Trends in maar crater size and shape using the global Maar Volcano Location and 1 Shape (MaarVLS) database

8 A maar crater is the top of a much larger subsurface diatreme structure produced by 9 phreatomagmatic explosions and the size and shape of the crater reflects the growth history of 10 that structure during an eruption. Recent experimental and geophysical research has shown 11 that crater complexity can reflect subsurface complexity. Morphometry provides a means of 12 characterizing a global population of maar craters in order to establish the typical size and 13 shape of features. A global database of Quaternary maar crater planform morphometry 14 indicates that maar craters are typically not circular and frequently have compound shapes 15 resembling overlapping circles. Maar craters occur in volcanic fields that contain both small 16 volume and complex volcanoes. The global perspective provided by the database shows that 17 maars are common in many volcanic and tectonic settings producing a similar diversity of size 18 and shape within and between volcanic fields. A few exceptional populations of maars were 19 revealed by the database, highlighting directions of future research to improve our 20 understanding on the geometry and spacing of subsurface explosions that produce maars. 21 These outlying populations, such as anomalously large craters (> 3000 m), chains of maars, 22 and volcanic fields composed of mostly maar craters each represent a small portion of the 23 database, but provide opportunities to reinvestigate fundamental questions on maar formation. 24 Maar crater morphometry can be integrated with structural, hydrological studies to investigate 25 lateral migration of phreatomagmatic explosion location in the subsurface. A comprehensive database of intact maar morphometry is also beneficial for the hunt for maar-diatremes on other planets.

as well as any trends between the landform and local and regional influences such as volcanic 39 setting, hydrology, and topography. To recognize universal characteristics of craters formed by 40 subsurface phreatomagmatic explosions, a global database of young maar crater size and 41 shape was created: Maar Volcano Location and Shape (MaarVLS). The size of maars have 42 been previously described in a few studies, but the source data for larger datasets was 43 unavailable (Cas and Wright, 1987), the measurements are limited to a few isolated craters 44 (Nemeth et al., 2001), or were limited to highly eroded craters or diatremes (Martín-Serrano et 45 al., 2009). This database contains the data of maar sizes and shapes from multiple eruptive 46 fields from a range of volcanic settings. This study focuses on pristine morphology and thus 47 does not contain morphometric data from all known maars; however, the discussion here also 48 considers other recognized features from the literature to apply morphologic observations to our 49 understanding of maar formation. This global survey of maars enables the study of bigger 50 picture processes of these volcanoes including influences on crater growth, implication of crater 51 shape, and the diversity of features within in and between volcanic fields. 52 Craters on Seward Peninsula AK, USA are the largest craters in the database (4-5 km diameters). C) Jabal 80 Simagh of the Harrat Kishb field in Saudi Arabia has an AR values of <0.5. D) Nejapa and Ticomo maars in 81 Nicaragua have elongation values of <0.45. E) Lake Leake of Newer Volcanic Province, Australia, and F) 82 Hora Lake Bishoftu Volcanic Field, Ethiopia show off the range of isoperimetric circularities in the 83 database. Images courtesy of Google and Digital Globe and CNES/Astrium. 84 The bulk of the explosive activity and deposits of a maar-diatreme occur in the 86 subsurface. As such, for young maars, the eruption, including depth and lateral position of 87 explosions, can only be reconstructed from the deposits that were successful ejected from the 88 crater to reach the tephra ring, and the shape of the crater. Maar craters grow through a 89 combination of explosive excavation and collapse (White and Ross, 2011;Sonder et al., 2015;90 Graettinger et al, 2016). The crater rim is only partly a constructional feature and therefore 91 crater shape, as measured here, is less susceptible to influences of outer slope stability and 92 wind than for scoria and tuff cones (Kereszturi et  List of references used to populate the database for a given crater 1 Polygon or shape file to indicate if the data was collected from Google Earth or using individual images in Arc GIS respectively. 2 Based on available literature, to be expanded in later versions. Elevation is measured by the level of the lake or low point of the crater. 141 MaarVLS provides a global perspective on maar craters and highlights potential for 142 comparative studies between multiple volcanic fields. This study identifies the unique 143 morphometric characteristics of maars that can be used to distinguish them from other similar 144 negative landforms such as kettle and permafrost lakes, impact craters, karst features and 145 volcanic collapse pits, and can ultimately be used to identify similar volcanic features on other 146 planets, such as Mars (Graettinger, 2016). .kml file and evaluated for morphometric analysis. Google Earth uses a cylindrical projection 156 that has significant warping at the poles. This first version of the database only includes craters 157 above 60 degrees latitude when alternative datasets such as Advanced Spaceborne Thermal 158 Emission and Reflection Radiometer (ASTER) imagery were already available in the author's 159 collection. Future versions of the database will take advantage of publicly available ASTER 160 imagery and other open datasets to include a larger population of maar craters at high latitudes. 161 All craters are Quaternary in age and have complete, or near complete (>75%), rims with limited 162 incision by erosion ( Table 1). Craters that have interacted with scoria cones or lava flows were 163 generally avoided, unless the 75% unobstructed crater rim criterion was satisfied. Compound 164 craters with > 3 separated basins are not included in the first version of the database due to the 165 high level of interpretation required for digitization (i.e. Katwe Volcanic Field, Uganda; Murray 166 and Guest, 1970). Modification by human activity is common for many of the volcanic fields 167 studied. When human activity made an obvious impact on the crater rim (i.e. quarrying), the 168 crater was not included in the database. 169 Craters were outlined manually from visible Google Earth imagery, ASTER images and digital 170 elevation models to produce polygons encompassing the crater along the rim. Crater outlines 171 were completed by four individuals and evaluated by one researcher for consistency. Polygons 172 of crater outlines were used to determine area, perimeter, and length of major and minor axes. 173 An average of the two axes is used as average diameter in this study. Shape parameters were 174 derived for each crater from these measurements. Shape parameters used in this study 175 describe the two-dimensional shape of the outline of the crater from the digitized polygon. 176 These include dimensionless ratios: aspect ratio, elongation, and isoperimetric circularity. 177 Aspect ratio (AR) is defined as the ratio of a crater's diameters: where Dminor is the length of the crater's minor axis and Dmajor is the length of the crater's 180 major axis. Here the minor axis is measured as the axis perpendicular to the major axis running 181 through the center point. An aspect ratio of 1 represents an equant shape around the center 182 point; as the disparity between the two axes increases, the aspect ratio decreases away from 1. 183 where A is the area encompassed by the crater rim as defined by the digitized polygon. 186 Elongation compares the area of a circle with the diameter of the major axis to the maar area. A 187 circle has Elongation equal to 1 and more elongate shapes have smaller values. Elongation 188 differs from Aspect Ratio as it better describes asymmetrical shapes, in fact, for ellipses the two 189 values will be the same. 190 Isoperimetric Circularity (IC) is defined as the area of a crater polygon divided by the 191 area of a circle with the same perimeter. 192 where A is the area encompassed by the crater rim and P is the perimeter of that same polygon. smallest maars (<200 m, <3% of database) occur in close proximity (<600 m) to other maar 226 craters including the Ukinrek West crater (Self, 1980) and Crater Z at Ubehebe (Fierstein and 227 Hildreth, 2017). In several cases these small craters are interpreted to be part of the same 228 eruptive sequence as the adjacent craters (Self, 1980;Fierstein and Hildreth, 2017). Very large 229 craters (>3000 m diameter and area >7 km 2 ) represent only 4% of the maar population. curvature. There is no apparent trend with size for any of the shape parameters (Fig. 3). There 270 is not a diagnostic crater shape related to craters produced by multiple co-located eruptions or 271 long-lived polygenetic eruptions. This relationship will need to be reevaluated as more dates 272 become available for these volcanic systems. 273 These shape values can be compared with ellipses that vary by the relationship between 274 the major and minor axis (Fig. 4) where 71% of craters occurred in fields with more than five maars. While roughly half of the 296 maars in the database occur in intraplate volcanic settings, they are also found in back arc 297 basins along subduction zones, continental rifts, on ocean islands above hot spots, and less 298 commonly in convergent or transpressional environments. 299 Maars in the database occur at sea level to elevations as high as 4000 m above sea 300 level (asl). At higher elevations the number of documented maars decreases, with most (90%) 301 below 2000 m asl (Fig. 3). Maars in the database cover a range of latitudes, but do not have 302 even distribution across all latitudes (Fig. 2). Maars above 200 m all occur between -30 and 40 303 North latitude. A comparison of crater diameter with distribution reveals that small craters 304 (<1000 m) occur globally at all elevations (Fig. 3), however, all exceptionally large maar craters 305 (diameter >3000 m, area >7 km 2 ) occur at elevations below 500 m asl. Crater shapes do not 306 present a clear trend with latitude, but isoperimetric circularity does increase (craters are more 307 circular) with increasing elevation (Fig. 3). 308

Fields 309
Quaternary volcanic fields with maar volcanoes contain anywhere from one to tens of 310 maar craters. In MaarVLS several fields are currently represented by only a sample of maars 311 due to limitations in available imagery, and the complete crater rim criterion. For volcanic fields 312 with five or more included maar craters, the size variability within individual volcanic fields is 313 high, but lower than the database as a whole (measured in meters; total stdev=861, for fields 314 with maars stdev=395; Table 3). Within a volcanic field, craters will typically fit between a 315 minimum and maximum crater diameter ratio of 0.36, meaning that the largest crater is less 316 than twice the diameter of the smallest crater. The shapes of craters within these volcanic fields 317 have similar average shape parameters, but narrower ranges than the overall database (Table  318 3). 319 320 321  The database also provides an opportunity to investigate the distribution of maar craters 323 relative to population centers (Table 4)  between crater size, shape and distribution with age were evaluated. There is no apparent 337 correlation between maar crater age with latitude, elevation, diameter, elongation or 338 isoperimetric circularity (Fig. 3). 339  The rhyolitic maars are all larger than 1000 m in diameter, but are not distinctively larger than 347 mafic maars as a population (Table 5). Intermediate magmas form maar craters 360-1400 m 348 across and fit within the scatter of mafic maar sizes. The shape of intermediate and rhyolite 349 maars is typically more circular and less elongate than mafic maars, but not enough to be 350 diagnostic. The largest craters (>3000 m) are limited to mafic magma compositions. Therefore, 351 based on this population, composition cannot be determined solely from crater size or shape. 352 Table 5: Comparison of size and shape trends for maars with known magma compositions. separating topographic lows are rare (or rarely preserved), and the organization of the 359 overlapping circles is variable across maars and volcanic fields with maars (Fig. 1). A few 360 anomalous populations of maar size (exceptionally large) and morphology (crater chains) stand 361 out against the main database characteristics. The database also highlights that while maars 362 typically represent a fraction of the volcanic constructs within a volcanic field there are a few 363 notable exceptions with abundant maars. Further, the maars studied occur in a wide range of 364 volcanic field types and tectonic settings reinforcing that while the conditions that form maar 365 volcanoes are specific, they are not limited to only one environment. Further, the global 366 distribution of maars highlights the proximity of numerous maar fields to major population 367 centers. In order to evaluate the potential for interpreting subsurface maar forming processes, 368 namely explosion location and number, from crater shape it is necessary to evaluate post-369 eruption modification, the completeness of the sample population, and the exceptional maar 370 populations mentioned above. 371

Role of post-eruption modification 372
Investigations of crater modification from the 1977 eruption of Ukinrek in Alaska revealed 373 a rapid increase in the major and minor axes of the crater and infill of the crater floor initially 374 after the eruption and stabilization with time (Pirrung et al., 2008). The shape of the crater 375 however, as measured by aspect ratio, was maintained. This suggests that absolute crater 376 diameters, depth, and internal slopes are susceptible to modification by erosion, but crater 377 shape is more stable over time. As the inclusion criterion for MaarVLS excluded maar craters 378 such as Kilbourne Hole, New Mexico, and Fort Rock, Oregon where the crater rim was 379 interrupted or missing, the shapes within the database are assumed to represent post-eruptive 380 shapes. Furthermore, comparison of Aspect Ratio, Elongation, and Isoperimetric Circularity for 381 those maars in the MaarVLS database with age indicates that there is no trend in crater shape 382 with age for Quaternary maars (Fig. 3). 383 Unlike scoria cones, which are constructional features with steep slopes, maar craters 384 have low angle tephra rings that extend away from the crater with the main structure cut into the 385 ground surface. When maars are erupted on complex topography this tephra ring will roughly 386 drape the surrounding topography (e.g. Dotsero, Leat et al., 1989; Bea's Crater, Amin and 387 Valentine, 2017). The crater is the result of excavation and collapse with limited deposits 388 escaping the crater to form the low angle tephra ring . The shape of the 389 crater is therefore less susceptible of the influence of outer slope stability and wind than for 390 scoria and tuff cones (Kereszturi et al., 2012;Kervyn et al., 2012). The low slope angle of the 391 tephra rings enables agricultural activities with less earth works than scoria cones or lava flows, 392 and consequently many roads and farms merely mantle the tephra ring deposits preserving 393 crater rim morphology. Based on these observations, for maars included in the database it is 394 reasonable to assume the crater shape, and to a lesser degree crater size, is dominated by a 395 signature of crater growth by eruptive and syn-eruptive processes. 396

Completeness of coverage 397
MaarVLS contains craters from a range of latitudes, with fewer craters at latitudes 398 greater than 60 degrees. High latitudes in both hemispheres are under-sampled as additional 399 imagery is required (due to warping of projections at the poles in WGS84 used in Google Earth) 400 and will be used to produce future versions of the database. The southern hemisphere is 401 represented by 64 craters, with limited craters from -10 and -20 degrees (Fig. 2). As the 402 southern hemisphere contains ~30% of the continental crust on Earth, the database has a 403 roughly proportionate distribution of maar craters between the hemispheres. Maar craters are 404 observed at a wide range of elevations (0-3500 m asl), but 60% of maars are at elevations of 405 1000 m or less, with half of that being at elevations below 250 m. As 75% of elevations above 406 sea level on Earth are less than 1000 m (Eakins and Sharman, 2012) the low abundance of 407 maars at high elevation is to be expected (Fig. 2). The number of craters included in the 408 database and the large range of elevations and latitudes on Earth provides a sufficiently diverse 409 population to establish what is typical of maar crater size and shape (Section 3.1) to recognize 410 any exceptional populations of maars on Earth (Fig. 2-3). 411

Unique populations 412
Very large craters 413 Of the shape parameters evaluated, diameter (Fig. 2) and area are the only parameters 414 to highlight a distinct outlying population. Very large maars occur in six volcanic fields that are 415 globally distributed, but all occur at low elevations (Fig. 5). The Espenberg maars (Beget et al., 416 1996)  and Ticomo maars in Nicaragua south of Lake Manuagua, occur along a linear trend that could 456 be extended to include the Xiola maar to the north (Fig. 1d) Qal'eh Ali field in Iran (Milton, 1977), all of these other fields occur in extensional settings with a 513 range of climates. This strongly suggests that while the availability of water is significant, the 514 tectonic setting and the subsurface structure are critical to the conditions leading to maar-515 forming eruptions. The MaarVLS database highlights the importance of lateral growth of craters in more 586 than one direction supporting field-based observations that lateral explosion location is common 587 and fundamental to the evolution of maar-forming eruptions. Additional work to relate shape with 588 host rock properties, regional faults and local hydrology is planned to further isolate the 589 influences on this lateral crater growth. Exceptional populations of large size craters, maar-590 dominated volcanic fields, and crater chains warrant further study as they have the potential to 591 provide unique insight into the role of regional structures, ground ice, lateral migration, and co-592