Assessment of climate change impacts on buildings, structures and infrastructure in the Russian regions on permafrost

The infrastructure inventory for selected regions was created using latest available official statistics from the Rosstat, which provides data for annual costs of residential real estate (thereafter buildings) and for the following five categories of capital fixed assets: (1) non-residential commercial and social structures (e.g. hospitals, schools, universities, airports, railway stations etc) (2) critical infrastructure (e.g. roads, railroads, pipelines, bridges), (3) heavy machinery and industrial equipment, (4) transportation machinery (e.g. vehicles, ships, trains), (5) intangible assets, such as intellectual property. The total original cost of each fixed asset category (e.g. structures and critical infrastructure) aggregated to the level of Russian administrative regions was used. Although almost half of Russian infrastructure has a residual value below 40%, the full cost of fixed assets, which does not account for the amortization and depreciation, is used in the official statistics since a substantial amount of infrastructure in use greatly exceeded its live expectancy and has a residual value of 0 (Porfiriev et al 2017). The spatial extent of some regions, such as Sakha—Yakutia Republic (which is comparable in size to the European Union), and the uneven distribution of regional development requires analysis at sub-regional scale. To alleviate this problem it was assumed that the spatial pattern of the fixed assets within a given region corresponds to that of population: higher and denser population indicates higher fixed assets. As a result, the cost of fixed assets in a given location (e.g. municipality) was estimated using the following equation:

$$\begin{eqnarray*}&&{{\rm{FA}}}_{{\rm{m}}}={{\rm{FA}}}_{{\rm{reg}}}* {{P}}_{{\rm{m}}}/{{P}}_{{\rm{reg}}},\end{eqnarray*}$$

where:

FA_m—cost of fixed assets in a municipality

FA_reg—cost of fixed assets in an administrative region;

P_m—population of the municipality;

P_reg—population of the administrative region.

Following the methodology of Porfiriev et al (2017), only stationary categories of fixed assets, such as structures and critical infrastructure (thereafter infrastructure) were considered for the analysis. It was assumed that other assets could be moved from the areas negatively impacted by permafrost degradation. The Rosstat data on costs of residential real estate was supplemented with the residential buildings inventory from the Russian State Fund for Assistance in the Reform of Housing and Communal Services (ReformaGKH 2018). This source provides data on major characteristics of the residential buildings, such as postal address, year of construction, number of apartments. In order to integrate the residential real estate into the geodatabase, postal addresses were converted to geographic coordinates within ArcGIS 10.5.

Since the exact extent of permafrost under infrastructure is unknown, it was implied that in areas of significant development, permafrost continuity corresponds to the lowest bounds of the permafrost continuity zones. In other words, we assumed that 90% of infrastructure is underlain by permafrost in continuous, 50% in discontinuous, and 10% in sporadic permafrost zones. For island permafrost, it was assumed that permafrost was avoided during construction. As a result, the total regional capital cost of fixed assets directly affected by permafrost was estimated using the following equation:

$$\begin{eqnarray*}&&{\rm{TC}}=0.9{C}+0.5{D}+0.1{S},\end{eqnarray*}$$

where

TC is the total cost of infrastructure affected by permafrost in the region,

C is the total cost of infrastructure on continuous permafrost,

D is the total cost of infrastructure on discontinuous permafrost,

S is the total cost of infrastructure on sporadic permafrost.

The outputs from six GCM models (CanESM2, CSIRO-Mk3-6-0, GFDL-CM3, HadGEM2-ES, IPSL-CM5A-LR, and NorESM1-M), which participated in the 5th Climate Model Intercomparison Project (CMIP 5), were used to assess climatic changes projected for the mid-21st century. These models were selected based on their ability to accurately reproduce observed surface air temperature trends for the Russian Arctic regions (Anisimov et al 2013, AMAP 2017). CMIP5 GCM output is available at 1 × 1° lat/long for the Northern Eurasia. Daily means of surface air temperature and precipitation, produced by six GCM models were averaged to compile the decadal climatology of daily temperature and precipitations for the present (2006–2015), and mid-21st century (2050–2059) periods. The Representative Concentration Pathway (RCP) 8.5 scenario was used for mid-21st century projections as it represents the most likely near-term scenario and provides the worst-case outcome for risk assessment and gives the upper limit of potential costs.

An equilibrium model of permafrost-climate interactions was used to estimate permafrost temperature and active layer thickness (ALT) within the study area under present and mid-21st century climatic conditions. The model is based on the Kudryavtsev solution of nonlinear heat transfer equations with phase changes through snow, vegetation, organic, and mineral substrate with variable thermal properties (Shiklomanov and Nelson 1999, Sazonova and Romanovsky 2003). Daily means of air temperature and precipitation were used as climatic forcing. Snow depth was estimated from precipitation using average snow density of 300 kg m⁻³. The model was extensively validated by empirical observations (Shiklomanov and Nelson 1999, Streletskiy et al 2012c) and was used in climate change applications in Russia (Anisimov and Reneva 2006, Shiklomanov and Streletskiy 2013, Streletskiy 2015b, Shiklomanov et al 2017b).

The differences in permafrost properties between two decades were calculated and scaled to 0.25° grid using ordinary kriging. The average, minimum, maximum and standard deviation of permafrost estimates produced with the climate outputs from six individual GCMs were calculated for each grid node to evaluate central tendency and variability due to differences in climatic forcing.

Within the framework of this study, we have considered two major risks to buildings and infrastructure associated with permafrost degradation under projected climatic conditions: (1) ground subsidence and (2) a decrease in the ability of permafrost to carry structural load (or bearing capacity). Ground subsidence is associated with the melting of spatially-heterogeneous ground ice, accompanied by the consolidation of sediments under progressive thickening of the active layer. This process can be a major hazard for critical infrastructure (e.g. roads, railroads) and, as a result, can negatively impact the connectivity and accessibility of northern communities by land. The bearing capacity of foundations on permafrost is dependent on permafrost characteristics. Permafrost warming, accompanied by increase in the ALT, can reduce the ability of foundations to support buildings and structures, leading to deformations and ultimately structural failure.

Spatial estimates of the potential thaw subsidence across Russian permafrost regions were produced using an approach developed by (Nelson et al 2001). The method utilizes ground ice content information from the IPA Permafrost map and relative climate-induced changes in the ALT, as estimated by the permafrost model forced by GCM-produced decadal climates. The following formula was used to estimate ground subsidence.

$$\begin{eqnarray*}&&{S}={\rm{d}}{Z}\times {I},\end{eqnarray*}$$

where

S—is thaw subsidence (m),

dZ the difference in active-layer thickness between contemporary and mid-21st century decadal periods (m)

I—ground ice content (%),

Bearing capacity was estimated using a geotechnical model developed by (Streletskiy et al 2012b). The model estimates the maximum structural load which can be carried by a standard foundation pile inserted to a given reference depth into the permafrost. To apply this model for spatial assessments of potential changes in stability of buildings and structures on permafrost, we assumed that these buildings and structures were erected on piles with ventilated crawl spaces. Such 'passive principle of permafrost construction' is aimed at maintaining the frozen state of the ground during construction and the lifespan of buildings (Shur and Goering 2009). It is the most commonly used approach to construction in Russian permafrost regions. The ground thermal regime is assumed to be unaffected by thermal input from the structure and no engineering solutions (except for the ventilated crawl space) were assumed to have been implemented to control ground thermal regime over the lifespan of the structures. No snow accumulation or vegetation growth was assumed possible under the buildings (Khrustalev et al 2011). The model input consists of spatially and temporally variable permafrost conditions (temperature and active-layer thickness) and standard pile characteristics. While foundation can vary between structures, the use of standard pile characteristics is suitable for assessing differences in geotechnical properties of permafrost between different territories and over various time intervals (Grebenets et al 2012). The relative changes in bearing capacity due to climate-induced changes in permafrost properties between two decadal climatic periods were used to assess the potential risks to infrastructure, regardless of structure-specific engineering designs. This method was previously applied for evaluation of ongoing (e.g. Streletskiy et al 2012a, Streletskiy and Shiklomanov 2016) and future (e.g. Shiklomanov et al 2017b, Streletskiy 2015b) changes in permafrost bearing capacity for several large cities on permafrost in Russia.

Spatial estimates of potential ground subsidence and changes in permafrost bearing capacity, attributable to changes in climatic conditions between the present and mid-21st century, were used to delineate areas with climate-induced risks to buildings and infrastructure.

To estimate the economic impacts of projected permafrost changes on human infrastructure we have applied the 'stressor-response approach' (Melvin et al 2017), which considers environmental stressors (e.g. ground subsidence, decrease in bearing capacity) and infrastructure replacement costs. The asset preservations strategy, which assumes maintaining the present quantity of all engineered structures to the mid-21 century, was used for analysis. Within the framework of the stressor-response model, the following stressor thresholds triggered the response: (1) for infrastructure: the ground subsidence exceeding 0.10 m; (2) for residential buildings and non-residential structures: ≥50% decrease in foundation bearing capacity. It was assumed that crossing these thresholds caused substantial damage to assets, requiring their replacement. The regional costs, associated with climate-induced permafrost hazards, were estimated as a ratio of cost of asset affected by environmental stressors to the total regional costs of the asset. The 2016 Gross Regional Products data were used to estimate the replacement costs of assets under the risk of being damaged due to permafrost changes relative to regional budgets. The costs were converted from Russian Rubles to USD using the reference Purchase Power Parity conversion of 24RUB/USD, as suggested by the Organization of Economic Cooperation and Development (OECD 2018). The integrative spatial analysis was performed using ArcGIS 10.5 and spatial database containing all available socio-economic (e.g. population, infrastructure, cost of assets) and environmental (climate, permafrost) data.

The average increase in the MAAT between decades of 2006–2015 and 2050–59 projected by ensemble of six CMIP 5 models under the RCP 8.5 scenario for Russian permafrost regions is 3.8 °C (figure 2(a)). The highest and lowest temperature increases are projected by the GFDL and the CSIRO models respectively. The spatial pattern of projected MAAT increase is consistent for six models: regions bordering the Arctic Ocean (e.g. NAO, YNAO, and CAO) have generally higher projected temperature increases. Below average increases are found in the southern permafrost regions (e.g. Komi Republic, KMAO, Kamchatka Krai, Magadan Oblast). The large North–South gradient in MAAT increases are evident within the large administrative regions such as Krasnoyarsk Krai and Sakha Republic: the MAAT in the northern parts of these regions are projected to increase by more than 5 °C while the southern parts by less than 3 °C (table 1). The models-produced precipitation changes between decades of 2050–2059 and 2006–2015 are less consistent (figure 2(b)). Similarly to temperature, GFDL projected the highest precipitation increases and CSIRO the lowest. However, the range of projected precipitation changes is much larger than MAAT, especially in coastal areas. Generally, regions influenced by Atlantic and Pacific maritime air masses are projected to have higher precipitation increases comparing to more continental areas. Komi Republic and CAO showed the highest precipitation increase of >85 mm, while central parts of West Siberian and Sakha Republic was characterized by <60 mm increases (table 1). The six-model ensemble average precipitation increase for the Russian permafrost regions is 70 mm.

Zoom In Zoom Out Reset image size

The projected air temperature and precipitation (snow) increases, will have significant effect on permafrost temperature and the thickness of the active layer (table 1). Permafrost changes are most pronounced in the continuous permafrost zone of northern regions, especially in NAO and CAO. The magnitude and spatial pattern of projected permafrost changes corresponds well to observed permafrost trends in Russia where permafrost temperature has increased by up to 2 °C over 30 year period, with some sites as much as 1 °C over the last decade (Romanovsky et al 2010, Drozdov et al 2015). ALT was estimated to increase by more than 0.5 m with northern parts of NAO, YNAO and CAO experiencing more than 0.7 m increase, while regions located on discontinuous permafrost have generally lower increases of up to 0.4 m. These trends are in line with historical trends characteristic of the Russian permafrost regions, where sites generally showed 0.1–0.2 m increases per decade (Streletskiy et al 2015b).

The projected increases in permafrost temperature and the thickness of the active layer will greatly affect engineering properties of permafrost causing decrease in its ability to support structures. Our estimates using mean of six-GCM ensemble suggest that the several Russian permafrost regions will experience decrease in bearing capacity of more than 50% by the decade of 2050–2059 (figure 3(a)). The pattern of bearing capacity change is somewhat different from projected changes in permafrost temperature. For example, the higher increases of permafrost temperature are expected in the most northern regions (e.g. Chukotka AO), while the highest decrease in permafrost bearing capacity is projected for the southern, most populated and developed regions of NAO, YNAO, KMAO and Komi Republic. This stems from nonlinear dependence of permafrost bearing capacity with temperature. Bearing capacity of cold permafrost is less sensitive to changes in ground temperature relative to warmer permafrost. This makes southern permafrost regions more vulnerable to loss of bearing capacity per degree of ground temperature change. The areas of substantial oil and gas development and cities like Salekhard, Nadym, Novyy-Urengoy, Vorkuta, and Norilsk and many municipalities located in the northern part of Krasnoyarsk Krai and Sakha Republic are expected to experience substantial decreases in bearing capacity.

Zoom In Zoom Out Reset image size

The assessment of projected active-layer increases in regions characterized by ice-rich permafrost was used as an indicator of potential ground subsidence. Ice-rich permafrost regions of Sakha Republic and YNAO are projected to have the highest ground subsidence. This is likely to decrease stability of critical infrastructure and negatively impact accessibility of the economic centers (table 1). For example, roads and railroads in the vicinity of Yakutsk (the major urban center in Republic of Sakha (Yakutia)) are expected to be negatively affected by ground subsidence despite relatively small changes in bearing capacity projected for the city buildings. In YNAO, the projected increase in ground subsidence is likely to impact linear structures associated with oil and gas production (e.g. pipelines) which can destabilize the economic activities in the regional centers like Nadym and Novyy-Urengoy (figure 3(b)).

According to our estimates, the total population of the Russian permafrost regions was 5.4 mil, or about 4% of the total Russian population in 2016. Almost five million people resided in the nine regions selected for the analysis in this study. Four regions: Komi Republic, YNAO, KMAO, and Sakha have a combined population of almost 4 million (table 2). In YNAO, KMAO and Magadan Oblast, more than 80% of population was urban, while CAO and Sakha had a slightly lower urbanization of 70%. Several large cities located in the study area: Magadan, Labitnangy, Nadym, Norilsk, Novyy Urengoy, Salekhard, Vorkuta, and Yakutsk had combined population of almost 0.9 million, which is comparable to the total population of the North American Arctic (Alaska and Northern Provinces of Canada).

The total value of fixed assets in nine administrative regions on permafrost was 1.29 trillion USD, or about 17% of Russia's total. Almost 79% of fixed assets in permafrost-affected regions were in non-residential buildings and critical infrastructure categories worth of 140.9 bln USD and 884.5 bln USD, respectively. Residential real estate was valued at 279.2 bln USD. The largest contribution to overall cost of assets was provided by the West Siberian regions of YNAO and KMAO. The values of fixed assets in CAO and Magadan Oblast were the lowest among nine regions. Details on distribution of fixed asset costs among regions are shown in table 2.

The amount and cost of infrastructure affected by permafrost varies among nine study regions (table 2). Although the Komi Republic is mainly located outside the permafrost region, the highly-developed coal industry center of Vorkuta is built on permafrost. As a result, as much as 8% of unmovable fixed assets in Komi are affected by permafrost. On the other hand, due-to the large expanse of permafrost in the neighboring Nenets Autonomous Okrug (NAO), approximately 75% of its infrastructure is erected on permafrost while the total value of fixed assets in NAO is four times smaller than in Komi Republic. Permafrost regions of YNAO, located in West Siberia, contain 17.23 bln USD worth of structures and 137.55 bln USD worth of critical infrastructure, which corresponds to 44% and 66% of the total cost of these types of assets on permafrost in Russia. The high cost of infrastructure on permafrost in YNAO is attributable to intensive oil and gas development. The infrastructure and population are concentrated in several areas of continuous (Yamalskiy and Tazovskiy) and discontinuous (Salekhard, Labytnangy, Novyy Urengoy, Nadym) permafrost zones of YNAO. Further south, in the Khanty-Mansi Autonomous Okrug (KMAO), substantial population resides within the sporadic permafrost zone (Kogalym, Beloyarskiy, Raduga). Although oil- and gas-related infrastructure transverse permafrost regions of KMAO, the value of permafrost-affected assets in KMAO is less than 2% of the total regional cost of fixed assets. In Central Siberia, more than a half of Krasnoyarsk Krai is occupied by permafrost. However, the permafrost regions of Krasnoyarsk Krai are sparsely populated except the Norilsk—Dudinka urban and industrial region, which contains almost 9% of the regional assets and located within continuous permafrost zone. To the east, the Republic of Sakha has more than 85% of assets located in permafrost areas, particularly in the cities of Yakutsk, Neryungri and Mirnyy. In the Far East of Russia the majority of CAO assets are in the continuous permafrost zone around the city of Anadyr. Although majority of Magadan Oblast is underlined by permafrost, the infrastructure is concentrated in vicinity of Magadan in a discontinuous permafrost zone. Overall, about 56% of assets are located in permafrost regions of Magadan Oblast. Although the value of assets in Kamchatka Krai exceeding that of CAO and Magadan Oblast combined, only less than 4% of these assets are affected by permafrost.

The total cost of fixed assets affected by permafrost in the nine administrative regions in 2016 was 248.6 bln USD or 7.5% of Russian GDP for that year. The total value of residential real estate on permafrost was estimated to be 52.6 bln. A highest total costs of real estate on permafrost are estimated for Sakha (25.9 bln), YNAO (9.1) and Krasnoyarsk Krai (7.9 bln). However, the relative proportion of buildings on permafrost is highest for CAO (92%), followed by Sakha (85%), and NAO (75%).

The cost of projected climate–induced permafrost changes is based on average values of the six climate models and provides the best point-estimate of buildings and infrastructure affected by loss of bearing capacity and ground subsidence. The agreement between the climate input and associated changes in permafrost estimates vary in-between and within the regions adding to various degree of uncertainty of estimated costs of climate impacts (table 1, figures 2 and 3). For example, Komi Republic is characterized by close agreement between climate models, while CAO has substantial variability in both temperature and precipitation estimates resulting in various degree of uncertainty of permafrost estimates. In order to provide standardized measure of variability about the mean that can be used across the study regions, the deviation of 5% and 1 cm around the mean were used for the bearing capacity and ground subsidence assessments, respectively. According to average estimates, the total cost of infrastructure impacts associated with the climate-induced permafrost changes in Russia would reach 105.07 bln USD by the mid-21st century. Among the nine Russian administrative regions with permafrost the highest costs are expected in YNAO and Sakha with 52.33 bln USD and 21.26 bln USD, respectively (table 3). Komi Republic, NAO, and Krasnoyarsk Krai would incur 8.5 bln to 10 bln USD in additional costs while projected costs for KMAO (1.46 bln USD), CAO (1.90 bln USD), Magadan Oblast (1.0 bln USD) and Kamchatka Krai (0.1 bln USD) will be relatively low. However even relatively small changes in permafrost bearing capacity and ground subsidence may substantially increase these estimates, particularly in the regions east of Krasnoyarsk Krai (table 3).

The incurred costs, attributable to specific permafrost-related environmental stressor (e.g. ground subsidence and loss of bearing capacity), vary between the regions. In the Komi Republic and NAO about third of costs is associated with damages to residential buildings and non-residential structures due to the loss of permafrost bearing capacity. In YNAO and KMAO, majority of expected costs (about 73% and 96%) will result from the deformation of infrastructure due to ground subsidence. In Sakha, the ground subsidence and the loss of bearing capacity will have almost equal economic impact: 49% and 51% respectively. In Krasnoyarsk Krai, 60% of the projected cost is associated with potential damages to residential buildings due to the high susceptibility to permafrost-related infrastructure damage in major urban centers of Norilsk and Dudinka. This is the only region where relative impacts of permafrost degradation on residential housing is higher than on non-residential structures and infrastructure.

To consider the ability of regions to address potential climate impacts, it is critical to evaluate the additional costs incurred due to changing climate in relation to overall economic prosperity of regions. A region with high population, diverse and developed infrastructure, and stable tax base is likely to cope with negative economic impacts of climate change better than a less-developed region. Although the regional economic conditions can fluctuate, the present-day GRP can be used as a proxy for the potential prosperity. To assess the economic resilience of nine Russian permafrost regions to infrastructure damages associated with climate-induced permafrost damage, the estimates of the total regional costs of all potentially-damaged infrastructure were spread equally over the 45 years and related to 2016 GRP of each region. Although such approach does not account for potential transient increase in infrastructure damage and/or potential changes in GRP, it provides a crude measure of the economic ability of each region to mitigate impacts of climate change.

Based on our economic resilience estimates, the financial burden associated with the mitigation of negative impacts related to permafrost degradation is the highest for the NAO, YNAO, and CAO (4%–5% of annual GRP). Although NAO and YNAO are relatively prosperous regions, the amount of critical infrastructure expected to be affected by the ground subsidence in these regions is significant, comparative to their respective GRPs. In CAO, the projected amount of infrastructure damages is much lower. However, this less economically-developed region has a much smaller GRP resulting in high relative costs and low economic resilience. The relative cost of mitigating negative permafrost impacts are also high in Sakha and Komi Republic where additional 3.7% and 2.2% of annual GRP respectively will be required to maintain the existing level of infrastructure. Magadn Oblast and Krasnoyarsk Krai are expected to spend under 1% of their GRPs on mitigating impacts of permafrost degradation. However, for Krasnoyarsk Krai it accounts to a significant additional cost of approximately 10 bln USD. Krasnoyarsk Krai can be considered as a special case due to its enormous size, large GRP, and diversity of environmental conditions. Only norther portion of the region has permafrost, and the permafrost infrastructure is concentrated predominantly in the Norisk-Dudinka urban industrial complex resulting in high economic resilience of the overall region to projected climate-induced permafrost changes. Although KMAO and Kamchatka Krai are much smaller than the Krasnoyarsk Krai, majority of their infrastructure is also located predominantly in non-permafrost areas resulting in smallest relative economic cost of mitigating projected damage to permafrost infrastructure of just under 0.1% of their respective GRPs.

The goal of this research was to evaluate the role of climatically driven changes of permafrost on structures and infrastructure and associated costs at regional scales regardless of the variability in specific construction designs and building standards that were discussed in previous works (Shur and Goering 2009, Streletskiy et al 2012b). The role of non-climatic factors, both human (building standards, adequate maintenance, snow redistribution, waterlogging) and natural (changes in soil moisture and ice content, and frost heave, changes in geochemistry and salinity depressing freezing point of water) should further be considered for more granular analysis. Construction methods such as implementation of longer piles, use of thermosiphons, and proper planning and land use (such as regular snow removal, storm water management) may have significantly offset the negative changes associated with climate warming and decrease the projected costs. However, the improper operation of the structures, snow accumulation, waterlogging and absence of storm water management may accelerate the permafrost degradation and further deteriorate geotechnical environment, therefore increasing the cost of permafrost change.