r.debrisflow User manual and model outline
Contents
1. Raster map of elevation meters gt Soil depth map m 112131415161 Raster map of soil depth in meters If soil depth is not known it should be set to a high value For bedrock it has to be set to 0 gt Hydrological surface classes map in teger 111213141516 Integer raster map of the distribution of the hy drological surface classes 1 for defined flow channel 2 for multiple flow channel or uncon centrated overland flow Care has to be taken that the cell size of the map corresponds to the cell size of the simulation it shall be used for Particularly the defined channels class 1 have to be clean and continuous gt Channel width map 1 213 41516 Raster map denoting total width of flow chan nel s for each cell connected to the hydrological surface classes 1 defined flow channel width of the flow channel 2 multiple channels or un concentrated overland flow ratio between sum of width of all flow channels crossing a cell and cell width perpendicular to the steepest slope The advantage of this approach is to be largely independent from cell size at least for quite uni form distributions of channels Also here the cell size has to correspond to the cell size of the simulation it shall be used for Additionally it has to fit to the hydrological sur face classes map exactly in order to avoid serious problems during simulation gt Objects at risk map boolean 11 2 31 415 6 Bo2. 1 during at least one time step are considered to fail at the deepest failure plane identified for the cell during the event Failed soil with a sediment concentra tion of Crit lt Crax where Coi 1 H 1 s is con sidered to evolve into a debris flow In reality debris flows with higher sediment concentration do occur particularly in non cohesive soils The model therefore assumes that all failed soil with amp 0 develops into a debris flow also at at higher sediment concentrations e for every cell where runoff is modelled to evolve into a debris flow sediment concentration C is tested against C after each time step If C gt Cn the material is retained from sediment load All retained material is routed downslope as debris flow at the end of the last time step Before routing the debris flow downwards the vol ume and the size of each patch of cells of potential debris flow initiation are calculated If one of these variables or the depth of potential initiation is below user defined thresholds the patch or the cell respec tively is excluded from runout 1 5 2 Routing procedure The debris flow is not simply routed downwards the steepest slope Similar to surface runoff the routing algorithm is determined by the hydrological surface class e HSC 1 defined channel the debris flow is routed through the channel with only one possi ble downward direction from each cell As soon as deposition occurs in a cha
3. Resources Research 42 9 p Gamma P 2000 dfwalk Ein Murgang Simulationsprogramm zur Gefahrenzonierung Geographica Bernensia G66 144 pp In German Green W H amp Ampt G A 1911 Studies on soil physics Journal of Agricultural Sciences 4 1 24 Perla R Cheng T T amp McClung D M 1980 A Two Parameter Model of Snow Avalanche Motion Journal of Glaciology 26 197 207 Rawls W J Brakensiek D L amp Miller N 1983 Green Ampt infiltration parameters from soil data Journal of Hydraulic Engineering 109 62 70 Rickenmann D 1990 Bedload transport ca pacity of slurry flows at steep slopes Mitteilungen der Versuchsanstalt f r Wasserbau Hydrologie und Glaziologie ETH Z rich Nr 103 249 p Rickenmann D 1999 Empirical Relationships for Debris Flows Natural Hazards 19 47 77 Wichmann V 2006 Modellierung geomorphologischer prozesse in einem alpinen Einzugsgebiet Abgrenzung und Klassifizierung der Wirkungsr ume von Sturzprozessen und Muren mit einem GIS Eichst tter Geographische Arbeiten 15 231 pp In German Xie M Esaki T amp Cai M 2004a A time space based approach for mapping rainfall induced shallow landslide hazard Environmental Geology 46 840 850 2 User s manual 2 1 System requirements r debrisflow was developed and tested under Fedora Core 6 with GRASS 6 2 1 It probably works on mo st other UNIX systems as well as with other versions of GRASS The mod
4. is only required to a lesser extent for example by introducing maxi mum rates of detachment or by finding a model which suits better for steep terrain e The slope stability model as used in this version is in a strict sense only able to predict shallow translational slope failures in cohesionless soils For the future it would be desirable to include slope geometry and potential failure planes in a way that also deeper seated rotational failures in cohesive soils can be predicted e All the mobilized material is kept at its place until the last time step and is then routed downwards together This is more realistically for failed mate tial than for detached material but for both mechanisms a way should be found to let the al gorithm decide when a patch of unstable material moves downwards as debris flow e At the moment it is attempted to implement a deterministic runout model based on the Savage Hutter equation also used for snow avalanches to GRASS as a raster module named r avalanche 3 2 Ease of use One of the next steps shall be to improve the quality of display of the results Module 5 of r debris flow sh In its current version the operation of r debrisflow runs very much on text files and the command line For the future it would be desirable to have a user interface facilitating data management and at least partly replacing r debrisflow sh However such im provements is given a lower priority at the mome
5. reclass tables for deriving secondary parameters from input datasets the subfolder colors contains colour tables for display temp The temp directory contains temporary files created during the execution of r debrisflow sh Its content shall not be manipulated manually but only using the functionalities of r debrisflow sh results It contains some simulation results summary file documentation file compare below However the main results are stored as rasters in the active GRASS mapset docs The docs directory contains this manual 12 Appendix 3 Manuals 2 4 Installation r debrisflow has to be added to the GRASS raster li brary as a new module based on the source code of the file main c For performing this task log in as su per user su call the script install sh in the folder r debrisflow tools cd dir r debrisflow tools sh install sh dir may be any location in your home directory The following prompt is displayed Full path to GRASS source slashes at be ginning and end Here enter the path to the GRASS soutce for ex ample usr local src grass64_release The Makefile is created and compilation and installa tion are run automatically so that r debrisflow is ready to use Please note that you have to change to the tools di rectory as described above if just entering sh dir r debrisflow tools install sh an error message will display and r debrisflow will not be installed
6. 0 m 5 15 1 for wedge shaped towards the front 2 for even deg 20 30 14 Appendix 3 Manuals Table 3 Spatially distributed input parameters of r debrisflow specified in Module 1 or automatically created in Module 2 of r debrisflow sh respectively shortcut description unit catchment definition of catchment of interest boolean elevation elevation above sea level m soildepth depth of soil m soilclass class of soil integer landcovclass class of land cover integer hsc hydrological surface class integer chanwidth width of flow channel m ratio riskobj definition of objects potentially at risk boolean dinitdef depth of debris flow initiation m dscourdef maximum depth of entrainment m snow depth of snow cover m soil presence of soil boolean alpha local slope angle deg my friction coefficient for runout coeff icp interception capacity of the vegeta m tion croot root cohesion N m droot rooting depth m nman_add vegetation surcharge for Man summand ning s n textclass texture class of soil integer d30 d30 grain diameter of soil m d50 d50 grain diameter of soil m d90 d90 grain diameter of soil m skeleton skeleton content of soil gt 2 mm ratio gammad dry specific weight of soil N m csoil soil cohesion N m phi soil angle of internal friction deg nman_bas basic value for Manning s n summand pref factor for preferential macropore factor flow to be multiplied with infiltration capacity thetar soil residual wa
7. 2 5 Data management r debrisflow uses text files and rasters with predefined names as input The shell script r debrisflow sh serves for creating these datasets and for generating secon dary datasets from primary information e g hydraulic conductivity from grain size class It must be run from within the used GRASS mapset cd dir r debrisflow tools sh r debrisflow sh r debrisflow sh offers the following modules gt Parameter and data input gt Preparation of parameters gt Execution of simulations gt Post processing of model output Display of results gt Removal of result files gt Cleaning of file system gt Exit AINA HBWNHE l l v By entering a number the corresponding module is executed The modules are described in detail below r debrisflow sh has to be called from within the GRASS mapset with all the required raster maps 2 5 1 Parameter and data input Module 1 consists of a sequence of prompts for input data and parameters If no input is given for a prompt by just pressing ENTER the dataset specified earlier is kept The numbers behind the prompts denote the modes of simulation compare Chapter 1 for which the corresponding dataset is required gt Catchment map boolean 1 2 3 41 516 Boolean raster map defining the catchment of in terest identified by cell values of 1 areas out of the catchment are defined by 0 or no data gt Elevation map m 1 213141516
8. as the exceed the critical temperature gt Parameter file 1 2 3 4 516 File with list of parameters for simulation one per line The parameters have to be specified in a defined order Table 2 In the folder dir r debrisflow data a file with example parameters example_param txt is provided gt Legend file 1 213141516 File with user specified values serving as maxima for the display of the output maps In the folder dir r debrisflow data an ex ample for a legend file example_legend txt is 2 5 No provided The order shown in this file has to be kept for all legend files gt File with coordinates of control points 1 2 3 4 File with coordinates of some specific points for which some variables are documented for each time step Each line contains one coordinate first line the x coordinate of the first point sec ond line the y coordinate of the first point third line the x coordinate of the second point etc Please note that if you do not wish to specify control points the file has to exist anyway oth erwise r debrisflow will produce an error mes sage but may be empty 2 Preparation of parameters user inputs are required for this module which serves for the automatic derivation of secondary in put parameter maps from the maps specified in Mod ule 1 A slope raster map is derived from the eleva tion model r slope aspech and a boolean presence of soil map is deri
9. atically according to Eq 14 using the maximum flow velocity of the previous time step over all cells A is set to 20s if doit Vmax exceeds a threshold value If no runoff oc curs at all A yom is set to 120 s The depth of the water table R for each cell is com puted as follows R R T M ZIF OF Eq 15 where T is the effective rainfall M is the snow melt XIF stands for the total inflow from all the upslope cells directly draining into the considered cell and OF stands for the outflow all values in meters 1 3 Sediment transport model 1 3 1 Basic assumptions Surface runoff independently of occurring as over land flow or channel flow has a certain capacity to transport sediment If the actual load is below trans port capacity soil from the bed is eroded whilst sediment is deposited in the reverse case The follow ing assumptions are set in the model e only bedload is considered as relevant regarding the magnitude of sediment transport and the evolution of debris flows Suspended load is ne glected e runoff is considered to follow hydraulic princi ples to a certain threshold of sediment concentra tion at higher sediment concentrations it is con sidered as debris flow 1 3 2 Detachment and sediment concentra tion The Rickenmann 1990 equation is used in the model for estimating sediment transport because it is best suited for relatively steep channels and high sediment concentrations It only include
10. atio of the de bris flow and v7 is the debris flow velocity of the previous cell The factor amp and the coefficient 7 are derived as follows 6 g sina ucosa Bas _ 72h Eq 33 m D where g is gravitational acceleration 9 81 m s on the earth surface a is local slope angle x is the dimen sionless friction coefficient and L m is slope length cell size corrected for slope angle Aa is the differ ence between the slope angle of the previous cell and the slope angle of the considered cell which is set ot 0 for convex slopes or channels Wichmann 2006 For concave slopes v is corrected as the flow loses en ergy Eq 34 Vi V 1 0 cos a u a if ait gt aj The first term in Eq 31 determines if the flow accel erates gt 0 or decelerates lt 0 the second term provides the contribution of flow velocity to the final velocity M D being a surrogate for the inertia of the flow exerts a major influence on flow velocity while its impact on runout distance is small The latter is primarily determined by the topography and yp Gamma 2000 Wichmann 2006 The simulation is stopped as soon as v becomes undefined square root of negative value compare Eq 31 One problem regarding the calibration of this model is that different combinations of the two parameters to be calibrated M D and y may result in the same runout distance A common way for calibration is therefore to set M D to values lea
11. ators It is important how ever that it does exist and is separated from the ac tual value by a tabulator as only the part of the file after the tabulator is used Please consult the example files starting with example_ as reference e gt Precipitation file mm 1 213 4 Table 2 Input parameters single values for r debrisflow description grain specific weight of soil User s manual 13 File with precipitation values from measured data or hypothetical header information o first line elevation of rain gauge meters o second line duration of basic time step sec onds o third line downwards precipitation values mm sum of one time step per line gt Temperature file C 1121314 File with temperature values from measured data or hypothetical Header information o first line elevation of thermometer meters o second line daily minimum temperature CO third line daily maximum temperature C forth line temperature for computing snow melt C o fifth line critical temperature rain fall snowfall boundary C o sixth line vertical gradient for daily mini mum temperature C m o seventh line vertical gradient for daily maximum temperature C m o eighth line degree day factor for snow melt m CH o Ninth line downwards temperature values one time step per line unit examples of value s N m 26 500 for quartzitic soil exponent for weighting of
12. cepts as well as its mode of operation Every user is encouraged to report encountered bugs ot errors to 1 Model outline 3 1 1 General aspects 3 1 2 Hydraulic model components 4 1 2 1 Precipitation and snow melt 4 1 2 2 Interception 4 1 2 3 Evapotranspiration 5 1 2 4 Infiltration 5 1 2 5 Surface runoff 6 1 3 Sediment transport model 6 1 3 1 Basic assumptions 6 1 3 2 Detachment and sediment concentration 7 1 4 Slope stability model 7 1 5 Debris flow runout 8 1 5 1 Initiation 8 1 5 2 Routing procedure 8 1 5 3 Runout distance and deposition 9 1 6 Model validation 9 1 7 Acknowledgements 9 martin mergili boku ac at Furthermore the developer would be grateful for re ceiving comments regarding e experiences with the program shortcomings recommendations for improvements scientific concepts ease of use e parameters chosen for certain study areas e interest in cooperation in application and further development The model as applicable with GRASS is running under the GNU General Public License www gnu org rdebrisflow has been created with the purpose to be useful for modelling of debris flows It has been developed with care and much emphasis has been put on ensuring its scientific value Nevertheless every user has to be aware that it is only a computer program created by a human being which may contain technical and topical errors and shortcomings No responsibility can be taken by the develop
13. debrisflow itself in order to design satisfactory lay outs 2 5 6 Removal of result files All results rasters and text files produced by r debrisflow and Module 4 are deleted All temporary files created in the modules 1 and 2 are kept so that new simulations may be performed immediately 2 5 7 Cleaning of file system All temporary rasters and text files created by the modules 7 and 2 are deleted They must be re run in order to perform new simulations 2 5 8 Exit r debrisflow sh is exited and the default cell size is restored 18 Prospected improvements 3 Prospected improvements 3 1 Scientific concepts This document describes Version 1 3 of r debrisflow A number of optimizations are prospected for future versions e The channel parameters for the hydrological sur face classes T and 2 are rough estimates from field observations and orthophotos For the fu ture it would be useful to have a tool for auto matically extracting these features from a high resolution elevation model in order to allow for a more objective down scaling to the cell size at which r debrisflow is finally run e Water from rainfall and snow melt is assumed to immediately concentrate in the flow channels In troducing some sort of time of concentration could be useful e The sediment transport model has to be cali brated with field data at the moment ST to ST It would be desirable to improve the model in a way that this calibration
14. ding to realistic ve locities and then calibrating w in order to correlate simulated and observed runout distances Wichmann 2006 used values of M D 75 m The following relationship for y was found to be useful for computing the maximum runout length Gamma 2000 u 0 134 Eq 35 Model outline 9 where A km is the catchment size for the consid ered cell It is assumed that would decrease with in creasing 4 because the water content of the debris flow would increase This relationship was used in r debrisflow but with user defined factor and expo nent in order to allow calibration for other condi tions Following Gamma 2000 the range of values of u would be restricted to a maximum of 0 3 and a minimum of 0 045 overruling Eq 35 The two patameter friction model does not say any thing about the patterns of particle entrainment and deposition Instead of designing a more complex scheme like Wichmann 2006 simple thresholds of slope and velocity are used for delineating these proc esses in r debrisflow where entrainment as far down as to the wetting front is only assumed if both pa rameters are above the threshold whilst deposition is assumed to take place only if both parameters are be low the thresholds The calibration of the thresholds is connected to the same problems as the calibration of M D and 4 different combinations of parameter values 1 6 Model validation r debrisflow is mainly based o
15. e wetting front and do m is the depth of the wetting front before infiltration Eq 4 is derived from Darcy s law Xie et al 2004 Chen amp Young 2006 If no meas urements of the soil hydraulic parameters are avail able values for different texture classes can be ob tained e g from Rawls et al 1983 or from Carsel amp Parrish 1988 Two possible cases have to be distinguished fhas to be cotrected for volumetric stone content s which does not affect the maximum possible depth of infil tration but the volume that fits into the soil until this Model outline 5 depth For each time step it is tested which case is applicable e case 1 Ro gt f Atun 1 5 inflow to the cell ex ceeds maximum possible infiltration and depth of sutface water table after infiltration Ry m is ex pressed as R Ro SAt hor 1 s Eq 5 Infiltration is limited by the infiltration capacity and the depth of the new wetting front d m is computed as follows d do T At f A0 Eq 6 where Ad is the moisture deficit of the soil dif ference between saturated water content 6 and initial water content 6 all in m m 4 e case 2 Ro lt f Aton 1 5 inflow to the cell is equal or smaller than maximum possible infiltra tion capacity In this case the entire inflow infil trates d d R A s Eq 7 and no surface water table remains meaning that no surface runoff will develop from the consid ered cell infiltrat
16. ed for each cell as well as the vegetation surcharge of Manning s N Madd e the surface water table is characterized by depth R m and flow velocity vg m s e soil here rather understood as sediment cover than as mixture of residuals from weathering and decomposed organic matter is basically repre sented by a nominal soil class raster layer A tex ture class dry specific weight y N m3 stone content s m m the hydraulic parameters 6 0 m m 3 d m and K m st and the me chanical parameters soil cohesion N m and angle of internal friction p are assigned to each soil class as well as minimum and maximum val ues for Mas Soil depth d m is defined independ 4 r debrisflow ently from the soil classes All parameters are considered constant over the entire depth of the soil column e bedrock is considered unconditionally stable Its permeability for water is accounted for by the pa rameter p denoting the ratio of total effective rainfall and snow melt which percolates through the rock p is not defined as raster map but globally for the considered study area The following sections give a more detailed introduc tion to the physical and mathematical background of r debrisflow 1 2 Hydraulic model components 1 2 1 Precipitation and snow melt Precipitation P m and air temperature T C are read from the prepared datasets Precipitation is con sidered as rainfall P if it exceeds a user defi
17. els and soil in between flow channels In be tween flow channels ZIF and OF are zero compare Eq 13 and 14 The integral form of the Green Ampt approach compare e g Xie et al 2004 is not used as the model is run in short time steps with varying rainfall intensi ties 1 2 5 Surface runoff After computing infiltration the ponded water of the depth Ry is assumed to concentrate in the flow chan nels immediately and to run off superficially Strictly spoken R Apm Pre where Ayo m stands for the cross section of the flow and P m is the wetted perimeter but in the model R is approximated by flow depth Runoff velocity vg m st is computed using the Manning formula 1 2 t Vow a R 3 sin a man Eq 8 where a is the local slope angle in degrees and man is the surface roughness determined by vegetation soil texture and obstacles which is computed using Pinan m n ig naa Eq 9 where m is a factor accounting for meandering as is the basic value and faq is a surcharge for vegeta tion obstacles etc is automatically set to 1 0 since the model presented here is designed for steep terrain with poor meandering of the channels The water discharge per unit width q m s is com puted as follows 1 ae ee 4 VaR R 3 sin a N man Eq 10 Surface runoff is computed separately for each hydro logical surface class HSC e HSC 1 defined channel the water is routed through
18. er for any types of deficiencies in the program or in the present document or for the consequences of such deficiencies 1 8 References 9 2 User s manual 11 2 1 System requirements 11 2 2 Test dataset 11 2 3 File management 11 2 4 Installation 12 2 5 Data management 12 2 5 1 Parameter and data input 12 2 5 2 Preparation of parameters 14 2 5 3 Execution of simulations 15 2 5 4 Post processing of model output 15 2 5 5 Display of results 16 2 5 6 Removal of result files 17 2 5 7 Cleaning of file system 17 2 5 8 Exit 17 3 Prospected improvements 18 3 1 Scientific concepts 18 3 2 Ease of use 18 1 Model outline 1 1 General aspects The ideas for most of the model components were taken over from existing models partly in a modified form The model framework was named r debrisflow the r indicates that it is a GRASS raster module It was kept relatively simple in its first version presented here but was also designed in a way for allowing to be extended with more sophisticated modules in the future compare Chapter 3 r debrisflow was implemented using a 2 5D raster data model the vertical dimension plays an important role but is only quantified by attributes It combines physically based deterministic modules and modules based on empirical relationships r debrisflow couples a hydraulic model a slope stability model a sediment transport model and a debris flow runout model e The deterministic hydraul
19. ic model distributes the water from precipitation or snow melt among vegetation interception soil infiltration and surface runoff It then approximates the soil water status and the runoff variables e the deterministic slope stability model computes the factor of safety for each cell based on an in finite slope stability model and identifies poten tial starting areas of debris flows e the sediment transport model based on an em pirical approach provides an estimate for ero sion and deposition by surface runoff allowing to assess the tendency of bedload rich runoff to develop into a debris flow e the debris flow runout model finally routes the debris flow downwards to the area of deposition based on a two parameter friction model The modules are executed in a defined sequence for a user defined number of time steps during and after a rainfall or snow melt event Slope stability and runout are computed at the end of the last time step Not all modules have to be executed the following combinations modes of simulation are possible when running r debrisflow e 1 full mode all modules are executed e 2 geotechnical mode A the sediment transport model is excluded and only debris flows starting from slope failures are modelled for conditions where it is known that debris flows only develop from slope failures Model outline 3 e 3 geotechnical mode B the runoff and sediment transport models are e
20. ify 1 else 0 gt Height of monitor in pixels Please specify the height of the monitor for dis play a value between 500 and 800 is recom mended depending on the size of your monitor e gt Observed patterns of debris flow ini tiation line vector A line vector map with areas of debris flow initia tion observed in the field may be specified if available for facilitating the evaluation of the model results gt Observed patterns of debris flow depo sition line vector A line vector map with areas of debris flow de posits observed in the field may be specified if available for facilitating the evaluation of the model results A monitor opens and a prompt with instructions ap pears in the terminal The maps are displayed in a de fined sequence please enter the number of steps to move forward or backward or exit to quit If you have moved a defined number of steps fore or backwards and would like to apply the same action again you can just press ENTER to do so User s manual 17 If you have chosen to export the maps as jpg no prompts appear but all maps are displayed and ex ported automatically and the module is terminated Please note that the size of the monitor and the placement of some of the elements of the maps leg end bar scale are not suitable for all map width to height ratios It may happen that some of the place ment options have to be changed in the shell script r
21. in r debrisflow sh you have to copy the input files to the dir r debrisflow temp directory as prec txt precipitation temp txt tem perature param txt parameter file and ctrlpoints txt control points These files shall not contain labels but only the values 2 5 4 Post processing of model output All the resulting raster maps are cleaned cells outside of the defined catchment are set as no data and the sediment balance from debris flows is computed For the modes of simulation 1 and 4 the sediment bal ance from surface runoff as well as the sediment con centration for each time step and the maximum sediment concentration are computed Some of the major resulting maps are prepared for display com pare below In order to ensure comparable legends a maximum value is assigned to each type of legend depth of wetting front factor of safety sediment concentration failure detachment and entrainment deposition sediment balance debris flow index depth and velocity of surface runoff depth of load The legend file has to be stored in dir r debrisflow data and to be specified during the data and parameter input compare above The color tables have to be stored in dir r debrisflow data colors Furthermore three prompts do appear when calling the module Each task is accepted by typing T or de nied by typing 0 e gt Calculate statistics boolean Maximum values of runoff velocity runoff depth and load dep
22. ion limited by infiltration capacity Ving lt V surf infiltration limited by surface water table Ving V surf old wetting front Vint AO 1 s Vint Aos new saturated soil already saturated soil Lo eS Fos gt Fo i e F bs Hd 24S 45 j i potential depth of infiltration A L soil at initial moisture content Figure 2 Infiltration into the soil according to the Green amp Ampt 1911 model as applied for the present study Vsurt volume of surface water before infiltration Vint infiltrated volume Designed by M Mergili For case 1 s has no direct influence on d for case 2 d increases with increasing s The application of this method has to be considered as an approximation e The Green Ampt approach in its strict sense was developed for horizontal surfaces but is also applied for slopes Chen amp Young 2006 showed that on slopes until 45 the effect of slope angle is small compared to other sources of inaccu racy e stone content is not accounted for in the original model Therefore it was decided to disregard its influence on infiltration capacity but its role as 6 Model outline limiting the infiltrable volume was taken into ac count More rescarch would be necessary in or der to clarify the inaccuracies connected to this simplification Slope parallel seepage is neglected in the model The infiltration is computed separately for soil below flow chann
23. key 1 S 2 LS 3 SL 4 SCL 5 SC 6 L 7 CL 8 SIL 9 C 10 SICL 11 SI 12 SIC It is possible to skip Module 2 and instead create all the parameter maps manually In this case please note that the naming conventions have to be met ex actly in order to ensure the functionality of the simu lation compare Table 3 2 5 3 Execution of simulations Prompts for mode of simulation and cell size are dis played gt Mode of simulation gt Cell size m integer User s manual 15 The options for the mode of simulation are described in Chapter 1 the corresponding number has to be entered The cell size has to be chosen in accordance with the input datasets and the required level of de tail For test simulations it is recommended to choose larger cell sizes in order to reduce computing time After specifying these two parameters the GRASS raster module r debrisflow is called by pressing EN TER Additionally to an array of raster maps a sum marty file summary txf and a documentation file doc txt are written and stored in the dir r debrisflow results directory The summary file contains variables par ticularly volumes for each basic time step The documentation file contains variables for the coordi nates specified as control points cfripoints txt for each short time step compare Table 4 for the vari able names Please note that if you wish to run r debrisflow manu ally not from with
24. llows Coulomb s law s c oltan p Eq 23 and the corresponding shear resistance force is T N tangt c Ax cos Eq 24 where N is the normal force p degree is the angle of internal friction c N m is the cohesion soil co hesion c plus root cohesion c Ax m is the length of the considered slope segment in downslope direc tion and a degree is the slope angle N and T N are computed from the weight of the moist soil G N G y Ax d Eq 25 N G cosa Eq 26 T G sina Eq 27 where y N m gt is the specific weight of saturated soil and d m is the depth of the potential failure plane r 1 7 9 4 s 1 ya is the specific weight of dry soil and y is the spe cific weight of water both in Nm ya is derived from grain specific weight to be specified by the user 26 5 kN m for quartzitic material and 6 and s as surrogates for pore volume Eq 28 The seepage force exerted by the soil water is stated as F Ax d y sina Eq 29 Dry and cohesionless soils F 0 c 0 p ya are stable when a lt y and unstable when a gt The forces exerted by the surface water table R m are neglected in the model 1 5 Debris flow runout 1 5 1 Initiation Debris flows are supposed to occur within a certain range of sediment concentrations usually between Cyin 0 45 and Cmax 0 55 e At the end of the last time step all cells identified as potentially unstable with FOS lt
25. n physically based con cepts but nevertheless it contains empirical elements and an array of parameters some of which are diffi cult to measure or to estimate Therefore some type of calibration of some of the parameters is requited for each study area For this purpose datasets for validation are needed for example e if the debris flows hit a road records by the road authorities about volumes to be temoved after debris flow events connected to known meteoro logical conditions or e the distribution and extent of landslide scars and patterns of deposition from debris flows visible in the field 1 7 Acknowledgements Funding was provided by the Tyrolean Science Funds the Vice Rector for Research of the University of Innsbruck Doktoratsstipendium aus der Nach wuchsf rderung der LFU and the Austrian Acad emy of Sciences Furthermore the valuable remarks of Wolfgang Fellin Institute of Infrastructure University of Inns bruck are acknowledged 1 8 References Arcement G J amp Schneider V R 2000 Guide for Selecting Manning s Roughness Coeffi cients for Natural Channels and Flood Plains USGS Water supply Paper 2339 67 p 10 Model outline Carsel R F amp Parrish R S 1988 Developing joint probability distributions of soil water re tention characteristics Water Resources Research 24 755 769 Chen L amp Young M H 2006 Green Ampt infil tration model for sloping surfaces Water
26. ne vector observed patterns of deposition from debris flow line vector test_dscourdef test_init_observed test_depd_observed The test dataset is suitable to run r debrisflow at spa tial resolutions of 5m or 10 m Please note that the User s manual 11 reclass tables described below are suitable for the soil and land cover classes in the test dataset for other datasets they have to be modified 2 3 File management The file system behind r debrisflow consists of two parts e a GRASS mapset with all the spatially distributed input information as raster or vector maps and e a folder named r debrisflow which may be stored at any location in your home directory The inter nal structure of the folder r debrisflow may not be manipulated otherwise some of the function alities will fail The directory r debrisflow contains the following sub directories tools The tools directory hosts the scripts required for in stalling and running r debrisflow e main c the source code for r debrisflow e r debrisflow sh a shell script facilitating data in put management and output compare below e install sh a shell script helping to compile the soutce code main c data It contains the input files of the meteorological data the parameter file and the file with control points compare below and a file for scaling the legends of the resulting maps to be displayed The subfolder recl contains
27. ned tem perature threshold Ti usually ranging between 0 C and 2 C Snow melt M m is computed using a sim ple degree day approach with air temperature as the only input EM ddf T yy Eq 1 degree day air precipitation factor DDF temperature T R where XM is the daily snow melt m dt and ddf is the degree day factor m C d Tapis the tempera ture in C at a defined time of the day In order to estimate snow melt of shorter time intervals daily snow melt is distributed over the considered day fol lowing a linear relationship with temperature M T ZM T Eq 2 where To is the temperature during the cosidered time step XT is the daily temperature sum based on the length of one time step A s Only temperatures above Tir are included in the sum 1 2 2 Interception The interception capacity of the vegetation Ips m is extracted from the land cover dataset For each time step A s rainfall is retained as interception Ly m until the interception capacity is reached La Ipo The excess rainfall is added to the soil water table R m as effective rainfall P g m Water from snow melt is considered not interceptable by vegetation This worst case assumption was cho sen due to the often unknown vertical distribution of Ipor interception capacity input surface water depth R effective rainfall P eff parameter J working parameter major output soil hydr
28. nel s is corrected for detachment and deposition Eq 18 to 20 which are not part of the original Rickenmann model are based on two rough generalizations e the bedload moves at the same velocity as the water Model outline 7 e the bedload discharge immediately reaches an equilibrium only if ST ST 1 1 4 Slope stability model The hydraulic model components supply saturated depth d m It is assumed that e slope failures only occur at the depth d the wet ting front e if total soil depth is known slope failures are also allowed to occur at the soil bedrock interface but only if the entire soil is saturated mathemati cally identical to slope failures at the wetting front An infinite slope stability model Figure 3 is used for the calculations Therefore a wide ratio between slope length and depth of the failure plane is required in order to yield an acceptable approximation a condi tion that is usually met for shallow but not for deep seated failures Furthermore infinite slope stability models assume a translational failure mechanism which usually only occurs in cohesionless soils For cohesive soils the model may still derive reasonable approximations of the factor of safety but is strictly spoken not really applicable y submerged unit weight Yw unit weight of water R ponding depth hydr radius d depth of potential failure G S weiahtiof moist soil plane depth
29. nnel the corre sponding cells are considered as HSC 2 for the further simulation e HSC 2 no clearly defined channel a random walk weighted for downslope angle is applied for routing the debris flow The weight is deter mined automatically as a function of the steepest slope It is expressed as war pe where the exponent 2 has to be specified by the user values between 3 and 5 appear reasonable p degree is the slope angle where deposition starts compare below 0 for upslope angles Similar to runoff this algorithm accounts for the higher tendency of debris flows to take another than the steepest slope specified in the DEM on gentle slopes than on steep slopes Eq 30 Each cell containing starting material for a debris flow is passed through the routing procedure indi vidually Routing continues until the debris flow has stopped according to the criterion specified in the next section 1 5 3 Runout distance and deposition Runout is computed using a semi deterministic two parameter friction model developed by Perla etal 1980 which was modified by Gamma 2000 and applied by Wichmann 2006 in a raster based GIS environment It is not only applicable to debris flows but also to snow avalanches The deterministic element of the approach is the ve locity of the debris flow v m s which is computed for each raster cell y al en Hve cos Aa Eq 31 where M D m is the mass to drag t
30. nt than those regarding the scientific concepts behind the model
31. of wetting front gt N normal force ne G y Axd T shear force E a Fz seepage force T G sina F Axdyy sina az Shear restistance force p T N tang c Ax cosa FOS Tf T Fs 3 FOS gt 1 cell is stable i FOS lt 1 cell is pot unstable Figure 3 Mechanisms for infiltration and shallow slope failure as applied in the present study For a detailed explanation compare text Designed by M Mergili As discussed above the infiltration model only con siders vertical seepage Infinite slope stability models in contrast usually assume a slope parallel flow ex erting a destabilizing seepage force F N parallel to the slope compare Eq 29 In reality the direction of the seepage depends on the local conditions particu larly on the presence or absence of an impermeable layer Fully including the slope parallel seepage into the slope stability calculations is therefore a worst case assumption The slope stability model is exe cuted after the computation of the infiltration has been completed last time step so that a slope parallel seepage can be assumed without contradic 8 Model outline tion to the vertical seepage computed with the Green Ampt model The dimensionless factor of safety FOS is stated as FOS T T F Eq 22 where Ty is the shear resistance force of the soil T is the shear force and F is the seepage force all in N compare Figure 3 Shear resistance s N m fo
32. olean raster map denoting objects at risk 1 for presence of potential objects at risk like roads or buildings else 0 or no data gt Soil classes map integer 1112131415 Integer taster map with predefined soil classes compare Module 2 Preparation of parameters The numbers of the classes may be chosen freely but must fit to the corresponding reclass tables compare Table 3 gt Land cover classes map integer 111213141 Integer raster map with predefined land cover classes compare Module 2 Preparation of pa rameters The numbers of the classes may be chosen freely but must fit to the corresponding reclass tables compare Table 3 gt Estimated depth of mobilization of soil map m 6 Raster map showing the patterns of estimated potential debris flow initiation depth in meters gt Estimated depth of entrainment of soil map m 1516 Raster map showing the patterns of estimated potential depth of entrainment by debris flows meters gt Snow depth map m 1 2 3 4 Raster map denoting depth of snow cover me ters The remaining four required input datasets are the names of files which have to exist in the dir r debrisflow data directory In each of the files each line has to consist of a label first column and a value second column The content of the label only serves for enhancing readability it has no influence on the simulation but it may not contain tabul
33. ological parameter parameters S 0r Os W K model slope a Manning s depth of wetting model in older versions a flow directions routing procedure flow discharge q grain size distribution D30 D50 D90 sediment slope a grain spec weight yg Rickenmann formula infinite slope stability model v v concentration C of criteria Manning formula front d soil dry specific roughness n weight y s work flow hydrological all time steps surface class flow velocity v work flow angle of internal friction rooting depth d f 1 last time step v factor of safety FOS I debris flow volume runout distance runout velocity loss of elevation depth of detachment or deposition from surface runoff d w depth of detachment or deposition from debris flow slope a Figure 1 General model layout of r debrisflow Designed by M Mergili 1 2 3 Evapotranspiration Potential evapotranspiration Ey m is set to zero This is a worst case assumption again which was cho sen for two reasons e also the most simple equations for evapotranspi ration require highly dynamic parameters that are usually not available at sufficient accuracy and resolution humidity irradiation etc e the model is designed primarily f
34. or short and in tense rainfall events where evapotranspiration is rather negligible Regarding snow melt neglect of evapotranspiration may lead to more significant inaccuracies 1 2 4 Infiltration Infiltration runoff and sediment transport have to be calculated at much shorter time steps AZo than the other processes meaning that the entire sequence has to be repeated for various times within each basic time step AZ Anis determined according to Eq 14 The water input from effective rainfall and the snow melt are added to the surface water table of the cell at the beginning of each short time step At short P tg Eq 3 Ry Royo M where Ry is the depth of the surface water table at the end of the previous time step The infiltration of water into the soil is a complex process influenced by an interplay of factors like the depth of the surface water table the soil parameters and the local topog raphy It was chosen to use the Green amp Ampt 1911 approach assuming a sharp wetting front as the inter face between saturated soil above and soil at initial moisture content below Figure 2 The hydraulic pa rameters governing infiltration are derived from the grain size class of the soil Infiltration capacity f m s 1 can be stated as _ Ro y u Rte 0 Eq 4 where K m s is the hydraulic conductivity Ro m is the depth of the surface water table before infiltra tion m is the matric suction at th
35. r debrisflow f Version 2012 02 08 A model framework for simulating mobilization and movement of debris flows based on GRASS V GIS User manual and model outline by Martin Mergili Institute of Applied Geology BOKU University of Natural Resources and Life Sci ences Vienna Austria mattin mergili boku ac at www mergili at February 2012 2 r debrisflow r debrisflow is a GIS supported model framework for simulating the potential spatial patterns of debris flow initiation movement and deposition It is physically based in general but includes some empirical statistical components r debrisflow is designed as a raster module for the software GRASS The scientific concepts behind r debrisflow are summarized in Chapter 1 Model outline In contrast to most of the other GRASS raster mod ules management of the data and the parameters in put and output is not done by adding parameters to the r debrisflow command but by an additional shell script r debrisflow sh with various functionalities including derivation of secondary input parameters from primary ones This is required due to the com plexity of the model framework Instructions how to operate r debrisflow are given in Chapter 2 User s manual The model shows a large potential for refinements Chapter 3 Prospected improvements will give a short overviev of ongoing and prospected further de velopment of r debrisflow regarding the scientific con
36. s bedload The original equations mainly derived from labora tory tests yielded very high values of detachment when applied to the study areas Furthermore the equation does not say anything about detachment rates For these reasons the dimensionless calibration parameters ST ST STs and ST had to be intro duced 12 6 D90 ME eu a a Mina Fa 16 where qp m s t is the volumetric bedload transport per unit width s is the ratio between grain and fluid densities D90 and D30 m are the grain sizes where 90 and 30 per weight respectively are finer q m s is the fluid discharge per unit width and a degree is the local slope angle qo m s t is the threshold discharge for sediment transport qa ST 0 065 s 1 g D50 sine Eq 17 where D50 m is the median grain size and g m s is the gravitational acceleration Erosion detachment of soil or deposition d m depth of bedload m and sediment concentration C m m can then be detived d ST l q v fot h lt q v Eq 18 d ST I q v fot h gt q v Eq 19 l 1 d 4 v Eq 20 C 1 I R Eq 21 where m is the depth of bedload at the start of the time step Negative values of d Eq 18 indicate de tachment positive values Eq 19 indicate deposition Only saturated soil is allowed to be detached All the sediment deposited is considered as saturated and the depth of the wetting front below the flow chan
37. slope angle for surface runoff random walk exponent 3 exponent for weighting of slope angle for debris flow runout random exponent 4 walk percolation through rock ratio compared to effective rainfall plus ratio 0 0 1 0 snow melt factor for calibration of critical runoff formula factor for calibration of potential sediment load formula factor for calibrating detachment by surface runoff factor for calibrating deposition from surface runoff factor 1 0 factor 0 005 0 01 factor 0 1 cell sizes factor minimum sediment concentration for development of debris flow ratio 0 45 maximum sediment concentration for development of debris flowin ratio 0 55 cohesive soil minimum depth of unstable or detached soil for initiation of debris flowm various minimum number of adjacent failed cells for development of debris integer flow depends on cell size minimum volume of unstable or detached soil for development of de m various bris flow factor for computing u compare Eq 35 exponent for computing u compare Eq 35 lower threshold for u upper threshold for u mass to drag ratio slope threshold for entrainment deposition velocity threshold for entrainment deposition maximum depth of debris flow deposit options for distribution of material deposited from debris flow integer Minimum slope angle for initiation of debris flows factor 0 13 exponent 0 25 coefficient 0 045 0 15 coefficient 0 3 m 75 deg 15 ms 1
38. software products like ArcGIS The resulting files are stored in dir r debrisflow results asc In the output rasters the display during program exe cution and in the summary and documentation files every variable is addressed by a shortcut Ta ble 4 The names of the resulting raster maps have the prefix r_ Depths raster maps and documentation file start with d The volumes depicted in the display during simulation and in the summary file have the prefix vol_ instead of d The number at the end of the raster map names indicates the time step Rasters except those of runoff and sediment transport vari ables are only written for the pre last depth of wet ting front or last time step all other variables For example the raster r_ddepd20 shows the depth of deposition from debris flow for each pixel at the end of time step 20 while vol_depd indicates the vol ume deposited from debris flow over the entire study area 2 5 5 Display of results Some of the major resulting maps can be displayed using this module The following parameters have to be specified gt Azimuth of the sun for shaded relief map All maps are displayed with a shaded relief as background the azimuth of which has to be specified in decimal degrees recommended 315 gt Export maps to jpg 1 0 The displayed maps can be automatically stored as jpg graphics in dir r debrisflow results jpg If you wish to do so please spec
39. ter content ratio thetas soil saturated water content ratio psi matric suction at wetting front m k soil hydraulic conductivity ms derived from specified as input in Module 1 soildepth elevation land covclass soilclass textclass modes of simulation 23 45 6 name of reclass table dir r debrisflow data recl xxx 1 x x x x x x x x x KK KK KK XK x KKK KK KK x KK KK KK OK x KK KK KK OK x KK K XK x x x x x x Kx KK XK x Kx KK XK lt x x KK XK x KK K XK x recl_Icv_icp4_min txt recl_Icv_icp4_max txt recl_lcv_croot_minO txt recl_Icv_croot_max0 txt recl_Icv_droot_min1 txt recl_Icv_droot_max1 txt recl_nman_add_min3 txt recl_nman_add_max3 txt recl_soil_classO txt recl_soil_d305 txt recl_soil_d505 txt recl_soil_d905 txt recl_soil_s3 txt recl_soil_gammado txt recl_soil_c0 txt recl_soil_phi1 txt recl_nman_bas_min3 txt recl_nman_bas_max3 txt lt x Kx KKK XK x Kx Kk x x x KK KK KK X x x Xx Xx x Kx KK KK KK OX recl_soil_pref_min1 txt recl_soil_pref_max1 txt x x recl_hyd_thetar3 txt recl_hyd_thetas2 txt recl_hyd_psi5 txt recl_hyd_k9 txt x x XxX XK x x kK XK x x K XK x x K XK Temperature is only relevant when including snow melt or when it is lower than the critical temperature separating rainfall and snow fall If the raster map of snow depth is 0 m over the en tire catchment the specified values have no in fluence on the results of the simulation as long
40. th over the entire event are computed Please note that the option is not ap plicable to all modes of simulation 1 2 and 4 for the runoff variables 1 and 4 for load depth 16 Appendix 3 Manuals Table 4 Output from r debrisflow after running Module 4 r raster map s summary file d documentation file shortcut description unit versions time steps mode of simulation r s d 1 2 34 5 6 vflow flow velocity of surface runoff m s x x all xx x vflow_max maximum flow velocity ms x be X x dflow depth of surface runoff m x x all x xX x dflow_max maximum depth of surface runoff m x x xX x dload depth of sediment load of surface runoff m x x all x x dload_max maximum depth of load m x x x streampower stream power as additional information Nm s x all x xX x deltaddetw detachment by surface runoff short time m x x x step ddetw cumulative detachment by surface runoff m x xX all x x ddetw_val detachment by surface runoff basic time m x all x x step deltaddepf deposition from surface runoff short time m x x x step ddepf cumulative deposition from surface runoff m x xX all x x ddepf_val deposition from surface runoff basic time m x all x x step dbudget_wat cumulative sediment balance from water flow m x all x x dbudget_wat_val sediment balance from water flow time step m x all x x csed sediment concentration of surface runoff ratio x all x x csed_max maximum sediment concentration ratio x x x dwfront_chan depth of wet
41. the channel with only one possible downward ditection ftom each cell e HSC 2 slope with numerous small channels or no channels at all the water is routed down wards assuming the defined channel densities on a sub cell scale and a random walk weighted for slope angle w a Eq 11 where is the weight assigned to each potential flow direction and 7 is a user defined exponent values of 3 to 4 appear reasonable on gentle slopes inaccuracies of the DEM or landforms on a sub cell scale may exert a stronger effect on flow direction than on steep slopes For both cases inflow ZIF m is computed with the Manning formula in the same way i n v At i n 3 1 LPs RS ey L Aton R sin a 2 a d i i l A dpi i Eq 12 man i where is the number of contributing upslope cells d i m is the horizontal distance between the centre of the celli and the centre of the considered cell Outflow OF m is computed in an analogous way V fowAt short OF R Eq 13 h where dp m stands for the horizontal distance be tween the centre of the considered cell and the centre of the downslope cell The length of one short time step AZn S is defined as At ort dou y max P Eq 14 where a is a factor lt 1 set to 0 5 dey m is the cell size and vmx ms is the maximum runoff velocity over the entire area Too short time steps would un necessarily increase computing time A is defined by the program autom
42. ting front below flow channel s m x pre last X X X dwfront_int depth of wetting front between channel s m x pre last xx x os factor of safety below flow channel s ratio x last x x X x dfailpot potential depth of slope failure m x xX last x xX X x dinit_fail depth of debris flow initiation from slope fail m x xX last XXX x ure dinit_detw depth of debris flow initiation from detach m x xX last x x ment by surface runoff dinit total depth of debris flow initiation m x last X X XXX dscour depth of entrainment by debris flow m x xX last X Xs Xo Xe X X ddepd depth of deposition from debris flow m x xX last x xX X X xX xX dbudget_deb sediment balance from debris flow m x last X X XS EX X idepG indicator for entrainment or deposition by integer x last X Xuan X Kae X X debris flow according to the two parameter friction model idepR indicator for debris flow incidence according integer x last X X Xen X X x to Rickenmann 1999 equation rcoef runoff coefficient ratio x all x XXX deltat length of short time step s x all XxX X XX X X e gt Extract values for time steps boo lean This option is only applicable for the modes of simulation 1 and 4 Raster maps of detachment deposition and sediment balance for each time step are extracted from the rasters of cumulative values written by the simulation gt Export maps to ascii boolean The resulting raster maps are exported as ascii rasters in order to be usable with other GIS
43. ule itself should also be usable under cygwin but the related shell script r debrisflow sh compare below would probably not work Please make sure to have a proper installation of GRASS before installing rdebrisflow In case of doubt please consult www grass itc it 2 2 Test dataset A test dataset is provided together with the program consisting of some text files and a GRASS location with the name test_rdebrisflow All data is packed in the file test_rdebrisflow zip test_prec txt and test_temp txt contain precipitation and temperature data in the format described below while test_param txt is the required parameter file test_ctrlpoints txt contains the control points com pare below Table 1 shows the names of the raster and vector maps Table 1 Names of the maps of the test dataset description mask for study catchment raster elevation map at 5 m resolution raster elevation map at 10 m resolution raster soil classes raster soil depth raster land cover classes raster hydrological surface classes at 5 m resolution raster hydrological surface classes at 10 m resolution raster width of flow channel s at 5 m resolution raster width of flow channel s at 10 m resolution raster road object at risk raster depth of snow cover raster predefined depth of debris flow initiation raster predefined maximum depth of en trainment raster observed patterns of debris flow initiation li
44. ved from the soil depth map Sec ondary parameter maps are derived from the land cover and soil classes using reclass tables compare Table 3 r reclass All the reclass tables have to be stored under dir r debrisflow data asc and must exactly be named as shown in Table 3 All the reclass tables must correspond to the following pattern derived value_l derived value_2 original value_l original value_2 original value_n derived value_n end for example 1 34 3 5 10 37 end All numbers must be left aligned and all equal signs have to stand in one column The numbers before the txt extension of the reclass files denote the factor with which the original derived values have to be multiplied before writing them into the reclass table This is necessary because the r reclass module of GRASS is not able to cope with non integer values After reclassification the magni tude of the derived raster maps is corrected automati cally For parameters with max or min at the end of the name of the reclass table the pixel values of the cre ated raster map are randomly distributed between the minimum and the maximum values If this is not de sited the maxima and minima have to be identical Example reclass tables are stored in the abovemen tioned directory but the tables have to be modified for each study area except the tables for the hydraulic parameters which refer to grain size classes according to the following
45. xcluded like 2 but ex cluding the influence of runoff on infiltration for conditions where it is known that no surface runoff develops e 4 hydraulic mode the slope stability model is excluded and only debris flows developing from sediment laden runoff are modelled for condi tions where it is known that slope failures play no tole for the mobilization of debris flows e 5 fully saturated mode it is assumed that the en tire soil in the study area is saturated With this precondition the slope stability model and the runout model are computed e 6 runout only mode only the runout model is computed with defined areas of debris flow ini tiation for testing the plausibility of the runout model for events of known patterns of debris flow initiation and deposition The general model layout is illustrated in Figure 1 r debrisflow considers the slope or catchment under investigation as six layered system chatacterized by the following variables e the overlying atmosphere is described by air temperature T degree Celsius and precipitation P m e snow cover is defined as snow depth d m e land cover is defined by a raster layer represent ing nominal land cover classes Minimum and maximum values of interception capacity ICP m root cohesion Nm and rooting depth d m have to be assigned to each class A hydrological surface class and the width of the flow channels at a sub cell scale are defin
Download Pdf Manuals
Related Search