landslide susceptibility mapping to inform land-use management decisions in an altered climate

1
Landslide Susceptibility Mapping to Inform Land-use Management Decisions in an Altered Climate Muhammad Barik and Jennifer Adam Washington State University, Department of Civil & Environmental Engineering, Pullman, WA. ABSTRACT: The Olympic Experimental State Forest (OESF) is a commercial forest lying between the Pacific coast and the Olympic Mountains. As this area is critical habitat for numerous organisms, including salmon, there is a need to investigate potential management plans to promote the economic viability of timber extraction while protecting the natural habitat, particularly in riparian areas. As clear-cutting reduces the strength of the soil, and as projected climate change may result in storms with higher intensity precipitation, this area may become more susceptible to landslide activity. This may result in potentially severe consequences to riparian habitat due to increased sediment loads. Therefore, this study was performed with the objective to quantify the impacts of land cover and climate changes on slope stability. A physically-based hydrology model, the Distributed Hydrology Soil Vegetation Model (DHSVM) with the sediment module, was used for this analysis. To map areas susceptible for landslides, logging was done for different combinations of soil, vegetation and slope classes. To investigate the impacts of climate change on landslide susceptibility we applied downscaled output from two General Circulation Models (GCMs) with two greenhouse gas emission scenarios. An understanding of the areas most vulnerable to landslides in an altered climate and due to timber harvesting will help in the development of sustainable forest management practices. The study domain for this research is the Queets basin, located on the Olympic Peninsula in northwest Washington State (see inset at top). The DHSVM mass sediment module was applied to a select tributary of the Queets basin (shown at left). Area (km 2 ) Elevation (m) Average Annual Precipitatio n (m) Averag e Annual Flow (m 3 /s) 1560 0-2200 3.55 121 Table: Queets river characteristics Figure: Observed and simulated stream flow between 2001 and 2004. Peak flows and dry season flows are underestimated by the model. For the calibration period (2001-2005),the Nash Sutcliffe coefficient is 0.62 while the volume error is 8%; and for the evaluation period (1995- 1999) they are 0.58 and 11%, respectively. For evaluation of the mass wasting module, we compared observed and simulated landslide area for storms occurring in 1985 and 1981. 8.Conclusions We have presented an approach to quantify the effects of logging and future climate on landslide susceptibility, which can be developed as a decision-making tool for forest management. Findings from this analysis include: Landslide frequency increases substantially in harvested areas and the increment is related to geography, soil and vegetation types. Certain combinations of soil, vegetation, and slopes are more vulnerable to landsliding than others. Mapping of susceptible landslide areas based on the study shows vulnerable areas for timber harvesting. This information can be used for long-term forest management planning. For 2045 projected climate new areas come under high susceptibility class, thus climate change should be includes for long term planning. ACKNOWLEDGEMENTS: Funding for this project is being provided by the State of Washington Water Research Center (SWWRC). 2.Study Domain Model: DHSVM (Wigmosta et al. 1994) with its mass wasting module (Doten etal.2006) The key component for this study, the mass wasting module, is stochastic in nature. Projected climate change for 2045 was used with the same possibility between 1984 and 1990. Outputs are from CGCM3.1_t47 and CNRM_cm3 GCM models for A1B and B1 emission scenarios. Results obtained for these two models are averaged. Schematic diagram for DHSVM mass wasting module (Doten et al., 2006) 3.Model Description Q Q sed Slide Year Historic Landslides Total Surface Area(m 2 ) Total Surface Area(m 2 ) of All Cells Factor Safety <1 (From Modeled Run) 1985 10614 11400 1981 15257 13678 The figure on the left shows an susceptibility index calculated by weighting method and classified into there classes. 1990 logging scenario detected by Landsat-5 image, and 8yrs(1990-1997) of historic landslides were projected on the map and they showed significant relationship(see table on right). Figure (a) shows the landslides susceptibility classes for Queets with timber harvesting in the basin for historic climate. Red marks in Figure (b) & (c) show new areas with Increased susceptibility level due to climate change. (b) and (c) figures are respectively for A1B &B1 carbon emission scenarios. The table above shows number of historical landslides in the harvested areas of the basin for different landslides susceptibility zones. Higher range shows higher number of landslides. Analysis showed for 2045 projected climate areas with high landslide risk increased on average 7.1% for B1 and 10.7% for A1B carbon emissions scenarios. 1.The Goal 4.Model calibration & Evaluation c 7.Results of climate change From the table we can see that the model simulations were close to historical landslides of 1985 and 1981. DHSVM Hydrology Model Soil, vegetation, DEM and mask file input at 150m resolution Run Mass Wasting Module for historical case Run Mass Wasting Module for wide- spread logging in different slope, elevation, soil and vegetation classes Weighting factor (W f ) calculated for each cell of the basin and classified Landslides risk zoning map for timber harvesting Meteorology Input DHSVM Mass Wasting Module for 10m Resolution Compared to historic al landslid es 5.Methodology for Mapping Landslide Risk due to Timber Harvesting A weighting factor (W f ) is applied to each thematic layer (e. g. slope, soil, elevation, vegetation) to calculate the vulnerability, according to the following method: 1/9/2001 7/28/2001 2/13/2002 9/1/2002 3/20/2003 10/6/2003 4/23/2004 11/9/2004 0 500 1000 1500 2000 Observed and simulated streamflow Date Discharge (m3/s) 6.Landslides and Harvesting The objective of this study is to predict the long term effects of timber harvesting under projected climate change on slope stability. The overarching goal of this approach is to develop a decision making tool that can be used by forest managers to make long-term planning decisions. Wi = the weight given to a certain parameter class (e.g. a soil type, or a slope class). Densclas = the landslide density within the parameter class. Densmap = the landslide density within the entire map. Npix(Fi) = number of pixels, which Contain landslides, in a certain parameter class. Npix(Ni) = total number of pixels in a certain parameter class. Suscept ibility Class Segment ation No of landsli des cell in the suscept ibility class No. of total cells in the suscept ibility class Percent age of landsli des in a suscept ibility class Low(<.0 5) 621 28049 2.2 Medium( .05-.79 ) 617 25099 2.5 High(>0 .79) 627 19021 3.3 a b c

Upload: effie

Post on 11-Feb-2016

57 views

Category:

Documents


2 download

DESCRIPTION

Q. Q sed. Landslide Susceptibility Mapping to Inform Land-use Management Decisions in an Altered Climate Muhammad Barik and Jennifer Adam Washington State University, Department of Civil & Environmental Engineering, Pullman, WA. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Landslide Susceptibility Mapping to Inform Land-use Management Decisions in an Altered Climate

Landslide Susceptibility Mapping to Inform Land-use Management Decisions in an Altered Climate

Muhammad Barik and Jennifer Adam

Washington State University, Department of Civil & Environmental Engineering, Pullman, WA.

ABSTRACT: The Olympic Experimental State Forest (OESF) is a commercial forest lying between the Pacific coast and the Olympic Mountains. As this area is critical habitat for numerous organisms, including salmon, there is a need to investigate potential management plans to promote the economic viability of timber extraction while protecting the natural habitat, particularly in riparian areas. As clear-cutting reduces the strength of the soil, and as projected climate change may result in storms with higher intensity precipitation, this area may become more susceptible to landslide activity. This may result in potentially severe consequences to riparian habitat due to increased sediment loads. Therefore, this study was performed with the objective to quantify the impacts of land cover and climate changes on slope stability. A physically-based hydrology model, the Distributed Hydrology Soil Vegetation Model (DHSVM) with the sediment module, was used for this analysis. To map areas susceptible for landslides, logging was done for different combinations of soil, vegetation and slope classes. To investigate the impacts of climate change on landslide susceptibility we applied downscaled output from two General Circulation Models (GCMs) with two greenhouse gas emission scenarios. An understanding of the areas most vulnerable to landslides in an altered climate and due to timber harvesting will help in the development of sustainable forest management practices.

The study domain for this research is the Queets basin, located on the Olympic Peninsula in northwest Washington State (see inset at top).

The DHSVM mass sediment module was applied to a select tributary of the Queets basin (shown at left).

Area (km2) Elevation (m) Average AnnualPrecipitation (m)

Average Annual

Flow (m3/s)

1560 0-2200 3.55 121

Table: Queets river characteristics

Figure: Observed and simulated stream flow between 2001 and 2004. Peak flows and dry season flows are underestimated by the model.

For the calibration period (2001-2005),the Nash Sutcliffe coefficient is 0.62 while the volume error is 8%; and for the evaluation period (1995-1999) they are 0.58 and 11%, respectively.

For evaluation of the mass wasting module, we compared observed and simulated landslide area for storms occurring in 1985 and 1981.

8.Conclusions

We have presented an approach to quantify the effects of logging and future climate on landslide susceptibility, which can be developed as a decision-making tool for forest management. Findings from this analysis include:Landslide frequency increases substantially in harvested areas and the increment is related to geography, soil and vegetation types. Certain combinations of soil, vegetation, and slopes are more vulnerable to landsliding than others.Mapping of susceptible landslide areas based on the study shows vulnerable areas for timber harvesting. This information can be used for long-term forest management planning.For 2045 projected climate new areas come under high susceptibility class, thus climate change should be includes for long term planning.

ACKNOWLEDGEMENTS: Funding for this project is being provided by the State of Washington Water Research Center (SWWRC).

2.Study Domain

Model: DHSVM (Wigmosta et al. 1994) with its mass wasting module (Doten etal.2006)The key component for this study, the mass wasting module, is stochastic in nature.

Projected climate change for 2045 was used with the same possibility between 1984 and 1990. Outputs are from CGCM3.1_t47 and CNRM_cm3 GCM models for A1B and B1 emission scenarios. Results obtained for these two models are averaged.

Schematic diagram for DHSVM mass wasting module (Doten et al., 2006)

3.Model Description

Q

Qsed

Slide Year Historic Landslides TotalSurface Area(m2)

Total Surface Area(m2) of All Cells Factor Safety <1(From Modeled Run)

1985 10614 114001981 15257 13678

The figure on the left shows an susceptibility index calculated by weighting method and classified into there classes.

1990 logging scenario detected by Landsat-5 image, and 8yrs(1990-1997) of historic landslides were projected on the map and they showed significant relationship(see table on right).

Figure (a) shows the landslides susceptibility classes for Queets with timber harvesting in the basin for historic climate. Red marks in Figure (b) & (c) show new areas with Increased susceptibility level due to climate change. (b) and (c) figures are respectively for A1B &B1 carbon emission scenarios.

The table above shows number of historical landslides in the harvested areas of the basin for different landslides susceptibility zones. Higher range shows higher number of landslides.

Analysis showed for 2045 projected climate areas with high landslide risk increased on average 7.1% for B1 and 10.7% for A1B carbon emissions scenarios.

1.The Goal 4.Model calibration & Evaluation

c

7.Results of climate changeFrom the table we can see that the model simulations were close to historical landslides of 1985 and 1981.

DHSVM Hydrology Model

Soil, vegetation, DEM and mask file input at 150m

resolution

Run Mass Wasting Module for historical

case

Run Mass Wasting Module for wide-spread

logging in different slope, elevation, soil and

vegetation classes

Weighting factor (Wf) calculatedfor each cell of the basin and classified

Landslides risk zoning map for timber harvesting

Meteorology Input

DHSVM Mass WastingModule for 10m

Resolution

Com

pare

d to

his

toric

al

land

slid

es

5.Methodology for Mapping Landslide Risk due to Timber Harvesting

A weighting factor (Wf ) is applied to each thematic layer (e. g. slope, soil, elevation, vegetation) to calculate the vulnerability, according to the following method:

1/9/2001 7/28/2001 2/13/2002 9/1/2002 3/20/2003 10/6/2003 4/23/2004 11/9/20040

500

1000

1500

2000 Observed and simulated streamflow

Observed SimulatedDate

Dis

char

ge (m

3/s)

6.Landslides and Harvesting The objective of this study is to predict the long term effects of timber harvesting under projected climate change on slope stability. The overarching goal of this approach is to develop a decision making tool that can be used by forest managers to make long-term planning decisions.

Wi = the weight given to a certain parameter class (e.g. a soil type, or a slope class).Densclas = the landslide density within the parameter class.Densmap = the landslide density within the entire map.Npix(Fi) = number of pixels, which Contain landslides, in a certain parameter class.Npix(Ni) = total number of pixels in a certain parameter class.

Susceptibility Class Segmentation

No of landslides cell in the susceptibility class

No. of total cells in the susceptibility class

Percentage of landslides in a susceptibility class

Low(<.05) 621 28049 2.2

Medium(.05-.79) 617 25099 2.5

High(>0.79) 627 19021 3.3

a

b

c