, as well as 93 selected cave systems. The detailed map section in Fig. 2 also illustrates how these karst water sources and caves are shown on the full-scale map. Tables 1 and 2 present details. The selection criteria for springs and caves include regional relevance, size and discharge (details in Chen et al. 2017a). The map in Fig. 3 and the associated Table 3 present a small selection of the largest and most important karst springs on all continents, selected on the basis of discharge and on regional distribution and significance. Recently, a World Karst Spring Hydrograph Database (WOKAS) has been prepared (Olarinoye et al. 2020), complementing the data included in the WOKAM database. WOKAM also indicates permafrost boundaries (areal percentage > 50%), because karst development and karst aquifer behavior is very different and limited under permafrost conditions (Ford 1987).
Global and regional distribution of carbonate rocks
According to the analysis, 9.4% of the global ice-free land surface consists of continuous carbonate rocks (CC) and 5.8% are occupied by discontinuous carbonate rocks (DC) or carbonate rocks mixed with evaporites. This means that 15.2% or 20.3 million km2 of the land surface are characterized by the presence of carbonate rocks, representing potential karst aquifers that have surface or near-surface exposure.
Carbonate rocks are present on all continents (Table 4). The largest absolute surface area is found on the largest continent, Asia, where 8.35 million km2 continuous, discontinuous or mixed carbonate rocks are present, corresponding to 18.6% of Asia’s land surface. The highest percentage of karst is present in Europe, where 15.2% of continuous and 6.6% of discontinuous and mixed carbonate rocks were mapped, resulting in a total area share of 21.8% and an absolute area of 2.17 million km2. Substantial amounts of carbonate rocks are also found in North America (19.6% or 4.43 million km2) and Africa (13.5% or 4.05 million km2), whereas smaller percentages of carbonate rocks are present in Australia and Oceania (6.2% or 0.50 million km2) and South America (4.3% or 0.77 million km2).
As a next step, the distribution of carbonate rocks in the 20 largest countries in terms of surface area (A1–A20) and population (P1–P20) was analysed (Table 5). According to this analysis, China and Russia are the countries with the largest and nearly identical absolute karst surface areas, 2.55 and 2.51 million km2, respectively, corresponding to 26.5 and 14.7% of the land surface. Indeed, among the ten largest countries (A1–A10), China has the largest area percentage of karst, but USA and Canada also have high percentages of carbonate rock outcrops (21.3 and 16.6%, respectively). Among the other countries on this list, Iran has the largest area share (54.3%), followed by Egypt (45.2%), Nigeria (30.5%), Ethiopia (29.9%) and Vietnam (27.2%). At the other end of the scale, Sudan, Chad and Japan have no or very little identified karst on WOKAM. Despite the low percentage of carbonate rocks in Japan (0.1%), this country has some well-developed karst such as the Akiyoshi Plateau with more than 400 caves, and several large karst springs. This example illustrates that low percentage values do not mean absence of locally significant karst landscapes and aquifers.
The previous WOKAM paper (Chen et al. 2017a) included a similar table for all European countries, some of which have even higher percentages of total carbonate rock areas such as Montenegro (80.1%), Bosnia and Herzegovina (60.5%), Slovenia (49.5%) and France (35.0%). France also has the highest absolute carbonate rock surface area (0.19 million km2) of all European countries with the exception of Russia, which extends over Europe and Asia. The European part of Russia alone has more carbonate rock surface areas (0.45 million km2) than any other European country.
Karst in plains, hills and mountains, and coastal karst
The occurrence of carbonate rocks/potential karst aquifers in different types of landforms was also analysed. The simplest and most pragmatic classification of landforms at global scale is into plains, hills, and mountains. These three landforms are classified using slope class and relative relief, which are calculated based on a global digital elevation model with 250-m resolution (Sayre et al. 2014). According to the analysis, 31.1% of all karst occurs in plains, 28.1% in hills and 40.8% in mountains (Table 6).
In Australia and Oceania, 55.3% of all mapped karst occurs in plains, with the Nullarbor Plain in South Australia as the most prominent example. Very large karst plains also occur on the North American continent, where 44.2% of all karst forms plains such as the extensive lowland karst areas in the eastern and central parts of Canada, the USA, and Mexico. Figure 2 shows two prime examples of karst plains: Florida (USA), and the Yucatan peninsula in Mexico.
In South America, more than two thirds (68.5%) of all karst occurs in mountainous areas, such as the spectacular alpine karst systems in the Peruvian Andes. Asia also has a high percentage of mountain karst, 59.6%, including the mountainous karst areas in Iran, southern China and Southeast Asia. Although the percentage of mountain karst is a bit lower in Europe (28.9%), mountainous karst aquifers are very important freshwater supplies for many European countries and cities such as Vienna in Austria and Rome in Italy (Kresic and Stevanovic 2010).
As a large proportion of global population lives in coastal areas, coastal aquifers are particularly important freshwater supplies. At the same time, these aquifers are particularly sensitive to natural seawater intrusion (Arfib et al. 2007; Pinault et al. 2004), exacerbated by overexploitation resulting in the upconing of saline water. For the analysis on coastal karst, a coastline map at a 1:10 million scale from WHYMAP was used and a 5-km buffer (corresponding to 0.2 mm on the 1:40 million map) was applied to identify coastal karst. According to the analysis, the global total length of coastal carbonate rocksper karst is about 151,400 km, about 15.7% of the total global coastline apart from Antarctica (Table 7). It is clear the obtained absolute length depends on the scale of the coastline map, while the percentage is assumed to be a relatively reliable estimation. According to the analysis, about one quarter of these coastal carbonate rocks occur in the Canadian Arctic and Hudson Bay, with a total length of 37,167 km, although scattered on several islands and sections of the continental coastline, mostly far from human population and with largely unknown karst and aquifer properties. Important examples of well-known coastal karst include the Dinaric Karst along the Adriatic Coast (2,707 km), Florida (2,220 km), the Yucatan Peninsula in Mexico (1,807 km) (the latter two are shown in Fig. 2), the Nullarbor Plain in South Australia (1,507 km) and the Mediterranean coastline of Libya and Egypt (982 km). The thousands of small islands of Micronesia and other groups of carbonate islands would require a more detailed analysis.
Karst and climate
Carbonate rocks are present in all climatic zones, and the type and degree of karst development is influenced by the climatic conditions (Ford and Williams 2007). The intensity of karst processes can be expressed by the denudation rate, i.e., millimeters of carbonate rock that are dissolved per thousand years (mm/kyr). Field observations from all over the planet, complemented by theoretical and experimental work, have shown that the denudation rate depends on the climatic water balance, i.e., precipitation (P) minus actual evapotranspiration (ETa). Denudation rates often range between 10 and 100 mm/kyr but can be close to zero under extremely dry conditions and exceed 200 mm/kyr under extremely humid conditions (Gabrovsek 2009). However, as climate changes on different time scales, karstified carbonate rocks can also be encountered in zones where the present climatic conditions do not favor intense karst development: for example, the karst aquifers in Saudi Arabia and other arid regions, which are characterized by high rates of water use but very limited or no present-day recharge (“fossil groundwater”), resulting in chronic overexploitation and rapidly declining water tables (“groundwater mining”; Dirks et al. 2018; Schulz et al. 2016).
The relation between karst and climate is not straightforward, because every karst system is the result of a long and complex geologic and climatic history. Nevertheless, it is useful and interesting to analyze the occurrence of karst in different climatic zones. The Köppen-Geiger climate classification system is one of the most commonly used and has been updated and modified several times (Beck et al. 2018). It subdivides climates into five main groups on the basis of seasonal precipitation and temperature patterns: tropical, arid, temperate, cold, and polar (which also includes high mountain climates). These groups can be further subdivided into various climatic types, but for simplicity the statistical analysis focused on the five main groups. Results are displayed in Table 8 and on the map in Fig. 4.
The analysis revealed that 6.92 million km2 or 34.2% of all carbonate rocks are located in arid climates, followed by 28.2% in cold climates and 15.9% in temperate climates, whereas only 13.1 and 8.6% occur respectively in tropical and polar climates (Table 8). These percentages were calculated by dividing the karst surface area in the respective climate by the total global karst surface area.
The proportion of karst areas in each climatic zone was calculated by dividing the karst surface area in the respective climate by the total surface area of this climatic region (not shown in Table 8). The highest percentage was identified in temperate climates, where 19.1% of the land surface consists of carbonate rocks, followed by cold (16.8%) and arid (14.8%) climates, whereas only 8.8 and 7.7% of the land surface in the tropical and polar zones consists of carbonate rocks.
As a hypothesis, this distribution could be explained by the relatively good preservation of carbonate rocks in temperate, cold, and arid climates, compared to high chemical dissolution rates in tropical climates (see discussion by Ford and Williams 2007), and the widespread erosive removal of most sedimentary cover by the action of Pleistocene glaciers in the polar zone. However, the uneven distribution of rock types as a result of contrasts in local geologic history is a simpler and probably sufficient explanation.
The largest absolute areas of tropical karst occur in Africa (1.05 million km2) and Asia (0.67 million km2). Asia also has the largest carbonate rock areas in arid (3.01 million km2), temperate (1.47 million km2), cold (2.37 million km2) and polar climates (0.82 million km2), which also includes the Himalayas and Tibetan Plateau.
For water resources assessment, precipitation is the most relevant climatic parameter. Groundwater recharge can be expressed as specific flow rates (in L s−1 km−2), effective rainfall (P-ETa, in mm years−1) or percentage of precipitation. Several studies report relatively high percentages of recharge in karst areas, due to thin soils and open fractures in the epikarst (Chen et al. 2018; Malard et al. 2016). However, determination of actual evapotranspiration and recharge are associated with high uncertainties in karst areas, where the extensions of recharge areas and, consequently, water balances are difficult to define (Hartmann et al. 2012). Therefore, for pragmatic reasons, the global distribution of karst aquifers was analyzed with respect to annual precipitation using four classes, ranging from arid to very humid:
<5 L s−1 km−2 (<160 mm years−1), in arid and semi-arid areas
5–20 L s−1 km−2 (160–630 mm years−1), e.g., in Mediterranean and temperate areas
20–50 L s−1 km−2 (630–1,600 mm years−1), in humid areas
>50 L s−1 km−2 (>1,600 mm years−1), in mountainous and very humid tropical areas
Figure 5 shows the distribution of these four precipitation zones in karst areas (from light blue to dark blue) and in nonkarst areas (from light grey to dark grey). This map is a first step toward more accurately quantifying the role of carbonate rocks and karst aquifers in the global water cycle.
Karst and population
In addition to climate change, population growth is a major challenge to water resources management (Vorosmarty et al. 2000). Therefore, the relations between karst and population were analyzed. Results are shown in Table 9 and on the map in Fig. 6. The analyses revealed that, in 2015, a total number of 1.18 billion people lived on karst areas, corresponding to 16.5% of the global population. Since 2015, the global population has further increased to an estimated 7.7 billion people, so the best estimate of the current population (2019) living on karst is 1.3 billion. This number is about twice as high as the number of people using freshwater from karst aquifers recently estimated by Stevanovic (2019), but this is not necessarily a contradiction. The two numbers refer to different quantities and are not directly linked to each other, because geologic boundaries often do not coincide with water-supply boundaries.
The largest absolute number is in Asia, where 661.7 million people (15.1%) live on karst, with a regional focus on China, as it can be seen on the map in Fig. 6. The highest percentages are in Europe and North America, where 25.3 and 23.5% of the populations live on karst areas.