Paucity of legacy oil and gas subsurface data onshore United Kingdom : implications for the 1 expansion of low carbon subsurface activities and technologies ” 2 3

30 The decarbonisation of energy systems to achieve net zero carbon emissions will likely see the rapid 31 development of carbon capture and storage, energy storage in the subsurface and geothermal energy 32 projects. Subsurface data such as seismic reflection surveys and borehole data are vital for geoscientists 33 and engineers to carry out comprehensive assessments of both the opportunities and risks for these 34 developments. Here, for the first time, legacy subsurface data from onshore oil and gas exploration in 35 the UK is collated and analysed. We provide a description of the spatial coverage and a chronology of 36 the acquisition of key seismic reflection and borehole data, as well as examine data resolution and 37 limitations. We discuss the implications of spatial variability in subsurface datasets and the associated 38 subsurface uncertainty. This variability is vitally important to understanding the suitability of data for 39 decision making. We examine societal aspects of data uncertainty and discuss that when the same data 40 are used to communicate subsurface uncertainty and risk, the source of the data should also be 41 considered, especially where data is not easily publicly accessible. Understanding the provenance of 42 data is vitally important for future geoenergy activities and public confidence in subsurface activities. 43

Achieving a transition to net zero carbon emissions from energy systems is one of the most 47 pressing challenges facing society globally (Rogelj et al. 2015). The UK government has set a legally 48 binding target to reduce greenhouse gas emissions to net zero by 2050 (see Climate Change Act 2008), 49 which to achieve will require decarbonising both industrial and residential energy systems (e.g. Broad 50 et al. 2020;Cooper and Hammond 2018). There will likely be a need for subsurface activities, whether 51 as part of industrial clusters and the development of carbon capture and storage (CCS) (e.g.Alcalde et 52 al. 2019) or as part of decentralised energy systems and the use of geothermal energy (Lloyd 2018). 53 The exploration and production of shale gas has raised concerns not only about the compatibility with 54 low carbon energy systems and mitigating climate change (Partridge et al. 2017) but also the ability to 55 predict the subsurface as a result of induced seismicity (Bommer et al. 2015). A fundamental question 56 regarding the use of the subsurface for future decarbonisation pathways is whether there is enough 57 suitable data to assess the potential contribution and impact of subsurface activities and their role in a 58 net zero future.  In communicating subsurface risks, experts often discuss the degrees of uncertainty inherent in 81 subsurface characterisation, however this is often without considering the target audience. Importantly, 82

Subsurface UK data 118
The geology of the onshore UK contains a geological record all the way back to the Archean, 119 and includes a history of subduction zones, volcanic arcs, continental rifts and mountain belts 120 (Woodcock and Strachan 2012). While extensive geological mapping of the UK dates back to the 19 th 121 century and is summarised in the now famous map by William Smith (Smith 1815), it was not until 122 1918 that the first deep oil and gas well was drilled, Hardstoft-1 in Derbyshire, to a depth of ~950m 123 (Morton 2014). In the period preceding the Second World War (1939), there were a number of early 124 seismic reflection experiments by the then Anglo-Iranian Oil Company (Jones 1937). However, it was 125 not until after the war and the 1950s that geophysical data acquisition began in earnest for oil, gas and 126 coal exploration. Since these early investigations, seismic reflection surveys have become the primary 127 subsurface geophysical method employed for oil and gas exploration. Onshore UK both the acquisition 128 of seismic reflection data, and the drilling of deep boreholes continued, with the late 1980s being the 129 peak of onshore seismic data acquisition (see section Seismic Reflection for details). Much of the 130 current understanding of the subsurface onshore the UK comes the data that has been acquired by the 131 oil and gas industry. While undoubtedly this data has advanced our understanding of the geology in the 132 UK, there has been little consideration in the literature of the implications of the source of this data for 133 public trust and perceptions of risk. This study focuses on the coverage of both seismic reflection 134 surveys and boreholes, as well as the characteristics of the data. The data included in this study are those 135 held by the UK Onshore Geophysics Library (UKOGL), the British Geological Survey (BGS), and the 136 Oil and Gas Authority (OGA), the sources of which are listed in Table 1. Detailed description of the 137 data can be found in subsequent sections, but as an overview, there are ~76 136km of 2D seismic 138 reflection data, ~2400km 2 of 3D seismic reflection data and 2242 oil and gas exploration boreholes. 139 While the publicly available OGA borehole dataset contains records for 2242 wells (not including wells 140 completed in 2018 and 2019), the UKOGL borehole records also include deep wells from other 141 activities such as coal mining. There is no single consolidated record of all onshore boreholes, therefore 142 the present analysis has used records from the BGS, OGA and UKOGL to provide a comprehensive 143 view. Throughout this study the term "borehole" is used to refer to both shallow and deep wells or wells 144 where total vertical depth (TVD) is not specified. For the BGS Geothermal Catalogue, the data had to 145 be digitised into a tabulated format from the scanned PDF file format available directly through NERC. 146 The location of data is specified either to 10 m or 100 m. In addition, there is no unique common well 147 identifier that allows the data in the catalogue to be matched spatially with the BGS Borehole Records. 148 Where comparisons have been made to the offshore areas of the UK Continental Shelf, the data 149 used was the "Surveys as Consented 2D", which is available from the OGA National Data Repository. 150 For the comparisons made with data from the Netherlands, the data is from NLOG, which is manged 151 by the Geological Survey of the Netherlands on behalf of the Ministry of Economic Affairs and Climate. 152 153

Methodology 154
This study describes the spatial distribution and characteristics of data available from the BGS, 155 OGA and UKOGL (Table 2). It used primarily geological and geophysical data collected for oil and 156 gas exploration and production. Geological parameters and concepts are often not directly observed or 157 measured, but interpreted from these data (Pérez-Díaz et al. 2020), In this context, this study considered 158 data to be measurements in wells, and post-stack seismic reflection data (see Table 1 and global Moran's I), the extent and distribution of geospatial subsurface data have been quantified. 163 The results of this analysis have then been assessed in terms of the possible impact on subsequent 164 interpretation and analysis. All spatial statistics were computed in ArcMap 10.6.1. The quadrant 165 analysis point density and total line length was computed for a given area of 10km by 10km area (area 166 of 100km 2 ). While for individual well locations point density was used, for the BGS Geothermal 167 Catalogue, where individual wells have more than one measurement, a quadrant analysis was used. For 168 spatial statistics, the P-value is used to assess if there is spatial pattern among the features and therefore 169 the probability that the observed spatial pattern was random. A small P-value is indicative of a low 170 probability that the observed spatial pattern is the result of random processes. The Z-values are standard 171 deviations, for example for a value of 2.5 the result is 2.5 standard deviations. The study has 172 differentiated between shallow and deep wells based on true vertical depth (TVD), a deep well being 173 defined as one completed to a depth >300m. 174 These statistics have then been used to asses the distribution of the legacy data and discuss the 175 suitability of for future use of the subsurface, specifically geothermal and unconventional hydrocarbon 176 extraction. The coverage of both well and seismic reflection data have been analysed with respect to 177 the domestic and non-domestic heat demand to assess the data available for geothermal resource 178 characterisation in demand hot spots. The study has used heat demand data for the year 2009 from 179 Taylor et al. (2014). The original data is annual heat demand provided at a 1km by 1km resolution and 180 in units of kWh/km 2 . In this study, the data data were reduced, using an aggregated mean, to a 5km by 181 5km resolution to simplify the boundaries of heat demand, and then converted to MWh/km 2 . These data 182 have been used to compare areas of heat demand to the distribution of subsurface data. There is a difference of 643 between the UKOGL data and the OGA records which reflects that only 208 those specifically identified as oil and gas exploration and production boreholes are included in the 209 OGA records, while the UKOGL includes other deep boreholes. Oil and gas exploration and production 210 boreholes account for less than 1% of all the boreholes drilled in the UK. The spatial density of the 211 shallow boreholes between 30 and 500m onshore the UK can be seen in   For the wells in UKOGL database, nearest neighbour analysis estimates a Z-score of -82.77, indicating 221 that the data are clustered and there is a less than 1% likelihood that this clustered pattern is random. 222 Global Moran's I analysis, indicates that wells are clustered with respect to depth, with a Z-score of 223 53.57, and less than 1% likelihood that the clustering is random. A histogram of wells drilled onshore 224 the UK by year shows that over ~70% of the onshore wells in the UK were drilled prior to 1990. Since 225 3D seismic reflection data acquisition onshore UK did not start until the 1990s, that means that all these 226 wells were drilled based on 2D seismic reflection data. As would be expected there is a spatial 227 coincidence of both boreholes and seismic reflection data. A total of 644 boreholes are co-located with 228 3D seismic reflection data, and 1578 wells located within 100m of a 2D seismic reflection line. 229

Core and downhole log data 230
The BGS maintain a database of over 10 000 onshore borehole samples, which comprises a 231 range of materials including core, core samples, individual hand specimens, bulk samples, unwashed 232 cuttings, washed and dried cuttings, plugs, powders and bulk samples, including those collected as part 233 of onshore oil and gas exploration and production borehole drilling. The relative spatial density of these 234 data can be seen in Figure 6a. This database can be searched online. The BGS hold an archive of digital 235 geophysical downhole log data from boreholes distributed across the UK. Basic well information, such 236 as location and spud and completion date, is also held by UKOGL, but access to digital log data is 237 through formal release agents. There is no single record of all downhole logs onshore UK. The BGS 238 hold a record of ~5963 wells with digital geophysical logs, which includes both oil and gas exploration 239 wells and other boreholes including mine gas and coal bed methane wells. The spatial density of these 240 data is shown in Figure 6b. In addition to the BGS records of geophysical logs, well data is available 241 through the OGA's appointed data release agents, who hold an inventory of digital log data for onshore 242 wells. However, the type of data available and the quality vary from well to well and the exact nature 243 and number of wells is a commercial product. 244

Temperature data 245
The BGS Geothermal Catalogue is a published compilation of temperature and heat flow 246 measurements from across the onshore UK. Figure 7a shows the location of individual wells with 247 temperature measurements and Figure 7b shows the number of temperature measurements in a 10km 248 by 10km quadrant. Average nearest neighbour analysis returns an observed mean distance of 1668m 249 compared with an expected mean distance of 9538m. This returns a nearest neighbour ratio of 0.1879, 250 with Z-score of -60.31 and less than 1% likelihood that this is random indicating that the data are 251 strongly clustered. Global Moran's I analysis, indicates that location of temperature measurements are 252 clustered with respect to depth, with a Z-score of 35.303, and less than 1% likelihood that the clustering 253 is random. As well as spatial clustering, the measurements of temperature in the boreholes are also over 254 a limited depth range. As described by Rollin (1995), there are ~2600 temperatures at over 1150 sites. 255 Of these, geothermal gradients are estimated in the dataset for ~1700 measurements. Over 90% of the 256 temperature data are from depths less than 2000m and ~27% are from a depth shallower than 500m 257 (Figure 8a). These data indicate that less than 10% of the measurements were made at depths greater 258 than 2km. Figure 8b is a plot of temperature and depth. While the dominant trend is one of increasing 259 temperature with depth, there is no simple relationship. These temperatures in the catalogue are used to 260 estimate geothermal gradients using a modified air surface temperature. These estimates of geothermal 261 gradient were not used in this study, as the method of determining land surface temperature is an 262 oversimplification an not accurate without correction. There are only 116 temperature measurements 263 from depths greater than 2000m. There is a very significant vertical sampling bias, as well as the spatial 264 bias shown in Figure 8a. The location, line length (in the case of 2D) and area (in the case of 3D) of seismic reflection 280 data onshore UK have been analysed to determine the spatial distribution of the data. Three-dimensional seismic reflection data onshore UK is limited to just 32 surveys (Figure 12) 293 covering an area of ~2400km 2 . As a comparison, the Netherlands has a land area of ~ 41 543km 2 across 294 which there is ~14 000km 2 of onshore 3D seismic reflection data. Onshore the UK the largest onshore 295 3D survey is 363km 2 , which is the Lincswold02 3D survey. Using the current (as of April 2020) 296 These prospective areas total ~20 000km, however there has only been 452km 2 of new 3D seismic 302 acquisition in these areas, which amounts to ~2% of the total prospective areas. 303 When the coverage of 2D and 3D seismic reflection data is compared with the domestic and 304 non-domestic heat demand across the UK., only ~500km of the existing 2D seismic reflection data 305 intersect areas of domestic heat demand above 10 000 MWh/km 2 annually. There is no 3D seismic 306 reflection data in these areas. This is <1% of the 2D seismic reflection data. Table 3 summarises the 307 coverage of data and the total length of 2D seismic data and the number of wells within the ten largest 308 areas where heat demand is >10 000 MWh/km 2 . As well as the limited availability of 2D seismic 309 reflection data, there are also only a handful of deep wells and wells with temperature measurements in 310 these areas. Figure 9 shows the four largest areas, London, Birmingham, Manchester and Liverpool, 311 and the coverage of deep data. These data indicate that there is notable paucity of well and seismic data 312 for geothermal exploration in these areas. 313 314

Seismic reflection data quality 315
The study has looked at the quality of 3D seismic reflection data specifically within the PEDL 316  Figure 13 shows how 325 the frequency spectrum for the 3D data varies by depth (in two-way-time [TWT]) of investigation. To 326 examine the impact of frequency content on the quality of the seismic reflection data , Figure 14 shows 327 example seismic sections of the original post-stack seismic volume (Figure 14a) with different high 328 frequency cut offs applied at 90 Hz (Figure 14b), 60 Hz (Figure 14c) and 40 Hz (Figure 14d). The 329 difference (original minus filtered) seismic is shown in Figure 15. Filtering out the high frequency 330 component (>90 Hz) of the 3D survey (Figure 14b) makes almost no difference to the seismic image 331 (Figure 15a), aside from some high frequency noise in the near surface section (upper most 500ms 332 TWT) section. Filtering out the component >60Hz removes some coherent energy above 1500 ms, but 333 below this there is very little difference (Figure 15b). Filtering out >40Hz component results in 334 removing coherent energy in the interval shallower than 1500ms as well as some deeper coherent energy 335 (Figure 15c). In this area, the exploration targets were at ~1000ms. While there is overall a higher 336 frequency content at shallower depths, this does not contribute to improving the overall interpretability 337 of the data and suggests that much of the higher frequency content could be noise rather than coherent 338 energy. Frequency is a key parameter controlling the resolution of faults in seismic images. The 339 maximum vertical resolution is directly related to the ability to distinguish individual reflecting surfaces 340 (Yilmaz 2001) and in the case of the Bowland-12 survey is approximately 60 m at the target intervals. 341 For the horizontal resolution, assuming that the Fresnel zone is reduced to a small circle by 3D migration 342 The ability to create accurate models of the subsurface relies on data being representative of 359 the area of interest. Data acquisition in oil and gas exploration is location biased, and often clustered, 360 because it is acquired to test a geological scenario that may have multiple objectives. This clustering is 361 having been demonstrated through the use of spatial statistics. Onshore oil and gas exploration wells 362 exhibit significant clustering, as do the temperature data that are frequently acquired in these wells. Of 363 the total onshore area of the UK, i.e. ~243 000km², the 76 136km of 2D seismic data covers an area of 364 ~109 900km 2 . This means that just under half of the total onshore area of the UK is covered by a 365 subsurface image. As noted previously, when compared with the offshore of the UK, where in many 366 respects seismic acquisition is easier, there is a relative paucity of both 2D and 3D seismic reflection 367 data. 368 3D seismic reflection data cover a total of 2400km 2 of the onshore UK. The limited extent of 369 any single 3D seismic survey onshore the UK limits the ability to map or extend our geological 370 knowledge and understanding. The largest onshore survey is 363km 2 (Lincswold-02) and is 371 approximately 30km by 12km. Similarly, the limited extent to which surveys are adjacent to one another 372 and form a patchwork from which larger areas can be mapped is in the same location where the 373 Lincwold-02 is adjacent to and overlaps with the Saltfleetby-99 survey and together cover ~380km 2 . 374 Despite the UK Government encouraging and overseeing shale gas exploration and the and numerous 375 companies embarking on shale gas exploration programmes (see Selley 2012) there has been only 376 638km 2 of 3D seismic reflection data have been acquired across ~20 000km the prospective areas since 377 2010. Overall, the paucity of 3D seismic data onshore the UK limits the ability to interpret geological 378 structure and trends beyond a handful of areas. Despite the critical role that 3D seismic reflection data 379 have in exploration and exploitation, and their importance in future geoenergy activities such as CCS, 380 there is a limit to their resolution and therefore the features that can be resolved to characterise the full 381 complexity and heterogeneity of the subsurface. For future geoenergy projects, a consideration could 382 be that operators should report the parameters and resolution of their seismic reflection surveys ahead 383 of consents being given, for example to hydraulically fracture. 384 As is now well documented, induced seismicity felt by the local population has been associated 385 with both shale gas sites in the UK where hydraulic fracturing has been carried out ( drilling or hydraulic fracturing it would have been possible to report that the data would not be suitable 406 for interpreting faults with either vertical (throw) or horizontal (heave) displacements below the 40m 407 and 60m estimated resolutions respectively. In addition, it is possible that the resolution of the data is 408 lower than estimated from the seismic frequency because the analysis presented indicated that the higher 409 frequencies in the Bowland-12 3D data do not contribute to the overall interpretability of the data 410 (Figure 15a-c). The interpretation of a fault with a vertical offset of less than 50m would be highly 411 uncertain. The overall interpretability of the 3D seismic reflection data for structural interpretations is 412 limited by the vertical and horizontal resolution of the data. 413 For geothermal energy this study highlights that in areas of high heat demand there is limited 414 existing subsurface data (see Table 3). Both well and seismic reflection data show significant clustering, 415 and the well data also have a sampling bias with respect to depth. The ability to predict subsurface 416 properties, such as temperature, relies on calibrating models against existing data. If the existing data 417 are clustered, and there is a significant sampling bias then making predictions, based on models, away 418 from data rich areas inevitably comes with an increased uncertainty. As discussed for interpretation 419 uncertainties by Bond (2015), the way in which these uncertainties are communicated in geosciences is 420 important from a social and economic perspective because the public are increasingly concerned with 421 the decision-making processes and the risks and uncertainties. 422 The subsurface will likely be required to deliver a low carbon energy transition in the UK, for 423 example the deployment of CCS, energy storage (methane and hydrogen), for the continued, but 424 sustainable extraction of natural resources (Stephenson et al. 2019) and likely vital for long term 425 disposal of radioactive waste. However, our ability to sustainably exploit the subsurface relies on our 426 ability to predict and model it accurately. Given the vintage of much of the existing seismic reflection 427 data, a consideration of future geoenergy projects should be whether existing data are suitable or 428 whether a step change in onshore seismic data quality (and coverage) will be required to both fully 429 understand the opportunity and to demonstrate that activities will have a low impact on communities 430 and the environment. The variability in the extent and quality of existing data across the UK means that 431 decision makers should include an assessment on the suitability of data from the project inception phase. The exploration and production of unconventional hydrocarbons which use hydraulic fracturing 453 methods have brought into sharp focus the challenges in confidently predicting the subsurface. There 454 is typically a larger uncertainty in subsurface interpretations using 2D seismic reflection data compared 455 with 3D seismic reflection data, with reduced uncertainty a function of both improved areal coverage 456 and the benefits of 3D migration (Bacon et al. 2007). The Consolidated Onshore Guidance (Oil and Gas 457 Authority 2018) specifies that "a map and seismic lines showing faults near the well and along the well 458 path" should be included but makes no specific reference to demonstrating the suitability of the 459 underlying data on which those interpretations are made. There is no requirement for the operators to 460 demonstrate that the seismic refection data are specifically suitable for the activity that is being planned. 461 The required information relates to primarily to interpretations (or knowledge). Nevertheless, not all UK regions and communities are equally exposed to subsurface development. That 475 is, there are significant regional and community variations in subsurface development as well as 476 uncertainty surrounding the risks that can be modelled using the data in this analysis. This unequal 477 distribution of subsurface risk is also compounded by various interpretations of risk. Social science 478 research suggests that variations in perceptions of risk are explained by geography, culture, 479 socioeconomic status, ethnicity, race and gender (Flynn et al. 1994). As just one example of the 480 importance of context, consider the case of hydraulic fracturing in Oklahoma (USA), a state highly 481 dependent on oil and gas development. The perceived risks associated human induced seismicity among 482 Oklahoma residents are less of a concern than perceived risks associated with pollution, especially to 483 water and poisoning of livestock (Campbell et al. 2020). Thus, when subsurface data is mapped out 484 across the UK it demonstrates the potential for enormous variation in interpretation of risk according to 485 the spatial location of wells as well as the constellation of community and demographic combinations 486 that may together shape risk perceptions (e.g. Kropp 2018). This distribution of perception of risk has 487 yet to explored in the UK using subsurface data, though ecosystem services suggests there are good 488 reasons to undertake such an analysis in the future. 489 There is an increasing public demand for high quality information that is accurate, consistent, 490 complete, timely and representative (e.g. Wang and Strong 1996). This analysis suggests that seismic 491 reflection and borehole data represent an information source that can be used to contribute to 492 information quality and aid in the communication of subsurface risk. However, simply reporting 493 information, even high-quality information, is probably not enough. In particular, social science 494 research suggests that credible information sources are highly important in conveying actual risk (Renn 495 and Levine 1991). Thus, where data are uncertain or complex the public is likely rely on experts to help 496 them make sense of subsurface risks that may be reflected in those data. As a result, trust in the experts 497 and institutions is likely to have an important impact on general perceptions about risks associated with 498 subsurface development. 499 The interpretation of these data open up an important opportunity for geoscientists to help 500 engage UK citizens about the levels of uncertainty and subsurface risks associated with energy 501 development (e.g. Buchanan et al. 2014). However, with opportunities also come challenges. That is, 502 while this study is one of the first to map the onshore UK subsurface, much of the underlying data are 503 produced by industry. Thus, information presented by geoscientists will be constantly evaluated within 504 the context of industry trust (Seeger et al. 2018;Wachinger et al. 2013;Wray et al. 2008). The challenge, 505 then, is to convey meaningful information about uncertainly and risk when data generated may be 506 viewed as suspicious, especially when it is not publicly accessible. Therefore, one of the biggest 507 obstacles in conveying accurate perceptions of risk to UK residents may rest in the fact that much 508 subsurface data are generated by industry (Wachinger et al. 2013). Such challenges, however, are not 509 usual in risk analysis as researchers find that stakeholders are often perceived to communicate risk 510 through the selective use of data that advances their own interests (Leiss 1995 Despite over a century of subsurface data collection onshore UK, this is the first synthesis of 519 the key datasets that can be used to interpret the geology of the deep subsurface. The study highlights 520 that there is a paucity of both well and seismic data across the onshore UK. All subsurface 521 interpretations, be it for well-established activities such as conventional oil and gas exploration and 522 production, or new activities as part of the energy transition, rely on these geophysical or geological 523 data. These interpretations and models are fundamentally limited by the inhomogeneous datasets and 524 the resolution of them. Onshore oil and gas production in the UK currently accounts for <1% of the 525 total production from the UK (OGA, 2020) and the limited scale of resources, when compared to the 526 offshore, that has restricted further data collection, with companies prioritising the offshore areas of the 527 UK Continental Shelf. The lack of extensive and high-quality data could be a fundamental limitation 528 on the expansion of nascent low carbon subsurface activities and technologies. The attention with which 529 the public are now putting on all new energy activities will require geoscientists to clearly articulate the 530 limitations of currently available datasets, and these limitations should highlight areas where new data 531 collection is needed, both to improve coverage, and to improve resolution. The ability to understand 532 and quantify uncertainties in a subsurface description is key to effectively reducing safety, 533 environmental, health and economic risk. Gaining new knowledge through data acquisition cannot be 534 guaranteed to de-risk a subsurface outcome, however, the new knowledge can be vital in the decision-535 making process. 536 The analysis and statistical measures shown here for the onshore UK subsurface datasets can 537 be used to determine priority areas for future data collection. But the analysis does not address what is 538 enough data for a given activity. There needs to be a concerted effort across geosciences and social 539 sciences to understand what defines an acceptable level of uncertainty, financial risk, and environmental 540 risk. This study raises the question as to whether for subsurface activities where there could be a 541 substantive impact on communities or the environment, is there a need for regulators to demand 542 minimum data standards as part of the planning process? There is more than ever a social dimension to 543 subsurface uncertainty. Never has the spotlight been so focused on the ability of geoscientists to predict 544 the subsurface. 545 546

Acknowledgments 547
The authors are grateful to the UK Onshore Geophysics Library for access to information relating 548 onshore oil and gas exploration wells, and to access to the 3D seismic reflection data analysed as part 549 of this study. Some of the analysis shown is based upon data provided by the British Geological Survey 550 © UKRI. All rights reserved. Maps throughout this study were created using ArcGIS software by Esri.    Table 3. Data coverage for 10 largest areas of annual domestic heat demand above 10 000 MWh/km2. 644 Both extent of 2D seismic reflection data and number of deep wells are quantified within these areas.