D422Lot1.SMHI.5.1.1B: Detailed workflows of each case ... · from 1951 to 2012. Reduced winter ice...

C3S_422_Lot1_SMHI – D5.1.1B |

Copernicus Climate Change Service

D422Lot1.SMHI.5.1.1B:

Detailed workflows of each case-study

on how to use the CDS for CII

production and climate adaptation

Full Technical Report:

Road and River Transport in Central

Yakutia, East Siberia, Russia

L. Lebedeva1, D. Gustafsson2, O. Makarieva1,3, N. Nesterova3

1Melnikov Permafrost Institute, 2Swedish Meteorological and Hydrological

Institute, 3St.Petersburg State University

REF.: C3S_422_Lot1_SMHI D5.1.1B

C3S_422_Lot1_SMHI – D5.1.1B |


Summary

Mean annual air temperature in Yakutsk has increased from -10.4 °C to -8.7 °C from 1951 to 2012. Reduced winter ice road access and land road stability have

numerous negative consequences for communities and the industry sector in Eastern Siberia. Data from the Copernicus Climate Change Service can be used as a basis for producing climate impact indicators (CIIs) for historical and future

periods up to 2100 on local scale in combination with impact models and locally observed data. Such CIIs are essential for development of a long-term strategy

for adaptation to the climate change in Yakutia. Principal stakeholders on federal, regional and local scale admit urgent need for such strategy in climate-

dependent transportation sector. CIIs were produced for the ice roads and ferry at the main local transport hub – Lena River at Yakutsk and seven other official river ice road crossings along the three federal roads. Road stability was

evaluated for the federal roads in the most densely populated central part of Yakutia in relation to different permafrost landscape settings. Produced CIIs on

thawing depths for 21st century show contrasting response of different permafrost landscapes to expected climate change. Substantial simulated changes of thawing depth by the end of 21st century in some environmental

settings could lead to critical modification of road construction and exploitation techniques. In other more resilient landscapes even small deepening of thawing

layer could lead to high risk of road damage and collapse due to hazardous cryogenic processes - surface subsidence and thermokarst intensification. River ice thickness have been reduced due to the higher winter temperatures during

the second half of 20th century and beginning of 21st century. For future periods, the model simulations show that maximum annual ice thickness is

reduced by 36-50 cm a the Tabaga station, and ice cover period may be shortened by 20 or even up to 50 days. Produced CIIs would contribute to development of both short-term adaptation measures for safe construction and

exploitation of transportation routes in Eastern Siberia and long-term strategy of adaptation and mitigation to climate change on local and regional scale.

C3S_422_Lot1_SMHI – D5.1.1B |


Contents 1- Case study description 4

1.1 Issue to be addressed 4

1.2 Decision support to client 4

1.3 Temporal and spatial scale 4

1.4 Knowledge brokering 5

2- Potential adaptation measures 5

2.1 Lessons learnt 5

2.2 Importance and relevance of adaptation 5

2.3 Pros and cons or cost-benefit analysis of climate adaptation 6

2.4 Policy aspects 6

3- Contact 6

3.1 Purveyors 6

3.2 Clients/users 6

4- Data production and results 7

4.1 Step 1: Local data collection 7

4.2 Step 2: Global data collection and adaption 9

4.3 Step 3: Data analysis 10

4.4 Step 4: Local impact model simulations and CII production 12

4.5 Step 5: Communication and Dissemination 29

5- Conclusion of full technical report 30

References 31

C3S_422_Lot1_SMHI – D5.1.1B | 4


1- Case study description

1.1 Issue to be addressed

Rivers are the most important transportation routes in many parts of Siberia due

to the absence of railway, bridges and the limited number of roads on land. Many remote settlements and industrial facilities critically depend on delivery by winter

and ice roads, that exist only for a limited period from December-January to April. Up to 90% of goods are brought in by trucks during that short winter time. Due to global warming, mean annual air temperature in Yakutsk has increased

from -10.4 °C (1951-1978) to -8.7 °C (1979-2012). As a consequence, roads on frozen ground are prone to risks of subsidence and structural weakening as

permafrost thaws. Reduced winter ice road access and land road stability have numerous negative consequences for communities and the industry sector. As often winter roads are the only land links around and between remote

settlements, a shortened winter road season implies higher costs of goods in these impacted areas. When winter roads fail or close, overland travel may

become dangerous or impossible and unless located near a navigable waterway, communities face steep price increases as supplies must be delivered by air. Mining, energy and timber interests face shorter time windows to transport

necessary equipments and products; and the coordination of the supply chain is getting complex due to the interannual variability in the winter road season

length.

1.2 Decision support to client

The Lena River Basin Water Management Administration operates ferry routes and ice crossing at the main regional transportation hub in Yakutsk. This administration plans the construction of seasonal roads and set up strategies for

their exploitation. Information on observed and projected durations of navigation season, operation of ice roads and assessment of the stability of roads

constructed on permafrost this case study provided, will support the planning of the transportation schemes, including investment into roads maintenance and influence policies.

In addition, the department of permafrost engineering at the Melnikov Permafrost Institute (MPI), that develops new technologies for safe construction

and exploitation of roads on land, will use the projections of roads stability under climate change in near future to improve their concepts of road development and maintenance.

1.3 Temporal and spatial scale

The local impact model simulations and CII production covered a historical

reference period (1971-2001) and three future 30-year periods (2011-2040, 2041-2070, and 2071-2100) representing the emission scenarios RCP4.5 and

RCP8.5. CIIs were produced for the ice roads and ferry at the main local transport hub – Lena River at Yakutsk. Road stability was evaluated for three main roads in the

most densely populated central part of Yakutia.

C3S_422_Lot1_SMHI – D5.1.1B | 5


1.4 Knowledge brokering

Interaction with the clients in our attempts to fill the gap between observational data, science and the client needs are organized through official correspondence

and personal communication every 1-2 months. Telephone and email communication are used to share information about on-going work, make clarifications and decide details for coming steps. The project outputs were

delivered to interested stakeholders through face-to-face meetings and short final report. Client feedback was asked for and considered for quality control and

assurance that we provide the decision support that the client actually needs.

2- Potential adaptation measures

2.1 Lessons learnt

We evaluated the value of the globally available climate information from C3S_422_Lot1_SMHI for producing local CIIs for transportation in Eastern Siberia for present and future. Mean annual air temperature in Yakutsk has

increased from -10.4 °C (1951-1978) to -8.7 °C (1979-2012). River ice thickness at the Lena River near Yakutsk has decreased in March and April for the period

1955-2015 by 27-49 cm, which implies a reduction of safe ice crossing operation season. Although both ground temperature and thawing depth are stable for the period 1982-2012, the projected future climate warming could lead to deepening

of thawing depth and cause ground subsidence, road damage and collapse in some environmental settings. Two impact hydrological models Hydrograph and

HYPE showed applicability to assess CIIs based on global and local data. Combination of local data and verified impact models contributes to successful usage of the climate service on local scale. CIIs will contribute to decision-

making process regarding transportation policy and investment into roads maintenance. Absence of existing long-term strategy of adaptation to climate

change on local and regional scale in Yakutia limits, to some extent, the quick practical applicability of produced CIIs but makes them useful for development of such strategy.

2.2 Importance and relevance of adaptation

Traditional methods of ground heat stabilization (e.g. thermosyphon) are current

adaptation measures for road stability in permafrost regions. They are typically installed without consideration of possible climate-induced changes in the future.

Conventional techniques of hydrological forecast are used for assessment of expected navigation period length and river ice thickness. Such methods and techniques were developed solely on historical data although local communities

notice climate change influence on their life. In winter 2016-2017 winter roads in North-Eastern Yakutia were opened in March only that didn’t allow accomplishing

of Northern Supply plan in full. Long-term adaptation strategies could be developed on the basis of CIIs produced using global climate data from the climate service, impact models and local up-to-date data.

C3S_422_Lot1_SMHI – D5.1.1B | 6


2.3 Pros and cons or cost-benefit analysis of climate adaptation

The probable adaptation measure to shortened ice roads seasons and thinner ice

conditions is an increased investment into artificial ice freezing at the ice crossings. The most popular and well-known method is ice-building by watering

with the Grad (“hail”) sprinkling machine. Another possible measure is to prolong the ferry navigation season to October that is subject to high year-to-year variation and danger of ice formation on the ferry when operating at low

temperatures. Longer periods between the end of ferry use due to ice freeze up and formation of ice cover thick enough for driving, imply substantial higher

costs for operating hovercrafts for people transportation and delivery of goods by air. Ground subsidence, road damage and collapse due to deeper ground thawing

lead to temporal road closure, higher investments to road repair and danger of accidents. Potential adaptation measures include implementation of special

techniques for ground thermal stability such as thermosiphon cooling and winter snow removal. Additionally, road constructions in particular environmental settings such as areas where risk of ground subsidence due to thawing is high,

should be reconsidered.

2.4 Policy aspects

Government of Sakha (Yakutia) republic enacted the bill about protection and preservation of permafrost on 22nd of May 2018 (source: Decree of head of

Sakha (Yakutia) Republic). The document discusses possible risks of permafrost thaw and related damages of roads and other infrastructure in Yakutia. Scientists from the Melnikov Permafrost Institute (MPI) consulted policy-makers about

tendencies, present and future states of frozen ground in different natural settings. The CIIs and conclusions about road stability in different landscapes of

Central Yakutia is a useful input for practical implementation of the bill. Preparation of such document on the governmental level highlights the importance of possible climate change influence for local policy-makers and

stakeholders. Some federal states of Russia (St. Petersburg, Murmansk oblast) has started preparations of climate adaptation strategies. CIIs could be useful

input for development of such strategy for Sakha (Yakutia) republic.

3- Contact

3.1 Purveyors

Lyudmila Lebedeva, Melnikov Permafrost Institute (MPI), Yakutsk, Russia

3.2 Clients/users

Lena River Basin Water Management Administration (Federal Water Resources Agency), Russia; Laboratory of Permafrost Engineering, Melnikov Permafrost Institute (MPI),

Yakutsk, Russia; Emergency Agency of Sakha (Yakutia) Republic, Yakutsk, Russia

about:blank

http://2605.pdf/

http://2605.pdf/

C3S_422_Lot1_SMHI – D5.1.1B | 7


4- Data production and results User needs and relevant global and local datasets were collected in the beginning

of the project. Local stakeholders do not have any long-term strategy of adaptation to climate change on local and regional scale in Yakutia and they

admit importance of development of such a strategy. For this purpose, they are interested to have scientific data describing observed climate change impacts in

the historical and future situation for the climate dependent transportation sector. The aim of the workflow is thus to provide CIIs relevant to roads on land and rivers in Central Yakutia.

Collected local river ice and ground thawing data (observations) were analysed

for trends to quantify the ongoing and historical changes, and were further used for verification of impact models for historical time period before application of these models for future periods.

Global data sets on essential climate variables (ECVs, in this case air

temperature and precipitation) for historical and future periods available in the C3S Climate Data Store (CDS) was collected for the Yakutian region, and used as input to the impact models used for simulating road and river ice conditions.

Initial quality controls comparing the global data sets to local observations for historical periods were used to select ECV data for the CII production for future

periods. Local applications of the impact models were setup for simulation of ground

thawing conditions in typical landscapes, and of river ice conditions at river ice road crossings in Central Yakutia. CII production by impact models covered a

historical reference period and three future 30-year periods up to 2100 representing the emission scenarios RCP4.5 and RCP8.5. CIIs for future and historical periods based on both observations and impact model simulations are

disseminated to the public through the showcase interactive climate atlas. A more in-depth analysis of the results is presented to the local stakeholders

through face-to-face meetings and written communication.

4.1 Step 1: Local data collection

Description:

C3S_422_Lot1_SMHI – D5.1.1B | 8


Long-term historical time-series starting in the 1950s’ of meteorological

measurements (air temperature and precipitation on daily time scale), river ice observations (thickness, break-up and freeze-up dates), and measurements of ground temperature and thawing and freezing depths for typical permafrost

landscapes were collected from hydro-meteorological and permafrost monitoring stations in Central Yakutia.

Landscape, land use, soil and vegetation cover maps, and digital elevation models were compiled from national and international databases. Stakeholder needs were collected from regional road and river transport

authorities through written communication and face-to-face meetings.

Results: There are three federal roads “Lena”, “Kolyma” and “Viluy” in Yakutia, including

several river crossings dependent on ferries in summer and ice roads in winter. The Lena river crossing in Yakutsk is the main transportation hub in the region. On land these roads cross different landscapes which are contrasting in

permafrost properties and are expected to react differently to climate change. Local meteorological observations namely daily air temperature, humidity and

precipitation were collected for Yakutsk, Churapcha, Berdigestyakh and Pokrovsk meteorological stations (Fig. 1) for 1966-2012. The stations are representative for Central Yakutia region. The Lena River ice break-up and freeze-up dates and

river ice thickness were collected for Tabaga, Yakutsk and Khangalassy hydrological gauges for 1950-2012. Ground temperature and ground thawing

depth were collected at the MPI research sites “Chabyda” (Varlamov et al., 2012), Spasskaya Pad’ and Neleger (Fedorov et al., 2006) for different typical permafrost landscapes: pine forest with sandy soil, larch forest with sandy soil,

larch forest with loamy soil, alas depressions covered by grassland with loamy soil, interalas areas covered by larch and birch forests and marsh. Landscape

distribution for the three roads was analysed based on the Permafrost Landscape Map of Yakutia ASSR (Fedorov et al., 1991). The most frequent landscapes along the roads are larch and pine forests with sandy and loamy soils and grasslands at

the thermokarst depressions (alases). Larch forests with loamy soils surrounded by alases are characterized by presence of ground ice and therefore the most

vulnerable landscape under climate change conditions. Location of meteorological stations, federal roads and landscape distribution in Central Yakutia is shown at the Fig. 1. Ice and ferry river crossings along the federal roads in Yakutia is

presented at the Fig. 2.

C3S_422_Lot1_SMHI – D5.1.1B | 9


Fig. 1 Scheme of landscape distribution in Central Yakutia along three federal roads, modified after

Fedorov et al. (1991): 1 – larch forests with sandy soils, 2 – pine forests with sandy soils, 3 – marsh, 4 – interalas areas covered by larch and birch forests, 5 – alas depressions covered by grassland with loamy soil, 6 – larch forests with loamy soils, 7 – pine forests with loamy soils, 8 – mixed pine and larch forest, 9 – water.

Fig. 2 Ice and ferry river crossings along the federal roads in Yakutia

4.2 Step 2: Global data collection and adaption

Description:

Essential climate variable data (ECV, air temperature and precipitation on daily time scale) available in the C3S Climate Data Store (CDS,

https://climate.copernicus.eu/climate-data-store) was used as input to the local impact models for historical and future periods. The data used in the case study includes the HydroGFD reference data (1961-2015) and the ensembles of bias-

corrected CMIP5 climate model outputs (1951-2100). Quality control of global ECV data was performed to define data to be used for

future climate scenarios: the HydroGFD reference data and the bias-corrected

https://climate.copernicus.eu/climate-data-store

C3S_422_Lot1_SMHI – D5.1.1B | 10


CMIP5 ensembles were compared with the locally observed air temperature and

precipitation to assess adequateness of the global data for local scale. Results:

Fig.3 shows comparison of mean monthly temperature and mean monthly precipitation sum between observations at Yakutsk meteorological station,

HydroGFD data and 18 GCMs for reference period 1970-2012. The same analysis was conducted for other three meteorological stations in Central Yakutia. Simulated mean temperature shows high agreement with observations on

monthly scale. Modelled precipitation has higher discrepancies with observations. GCMs underestimate precipitation amount for the period from October to January

and overestimate it in June. Our target CIIs relate to ground thawing and river ice that depend on air temperature much more than on precipitation. Based on

the quality control of climate data it was decided that simulations of all GCMs would be used as input to impact models for CIIs production for future periods. Both RCP 4.5 and RCP 8.5 scenarios are used for CIIs production.

Fig.3 Comparison of mean monthly temperature, degree Celsius, (left) and precipitation, mm, (right) produced by bias-corrected towards HydroGFD global climate models with HydroGFD product and in-situ meteorological observations at the Yakutsk station for reference period 1970-

2012

4.3 Step 3: Data analysis

Description:

Trend analysis and quality control of local observational data: river ice conditions (break-up and freeze-up dates and ice thickness), ground temperature and

ground thawing and freezing depth in typical landscapes was performed to assess on-going changes and prepare data for impact model verification.

Results: The mean ice thickness of the Lena River at Yakutsk in March and April is 137-

143 cm (Table 1). Trend analysis showed that ice thickness has decreased in March and April during the period 1955-2015 (Fig.4). Depletion of river ice thickness intensifies from March to April and reaches the rates from 27 to 49 cm

(20-35%). The changes are statistically significant. Changes in other months are not statistically significant. Dates of river breakup are shifting. The first ice

movements occur earlier than before. Last day of complete Lena River freeze-up

C3S_422_Lot1_SMHI – D5.1.1B | 11


are shifting from second half to the first half of May (Fig.5). More trend analyses

on the observed river ice data is presented as part of Step 4 below. Collected

local data on ground temperature and thawing depth showed that they are relatively stable under the observed air temperature increase in Central Yakutia for the period

1982-2012 for the several landscapes (Fig.6). Long-term data are available only for

marsh, larch and pine forests.

Fig.4 Maximum annual Lena River ice depth at the Tabaga hydrological gauge and its trend estimate for historical time period

Fig.5 Last day of complete Lena River freeze-up at the Tabaga, Yakutsk and Khangalassy hydrological gauges for historical time period

C3S_422_Lot1_SMHI – D5.1.1B | 12


Fig.6 Mean annual ground temperature and maximum thawing depth at different landscapes in Central Yakutia for historical time period (Varlamov et al., 2012)

Table 1 Long-term mean values and changes of river ice depth at the Lena River at Tabaga in March and April, 1955-2015

March April

Date 10 20 31 10 20 30

Mean, cm 137 140 142 143 143 140

Changes, % -20 -21 -23 -29 -33 -35

Changes, cm -27 -29 -32 -42 -47 -49

4.4 Step 4: Local impact model simulations and CII production

Description: Local applications of the hydrological models Hydrograph and HYPE were setup

for simulation of ground temperature and river ice conditions in typical permafrost landscapes in Central Yakutia, and for 7 official river ice road crossings of the Lena, Aldan, and Vilyuy rivers, respectively.

Model input data (mainly from step 2) and parameters were adjusted using the local information and observational data obtained in step 1 and analyzed in step

3 to improve the ability of the models to represent the historical conditions. The local impact model simulations and CII production covered a historical reference period (1971-2001) and three future 30-year periods (2011-2040,

2041-2070, and 2071-2100) representing the emission scenarios RCP4.5 and RCP8.5. The uncertainty in the calculated CIIs were represented by the spread

generated from variation within the ensemble of bias-corrected CMIP5 climate model data used as input to the impact models.

Results - Ground thawing: Hydrograph model was applied in a lumped way to simulate ground thawing

depth, ground temperature and moisture in active layer in typical landscapes in

Central Yakutia, that are crossed by federal roads (Fig. 1): pine forest with sandy

soil, larch forest with sandy soil, larch forest with loamy soil, alas depressions

covered by grassland with loamy soil, inter-alas areas covered by larch and birch

C3S_422_Lot1_SMHI – D5.1.1B | 13


forests, and marsh. Model parameters were estimated based on field data at the

MPI research sites “Chabyda” (Varlamov et al., 2012), Spasskaya Pad’ and

Neleger (Fedorov et al., 2006) that are representative for above-mentioned

landscapes. Comparison of Hydrograph model simulation of ground temperature

in larch and pine forest at different depths with observations is shown at the

Fig.7. Larch and pine forests have contrasting thermal regime. There is thick

organic layer on the soil surface consisting from moss in larch forests that acts as

heat insulator and prevents soil from warming in summer. Maximum thawing

depth reaches only 1 m. On the contrary there is very thin organic layer in pine

forest. Soil thaws up to 3-4 m. Model satisfactorily represents ground

temperature in both larch and pine forests. The largest discrepancies with

observations up to 1.5°C are at 2-3 m depths in larch forest. Validated on the

historical observations Hydrograph model was used for future projection of

ground thawing and freezing in different landscapes.

Fig.7 Comparison of daily simulated and observed ground temperature in larch and pine forest at different depths

CIIs related to ground thawing namely annual maximum thawing depth, annual

maximum freezing depth, mean monthly thawing and freezing depths are

produced by the Hydrograph model forced by 18 GCMs for three future 30-year

periods (2011-2040, 2041-2070, and 2071-2100) representing the emission

scenarios RCP4.5 and RCP8.5. Fig.8 shows modelled annual maximum thawing

depth for six landscapes according to RCP4.5 for 2011-2100. Box-and-whisker

plot for every year shows spread between simulations forced by 18 GCMs.

Practically no changes of thawing depth are expected in alas and interalas areas.

Deeper thaw and its higher uncertainty is projected by the end of 21st century in

marshes, pine and larch forests with sandy soils. Deepening of mean monthly

thawing depth in marshes, pine and larch forests with sandy soils for period

2071-2100 would reach 66, 50 and 47 cm comparative to 2011-2040 period.

Another consequence of air temperature increase is shifting of month with

deepest thaw from October to November in marsh and pine forest with sandy

soils.

C3S_422_Lot1_SMHI – D5.1.1B | 14


Fig.8 Annual maximum thaw depth for six permafrost landscapes for 2011-2100, m, simulated by Hydrograph model forced by 18 GCMs for the RCP 4.5 emission scenario

Simulated annual maximum thawing depths according to RCP8.5 emission

scenario are presented at Fig.9. In contrast to modelling results by RCP4.5

significant increase of thawing depth is projected in all studied landscapes.

Uncertainty of model projections with RCP8.5 emission scenario to the end of

21st century is much higher comparative to RCP4.5. Deepening of mean monthly

thawing depth in alas, interalas and larch forest with loamy soil would reach 17,

15 and 27 cm for 2071-2100 period comparative to 2011-2040. Complete

permafrost thaw in upper 5 m is expected in other three landscapes - pine, larch

forests on sand and marsh during last third of 21st century according to RCP8.5.

Uncertainty of the projection in pine and larch forests is very high. Complete

upper permafrost thaw by the end of 21st century in marsh is simulated with 14

GCMs out of 18.

Simulation results suggest relatively high resilience of alas, interalas areas and

larch forests with loamy soils to future air temperature increase. It could be

explained by thick organic cover on soil surface and lower thermal properties of

C3S_422_Lot1_SMHI – D5.1.1B | 15


soil profile. Despite small projected changes in future high ground ice content in

soil profile in those landscapes leads to high risk of hazardous cryogenic

processes - surface subsidence and thermokarst intensification. They are

sporadically observed at the present time and could be caused even by

insignificant deepening of thawing layer. Substantial changes of thawing depth in

pine, larch forests on sand and marsh could lead to critical modification of road

construction and exploitation techniques. Significant difference in reaction to air

temperature increase between landscapes suggests high importance for

transport sector of taking into account landscape properties while road design,

construction and maintenance in permafrost areas.

Fig.9 Annual maximum thaw depth for six permafrost landscapes for 2011-2100, m, simulated by

Hydrograph model forced by 18 GCMs for the RCP 8.5 emission scenario

C3S_422_Lot1_SMHI – D5.1.1B | 16


Results - River ice:

A river ice impact model for calculating river ice CIIs for historical and future

periods were developed using the hydrological model HYPE. A preliminary model

setup was extracted for the Lena river basin from the pan-arctic Arctic-HYPE

model (v4.0.1, Andersson et al, 2015; Gelfan et al 2017; Macdonald et al, 2018;

http://hypeweb.smhi.se) to simulate river ice depths and freeze-up and break-up

dates for the federal river ice roads in central Yakutia. The main focus was on the

ice road crossings over the Lena river at the city of Yakutsk - the main transport

hub in Yakutia - where local river ice observations were also available. In

addition, model simulations were made for six other river ice crossings over the

Aldan and Vilyuy rivers (Fig. 2).

The river ice model was adapted to the local conditions as represented by the

local observations of river ice depth at the Tabaga station, and the dates of

freeze-up, ice formation, and break-up at Tabaga, Yakutsk and Khangalassy. It

should be noted that the ice observations as well as the simulation model

represent natural ice conditions where the ice road management practices are

not taken into account. The most important model improvements to the local

conditions were:

● introduction of internal melt of river ice as a function of solar radiation, to

improve river ice breakup simulation as a function of ice porosity

exceeding a critical stability threshold rather than ice thickness

approaching zero.

● introduction of a minimum heat flow from river water to the ice, to limit

the maximum ice growth for thick ice layers.

The original Arctic-HYPE river ice model was established using ice observations

from Swedish lakes and rivers using a zero heat flow from the water, and zero

ice thickness as spring break-up criteria. This was found to simulate the onset of

ice growth quite well for the Tabaga station. However, the annual maximum ice

depths were overestimated in general and the spring break-up date was not well

defined compared to the observations.

River ice CIIs were defined per ice season years as described in Tab. 2, based

on the variables included in the ice observations and variables used in ice road

operation practices. The ice season year is defined as the period from 1

September in the previous year until end of August in the current year.

http://hypeweb.smhi.se/

C3S_422_Lot1_SMHI – D5.1.1B | 17


Tab. 2 River ice CII definition by observations and calculation from model output.

CII Unit Definition by observations

Calculation from model simulations

Date of first ice (DFI)

day of year (1-366)

Date of first landfast ice

First date in the autumn with ice thickness larger than 0 in the autumn

Date of ice cover formation (DICF)

day of year (1-366)

Date of stable complete freezing for at least 20 days (freeze-up)

First date in the autumn with ice thickness larger than 0, ice cover fraction equal to 1, and air temperatures below -2.7°C in the previous 20 days.

Date of first ice flow (DIFL)

day of year (1-366)

Date of first ice movement, including slush flow on top of ice.

Date of ice break-up in spring, triggered by ice porosity larger than critical level set to 40% (in the model, ice thickness is reduced by surface and bottom melt driven by air temperature and heat flow

from the water, respectively, and porosity is increased by internal melt driven by solar radiation).

Ice period length (IP)

days Number of days between DIFL and DFI

same as for observations

Ice cover period (ICP)

days Number of days between DICF and DIFL

same as for observation

Ice-free period length (NIP)1

days Number of days between DIFL and

DFI for the next ice season year

same as for observations

Annual

maximum ice depth (IMAX)2

cm Maximum of

observed ice depth during the ice season year

Maximum simulated ice depth during the

ice season year

First day with more than 30 cm ice thickness (DF30)3

day of year First day with observed ice thickness larger equal to 30 cm

First day with simulated ice thickness larger equal to 30 cm

Mean monthly ice thickness, November -

April

cm Mean observed ice thickness in the month, excluding

missing values from mean value

calculation

Mean simulated ice thickness, including zero ice thickness values in the mean value calculations

1It should be noted that the day of first ice flow does not really represent ice-free conditions, so it

is not strictly correct to call the period between ice flow and first ice as ice free.

C3S_422_Lot1_SMHI – D5.1.1B | 18


2The ice season year is defined as the period from 1 September in the previous year until end of

August in the current year.

330 cm ice thickness is the recommended minimum ice thickness for starting ice road operations

and maintenance (ODM 218.4.030-2016 Guideline for assessment of load-bearing capacity of ice

road. 2017. 42 p). It also corresponds to safe allowable vehicle weights of 4-6 tons according to

the same report. For observations, DF30 is calculated by interpolating the made observations every

10th day, but only for years when the first ice observation is smaller than 30 cm.

River ice CII production and evaluation was made in two steps:

● In a first step, the river ice CIIs (Tab. 2) were calculated for a 54 year

time period (1962-2015) based on the available observations at Lena river at Tabaga, Yakutsk, and Khangalassy, and from model simulations forced

with the bias-corrected re-analysis dataset HydroGFD v2.0 provided by SMHI. The model based CIIs were evaluated versus the observation based with regard to mean value, correlation, root mean square error, and

trends analyses (Tabs. 3-4). Trend analyses was made using the Yue and Pilon (2002) method for prewhitened nonlinear trend analysis to take into

account lag-1 autocorrelation, as implemented in the R Package ZYP. ● In a second step, CIIs were calculated for the historical reference and

future 30-year periods as defined above, based on model simulations

forced by meteorological datasets derived from 18 CMIP5 GCM models and bias corrected towards the HydroGFD data. In this step, calculations were

made for all 7 river ice road crossings (Tab. 5)

In the following sections, the model evaluation, trend analyses, and future

projections are presented and discussed separately for a selection of the river ice CII.

Annual maximum ice thickness in the Lena river at Tabaga show a significant decreasing trend over the period 1962-2015, both in the observed data the

model simulations (Fig. 10). The trends in the observations was -4.5 cm/decade whereas the trend in the model was only -2.6 cm/decade. This underestimation

of the historical reduction in maximum ice thickness should be considered as additional uncertainty when interpreting the current results for future periods.

Fig. 10. Annual maximum ice thickness in Lena river at Tabaga station, from observations (black

filled dots) and model simulations (red open squares), slope of trend lines (cm/year).

C3S_422_Lot1_SMHI – D5.1.1B | 19


On the other hand, it can be noted the seasonal variability of river ice

thickness was quite well represented by the model simulations, with a low relative error and high correlation (2% and 0.088, respectively, Tab. 3, row ICED).

River ice CIIs for future periods are reported as changes in mean CII

between the reference 30 year period (1971-2000) to the future 30-year periods (2011-2040, 2041-2070, and 2071-2100). Mean CIIs for the 30 year periods are calculated for each ensemble member separately, and the spread in the

ensemble is presented using boxplots (as exemplified in Fig. 11) or shaded areas around a line representing the ensemble mean (as exemplified in Figs. 12-13).

The interquartile range (from 25 percentile to 75 percentile) is regarded as the “most likely” range of change. According to the model projections presented in Fig 11, the annual maximum ice thickness in the Lena river at Yakutsk 2071-

2100 (end of century) will thus most likely be reduced by approximately 35-55 cm under emission scenario RCP4.5, and 65-95 cm under RCP8.5.

Fig 11. Future changes in annual maximum ice thickness at 6 Yakutian river ice road crossings in the Lena, Vilyuy and Aldan rivers, simulated by HYPE model forced by 18 GCMs for the RCP 4.5 and

8.5 emission scenarios.

C3S_422_Lot1_SMHI – D5.1.1B | 20


The projected future reduction in the mean annual maximum ice thickness is

considerable lower at the other river ice road crossing, especially in the Viluy river (-30 to -60 cm depending on RCP) but also at the Khandaga crossing in the Aldan river (-30 to -80 cm). On the other hand, the reference period mean

annual maximum ice thickness was much smaller in the Viluy river crossings (~95-110 cm) compared to Lena at Yakutsk (156 cm) and Aldan at Khandaga

(178cm, Fig 11.), suggesting that towards the end of the century, the annual maximum ice thickness at the Vilyuy ice road crossing might be in the range 40-70 cm only, which might impact on the maximum vehicle weights. Also at the

Yakutsk the mean annual maximum river ice thickness might be considerably reduced compared to the current conditions (down to about 60-90 cm according

to the RCP8.5 results, Fig 11.).

For the web-based climate atlas we used a more graphical representation of the

changes in maximum ice thickness, as shown in Fig. 12 for the Lena river at Tabaga station.

Fig. 12. Projected change in annual maximum ice depth in the Lena river at Yakutsk between the reference period 1971-2000 to mid-century (2041-2070) for RCP8.5; -51 cm from 155 cm to 104 cm, and to end of the century (2071-2100); -81 cm from 155 cm to 75 m (ensemble mean), based on simulations with the HYPE model forced by air temperature and precipitation data from 18 bias-corrected CMIP5 GCM models.

Mean monthly ice thickness CII was calculated only from the model

simulations, and presented as changes from the reference period as exemplified in Fig. 13. Similar plots were generated for each river ice road crossing and

presented in the web-basin climatic atlas. It should be noted that in these plots, the uncertainty range is only represented by the interquartile range, for visibility

reasons. The data analysis in step 2 showed that significant decreasing trends in maximum ice thickness were found for March and April in the observations from Lena river at Tabaga station. The simulations of future period showed that the

changes continue to be largest in March-May (Fig. 13).

C3S_422_Lot1_SMHI – D5.1.1B | 21


Fig 13. Change in mean monthly ice thickness, Lena river at Tabaga, between reference period (1971-2000) and 3 future 30 year periods (2011-2040, 2041-2070, and 2071-2100) for emission scenario RCP4.5 and 8.5, as simulated with the HYPE model. The uncertainty range represent the 25 to 75%ile of the ensemble of model simulations forced by data from 18 CMIP5 GCM model

outputs.

First day with more than 30 cm ice thickness (DF30) is an important CII

representing the start of the ice road operation and management season. Unfortunately, the model evaluation showed that the quality of the model was

rather low for this CII (Tab. 3). The mean error was not too bad (3 days), but the ability of the model to explain the year-to-year variation was low (correlation =-0.01). These results further stress the room for improvement with regard to

modelling the seasonal ice thickness growth. It is interesting that the trend analysis on observation as well as on model simulations for the historical period

did not show any significant trend in DF30 and very few cases with significant trends in Date of first ice, DFI or Date of ice cover formation, DICF (Tab. 4), suggesting that so far the effects of increasing temperatures are mainly seen

in the ice thickness growth during mid-winter and not so much in the ice formation period in autumn. In the simulations of future periods, the shift if first

day with ice thickness larger than 30 cm are larger, up to 15 days delay under emission scenario RCP 4.5, and up to about 30 days towards the end of the century with RCP 8.5 (Fig. 14).

C3S_422_Lot1_SMHI – D5.1.1B | 22


Fig. 14 Future changes in first day with river ice thickness larger than 30 cm at 6 Yakutian river ice road crossings in the Lena, Vilyuy and Aldan rivers, simulated by HYPE model forced by 18 GCMs for the RCP 4.5 and 8.5 emission scenarios.

Date of first ice flow (DIFL), used here as indicator for ice breakup show at

significant decreasing trend in the observations (1.3 to 1.4 days earlier in spring per decade for the period 1962-2015) as shown both in Fig. 15 and Tab. 3. There is a tendency in the model simulations to predict ice breakup somewhat earlier

than in the observations, especially towards the end of the historical time period(Fig 15). On average, the simulated ice breakup is about 1 day too early

(Tab. 4) compared to the observations.

C3S_422_Lot1_SMHI – D5.1.1B | 23


Fig. 15. Day of ice flow in Lena river at Yakutsk river ice road crossing, from observations (black filled dots) and model simulations (red open squares), slope of trend lines (days/year).

In the future scenarios, the day of ice flow continue to shift to earlier days in

spring, however the rate of change seems to slow down compared to the historical period. The maximum change in ice breakup date is -15 days for all simulations between (1971-2000) and (2071-2100), except for emission scenario

RCP8.5 and the 30-year period 2071-2100 where the ice break up is 15-25 days earlier than in the reference period (Fig. 16). In general, the shift to earlier ice

breakup is larger at the Vilyuy river crossings than at the Aldan and Lena river crossings.

Lengths of ice period, ice cover period, and ice-free periods (IP, ICP, and NIP) all show significant decreasing and increasing trends in the historical period

1962-2015, respectively, either based on observations or on simulations (Tab. 4), as a consequence of the significant trends of earlier ice breakup plus non-significant trends in the ice formation dates (date of first ice and ice cover

formation). The model evaluation show that the model is able to predict mean value, year-to-year variation and trends in freezeup/breakup dates and ice

cover/ice free period lengths quite well (Tab. 3).

In the interactive climate atlas, future changes in freezeup/breakup dates was

illustrated with a so-called dash-board (Fig. 17) representing the different emission scenarios with either wind-mills (RCP 4.5) or factories (RCP 8.5). The

changes in the ice and non-ice cover periods were further illustrated using a simple but direct bar-diagram (Fig. 18).

C3S_422_Lot1_SMHI – D5.1.1B | 24


Fig. 16 Future changes in day of ice flow (ice breakup) at 6 Yakutian river ice road crossings in the Lena, Vilyuy and Aldan rivers, simulated by HYPE model forced by 18 GCMs for the RCP 4.5 and 8.5 emission scenarios.

C3S_422_Lot1_SMHI – D5.1.1B | 25


Fig. 17 Current conditions and future changes in day of river ice formation (freezeup) and ice flow (breakup) at the ice road over Lena river at Yakutsk, , simulated by HYPE model forced by 18 GCMs for the RCP 4.5 and 8.5 emission scenarios.

Fig. 18 Current conditions and future changes in ice cover period and days of river ice formation (freezeup) and ice flow (breakup) at the ice road over Lena river at Yakutsk, simulated by HYPE model forced by 18 GCMs for the RCP 4.5 and 8.5 emission scenarios.

Tab. 3. River ice CII evaluation based on observations and model simulations for period 1962-2015

Variable/CII Station Mean (1962-2015) Model validation

obs sim bias correlation rmse

Ice thickness

Daily (cm)1 Tabaga 102.4 104.8 2.4 0.88 21.6

C3S_422_Lot1_SMHI – D5.1.1B | 26


Maximum annual (cm) Tabaga 142.4 152.4 10.0 0.30 25.7

First day >= 30 cm (day of year)

Tabaga 323.7 321.2 -2.5 -0.01 25.7

Ice freeze-up/break-up dates (day of year)

First day with ice Tabaga 287.0 287.3 0.3 0.71 3.5

Yakutsk 288.1 286.4 -1.6 0.69 3.9

Khangalassy 287.6 284.2 -3.4 0.59 5.4

Day of ice cover formation, freezeup

Tabaga 306.8 306.9 0.0 0.55 6.7

Yakutsk 306.3 306.6 0.4 0.54 6.6

Khangalassy 305.4 305.0 -0.5 0.51 6.5

Day of ice flow, breakup

Tabaga 138.7 137.6 -1.1 0.78 3.7

Yakutsk 139.0 137.2 -1.7 0.76 4.0

Khangalassy 139.8 139.1 -0.6 0.76 3.7

Periods of ice/open water (number of days)

Ice period

(first ice-breakup)

Tabaga 217.0 215.6 -1.4 0.74 5.7

Yakutsk 216.1 216.1 0.0 0.76 5.1

Khangalassy 217.5 220.2 2.8 0.75 6.1

Ice cover period (freezeup-breakup)

Tabaga 197.1 196.0 -1.1 0.65 8.4

Yakutsk 197.9 195.7 -2.2 0.62 8.3

Khangalassy 199.5 199.3 -0.2 0.70 6.8

2Ice free period (breakup-first ice)

Tabaga 148 150 1.4 0.78 5.2

Yakutsk 149 149 0.1 0.77 5.0

Khangalassy 148 145 -2.8 0.78 5.5

1Evaluation based on the observations made every 10th day. 2It should be noted that the day of first ice flow does not represent ice-free conditions.

C3S_422_Lot1_SMHI – D5.1.1B | 27


Tab. 4. Lena river ice CII – trend analysis based on observations and model simulations for period 1962-2015. Variables with significant trends (Kendals p-value<0.05) are marked in bold.

Variable/CII (unit) Station Observations Model simulations

Trend/10yr p-value Trend/10yr p-value

Ice thickness

Maximum annual (cm) Tabaga -4.5 0.02 -2.6 0.02

First day>=30 cm (doy) Tabaga -0.3 0.98 -0.3 0.80

Ice freeze-up/break-up dates (day of year)

First day with ice Tabaga -0.2 0.70 0.8 0.03

Yakutsk 0.3 0.17 0.8 0.11

Khangalassy +1.1 4E-03 0.6 0.13

Day of ice cover formation, freeze-up

Tabaga 0.5 0.42 1.6 0.06

Yakutsk 0.6 0.27 1.1 0.16

Khangalassy 0.3 0.74 1.2 0.14

Day of ice flow, breakup Tabaga -1.3 3e-03 -1.9 2e-04

Yakutsk -1.4 2e-03 -2.0 7e-05

Khangalassy -1.4 0.01 -1.7 2e-04

Periods of ice/open water (number of days)

Ice period (first ice-breakup)

Tabaga -1.0 0.09 -2.6 4e-04

Yakutsk -1.8 0.01 -2.7 4e-04

Khangalassy -2.5 4e-05 -2.3 1e-03

Ice cover period

(freezeup-breakup)

Tabaga -1.8 0.02 -3.1 5e-04

Yakutsk -2.1 0.01 -2.9 9e-04

C3S_422_Lot1_SMHI – D5.1.1B | 28


Khangalassy -1.8 0.01 -2.9 2e-03

1Ice free period (breakup-first ice)

Tabaga +1.1 0.04 2.7 1e-04

Yakutsk +1.7 1e-03 2.5 3e-04

Khangalassy +2.7 7e-06 2.3 1e-03

1It should be noted that the day of first ice flow does not represent ice-free conditions.

C3S_422_Lot1_SMHI – D5.1.1B | 29


4.5 Step 5: Communication and Dissemination

Description: CIIs for future and historical periods based on both observations and impact

model simulations are disseminated to the public through the showcase interactive climate atlas. A more in-depth analysis of the results is compiled to

the local stakeholders through face-to-face meetings and written communication, including assessment of uncertainty of CIIs, recommendations for optimal operational period of ice road and ferry at Yakutsk, assessment of road stability

due to ground thawing and freezing in typical landscapes in Central Yakutia in present and future.

Results: Results of the case study were disseminated in several ways:

● showcase interactive climate atlas. Fig.19 shows comparison of mean monthly simulated ground thawing depth in pine forest with sandy soil for

the reference 1971-2000 and future 2071-2100 period for the RCP 8.5 emission scenario with spread between 18 GCMs from the atlas. Figs. 12, 13, 17, and 18 show how river ice CIIs were presented in the interactive

climate atlas (changes annual maximum and monthly mean ice thicknesses, and in freezeup/breakup dates and ice cover period length,

respectively). The atlas also includes an interactive map, where data from each river road crossing is provided by clicking the location in the map

(Fig. 20). The atlas is designed for general public and presents main findings in simple and illustrative way.

● internal meetings and other types of communications with clients.

● Russian and international scientific conferences and papers

Fig.19 Mean monthly thaw depth, m, in pine forest on sandy soil based on 18 GCMs, RCP 8.5 emission scenario

Based on communication with clients the probable short-term adaptation

measure to shortened ice roads seasons and thinner ice conditions could be an increased investment into artificial ice freezing at the ice crossings. Another possible short-term measure is to prolong the ferry navigation season to October

that is subject to high year-to-year variation and danger of ice formation on the

C3S_422_Lot1_SMHI – D5.1.1B | 30


ferry when operating at low temperatures. Longer periods between the end of

ferry use due to ice freeze up and formation of ice cover thick enough for driving, imply substantial higher costs for operating hovercrafts for people transportation and delivery of goods by air. Ground subsidence, road damage and collapse due

to deeper ground thawing lead to temporal road closure, higher investments to road repair and danger of accidents. Potential mid-term adaptation measures

include implementation of special techniques for ground thermal stability such as thermosiphon cooling and winter snow removal. Yakutia lacks long-term strategy of adaptation to climate change on local and

regional scale. Our clients admit urgent need of development of such a strategy. Produced CIIs will contribute to planning of adaptation and mitigation strategies

in present and future under climate change conditions.

5- Conclusion of full technical report The area of Yakutia is more than 3 million square km. Largest part of the

territory is covered by perennially frozen ground and does not have any year-round connection with other parts of Russia due to absence of bridges across the big rivers. Local stakeholders deal with design, construction and maintenance of

climate-dependent river ice crossings, ferry routes and roads on land in permafrost. Essential climate variables (in this case air temperature and

precipitation) for historical and future periods available in the C3S Climate Data Store was collected for the Yakutian region, and used as input to the impact models for simulating road and river ice conditions. Produced CIIs on thawing

depths for 21st century show contrasting response of different permafrost landscapes to expected climate change. According to extreme RCP8.5 emission

scenario permafrost would completely thaw in upper 5 m in three typical landscapes in Central Yakutia while thawing depth would increase only by 15-27 cm in other three landscapes. Historical observations of river ice thickness and

freezeup/breakup dates show significant trends of reduced ice thickness and shorter ice cover period lengths due to earlier breakup in spring (as a

consequence of the reduced maximum ice thickness). Model simulations for the future suggests that the river ice cover periods may shorten by up to 40 to 50 days under RCP8.5 towards the end of the century, but only up to 10-20 days in

the near future and under RCP4.5. Produced CIIs would contribute to development of both short-term adaptation

measures for safe construction and exploitation of transportation routes in Eastern Siberia and long-term strategy of adaptation and mitigation to climate change on local and regional scale.

C3S_422_Lot1_SMHI – D5.1.1B | 31


References Andersson. JCM., Pechlivanidis, IG., Gustafsson, D., Donnelly, C., Arheimer, B.

(2015) Key factors for improving large-scale hydrological model performance.

European Water 49:77-88.

Fedorov, A. N.,Botulu, T. A., and Varlamov, S. P.: Permafrost Landscape Map of

Yakutia ASSR, Scale 1:2500000, Moscow:GUGK, 1991 (in Russian).

Fedorov, A.N., Maximov, T.Ch.,Gavriliev, P.P. et al. Spasskaya Pad: Integrated

Investigations of the Permafrost Landscapes, Yakutsk, 2006, 210 pp.

Gelfan, A., Gustafsson, D., Motovilov, Y., Kalugin, A., Krylenko, I., Lavrenov, A.

(2017) Climate change impact on the water regime of two great Arctic rivers:

modeling and uncertainty issues. Climatic Change,141 (3): 499-515.

doi:10.1007/s10584-016-1710-5.

MacDonald, M. K., Stadnyk, T. A., Déry,S. J., Braun, M., Gustafsson, D., Isberg,

K.,& Arheimer, B. (2018). Impacts of 1.5 and 2.0 °C warming on pan-Arctic river

discharge into the Hudson Bay Complex through 2070. Geophysical Research

Letters, 45, 7561–7570. https://doi.org/10.1029/2018GL079147.

Varlamov S.P., Skachkov Yu.B., Skryabin P.N., Shender N.I. Thermal State of the

Upper Horizons of the Permafrost in Central Yakutia // Tenth International

Conference on Permafrost TICOP: Resources and Risks of Permafrost Areas in a

Changing World. Volume 2: Translations of Russian Contributions (Proceedings of

the Tenth International Conference on Permafrost. The Northern Publisher,

Salekhard, 2012). – Pp. 481-486

Yue, S., P. Pilon, B. Phinney and G. Cavadias, 2002. The influence of

autocorrelation on the ability to detect trend in hydrological series. Hydrological

Processes, 16: 1807-1829.

D422Lot1.SMHI.5.1.1B: Detailed workflows of each case ... · from 1951 to 2012. Reduced winter ice...

Documents

Transcript of D422Lot1.SMHI.5.1.1B: Detailed workflows of each case ... · from 1951 to 2012. Reduced winter ice...