Bob Adler Robert.F.Adler@nasa Yang Hong and George Huffman

Bob [email protected]

Yang Hong and George HuffmanNASA Goddard Space Flight Center, Greenbelt, MD 20771

National Aeronautic Space AdministrationNational Aeronautic Space Administration Goddard Space Flight Center, Greenbelt, Goddard Space Flight Center, Greenbelt, Maryland 20771Maryland 20771

Flood and Landslide Applications of High Time Resolution Satellite Rain Products

Motivation:

• Floods and landslides associated with heavy rainfall impact more people globally than any other type of natural disaster

• Detecting or forecasting such events globally is important to understand the causative processes, for mitigation, and potentially for warning

• Insufficient in situ data, long delays in data transmission and absence of data sharing in many trans-boundary river basins

• Recent satellite remote sensing data sets of precipitation and land surface characteristics (e.g., elevation, soil conditions, vegetation) are now available for research and potential applications in this area.

In this talk:

• A framework for global flood estimation

• A global landslide forecast technique--in early test phase

Global Monitoring/Warning Systems for Floods and Landslides

A Framework For Global-scale Flood Detection System

TRMM Multi-satellite Precipitation Analysis (TMPA)TRMM 3-hour real-time

Output/Decision Support System

Flood Inundation Map-

News ReportInventory

Satellite Inundation Information

Stream Flow Model

Hydrograph

WEB: 3-h update

Cell-to-CellFlow Routing

Basin/river networkFlow Accumulation

Implementation interface

Module. 1: Input

Rain-InfiltrationPartitioning

Qi

Pi(t)

Iii(t) Oik(t)QkQi

Grid-based Water Balance Model

Basin-based WaterBalance Model

Module. 3: Model

DEM, Topography, Slope gradient, Flow Direction, Flow Accumulation, River Network, watershed, Soil Property, Soil Moisture, Hydrological Soil Groups, Land Use, Vegetation cover, Field Capacity, Profile Water Content, Saturated Hydrological Conductivity

Land Surface Characteristics

Soil Parameters Routing Parameters Basin Parameters

Module. 2: Surface

Validation

Tropical Rainfall Measuring Mission (TRMM)Tropical Rainfall Measuring Mission (TRMM)TRMM Multi-satellite Precipitation Analysis

(TMPA) [3B42]Real-time update every 3-hour at http://trmm.gsfc.nasa.gov

7 day accumulation

TMPA uses polar-orbit microwave satellites (NOAA, DoD, NASA) and geosynchronous IR satellites, all calibrated by TRMM

Recent Typhoon-related heavy rains over Philippines

Module 1:Quasi-global rainfallModule 1:Quasi-global rainfall

From Space-borne Rainfall to Flood Potential Mapshttp://trmm.gsfc.nasa.gov

Simple rain amount thresholds: 24 hrs

ending 28 Oct. 2005 0000GMT

Threshold: 35 mm

131 mm MADRAS 214 mm NELLORE 234 mm CUDDAPAH 336 mm (13”) TIRUPATHI

From web site text page:

24 hr rainfall

NEWS STORY:More than 100 die in India floods More than 100 people have died in five days of heavy rains in the southern Indian states of Tamil Nadu and Karnataka, officials say.

More than 50,000 people have been evacuated from their homes in affected areas of Tamil Nadu.

Thousands of people have been displaced and air, rail and road services hit.

24 hrs ending 0300 GMT 19 Oct.

Potential Flood Areas from TRMM Web Site

http://trmm.gsfc.nasa.gov

Module 2: Development of Hydrologic Parameters at global scale Geospatial Database

(a) USGS GTOPO30 DEM (1km) (b) NASA SRTM DEM (30- or 90-m)

Topography Parameters:

Puerto Rico SRTM DEM (30-meter) Derived Slope (degree)

SRTM--Shuttle Radar Topography Mission

Global Slope (degree) Flow Accumulation (upper stream)

Global River NetworkFlow Length (cell to watershed outlet)

Module 2: Development of Hydrologic Parameters at global scale Level II Derived Flow Routing Parameters

Module 2: Development of Hydrologic Parameters at global scale Geospatial Database (not from satellite information)

Soil Parameters:

(b) Clay (%)

(c) Sand (%) (d) Silt (%)

(a) Soil Texture (b) FAO Soil Type Classification

Land Use/Vegetation Information from NASA MODIS

Spatial variation: each cell has its own CN Time-variant: Antecedent Precipitation Index

Surface Runoff Generated by uniform Rainfall (=100mm/hr) at normal moisture condition

Rainfall-infiltration Partitioning (Spatio-temporal variation)

CN=95

CN=85

CN=75

…..

CN=40

Fixed CN=75Various NAPI

Curve Number Approach

Module 3: Distributed Macro-scale Hydrological Model

Step I: Rainfall-infiltration Partitioning (Distributed and Time-variant)

Step 2: Flow Routing using Macro-scale Cell-to-Cell Algorithm

Step 3: Grid Point Hydrographs--Flood Inundation Mapping

Rain-InfiltrationPartitioning

Qi

Pi(t)

Iii(t) Oik(t)QkQi

Grid-based Water Balance Model

Basin-based WaterBalance Model

Constraints: computational efficiency, simplification of channel and hydraulic parameterization

Qi

Pi(t)

Iii(t) Oik(t)QkQi

Cell-to-Cell Flow Routing )()()()()( tEtOtQtIdt

tdSiiki

jji

i −−+=∑Cell-based Water Balance Model

Routing ParametersCell-based: Slope, soil type, Hydrologic conductivity, Porosity, Field Capacity, effective depth of soil column, flow direction, velocity, hydraulic radius, roughness coefficient, antecedent precipitation index;Watershed-based: flow length, area, flow accumulation, concentration time, flow time

Basin Rainfall (mm)

Water Depth [lower basin] (mm)

Case Study: Yangtze River flooding in Sept. 2005

Yangtze Basin-averaged Hydrograph for 2005

Water depth exceeding threshold= flood

Sept 01----- 05 (day)

Verified by Dartmouth Flood Archive and News: Sept 01-05, 2005 China – Typhoon Talim caused flooding and landslides. 129 dead, 30 missing. 1.84 million people evacuated. 62,000 houses collapsed, US$960 million damages.

Basin Area: 1,722,155 km2

Population: 386 million

Lower Basin

Preliminary Case StudiesYangtse River Basin Cell-based Flow Routing

2005 Day of Year

Cell Rainfall

Cell Water Storage/Depth (mm) after Flow Routing

(118.875oE, 31.125oN)

Grid (118.875oE, 31.125oN)

A grid-based hydrograph

Cell-based Hydrograph

19621970

Peru (Ancash)Peru (Ancash)

5,000 death18,000 death

10/30/1998 Nicaragua >2000 death

10/08/2005 Solola, Guatemala >1800 death

2/17/2006 Layte, Philippines buried entire village( 1500)

1/10/2005 La Conchita, CA 12 death

01/04/2006 Jakarta, Indonesia entire village

Landslides Death Toll

Landslides/Mudslides/Debris Flow• Landslides are one of the most widespread natural

hazards on Earth, responsible for thousands of deaths and billions of dollars in property damage every year.

• Rainfall is the primary causative factor.

• Currently, no system exists at regional or global scale to detect heavy rainfall that may trigger landslides.

Entire village buried.

GGlobal lobal RRainfall-induced ainfall-induced LLandslide andslide FForecast orecast SSystemystem

DEM, Slope, AspectTopography

Curvature, Concavity

Morphology

Lithological makeup Geology

Sand, Foam, Silt, ClaySoil Property

Shrub, barren, builtup Land Cover

Soil Moisture, FD, FA Hydrology

Landslide Susceptibility

Real-time Rainfall Estimation

NASA TRMM-based

Surface controlling factors

When

Where

How big

Risk

Damage

Detection/Warning

Slope-Stability Hierarchical Decision Tree

Decision MakingInventory Data

Rainfall Trigger Intensity-Duration

ClassificationSoil Moisture

SlidingProbability

Hong et al., IEEE TGRS (accepted)

Landslide Susceptibility Category-1: Water Bodies0: Permanent Snow/Ice1: Very Low Susceptibility2: Low Susceptibility 3: Moderate Susceptibility4: High Susceptibility (yellow)5: Very High Susceptibility (orange)

Landslide Susceptability Map

Hong et al., Submitted to J. of Natural Hazards

DEM, Slope, AspectTopography

Curvature, Concavity

Morphology

Lithological makeup Geology

Sand, Foam, Silt, ClaySoil Property

Shrub, barren, urbanLand Cover

e.g., Soil MoistureHydrology

Surface controlling factors

Percentage of Grid Boxes in Each Category

(a) (b)

Hour Day

Influence of Rainfall Characteristics on the Timing and Occurrence of Landslides

Philippines Landslide and TRMM Rainfall Accumulation

Philippines

Feb 8-17, 2006

1500 Deaths

Note that bars are the rainfall intensity and star denotes the timing of landslide occurrence.

Hong et al., Submitted to GRL

GGlobal lobal RRainfall-induced ainfall-induced LLandslide andslide FForecast orecast SSystemystem

Decision MakingInventory Data

Rainfall Trigger Intensity-Duration

SlidingProbability

Rainfall Map Intensity-Duration Susceptibility/ Landslide Warning

TRMM Near Real-Time Rainfall at location 76.875 W, 4.125 N, April 13, Columbia 1) the last 24 hour rainfall accumulation > 103mm 2) The Susceptibility Map shows high or very high susceptibilityNews Report: 13 Apr 2006, At least 34 people missing in Colombian mudslide

Summary and Future Work

NASA-based data sets of near real-time precipitation observations and land surface characteristics are being combined to develop a Flood and Landslide Monitoring Systems.

Next: 1) Evaluation/implementation of the first-cut, experimental systems for user

feedback (January 2007 is goal to have real-time experimental systems running in real-time);

2) Flexible Module Structure: open for new component plug-ins for testing;

3) Use of multiple precipitation estimates for ensemble forecasts;

4) Use of NWP model(s) precipitation forecasts to lengthen forecast applicability

NASA TRMM-based Global-scale Flood/landslide System

Watershed-based VIC Simulation (1998-1999): La Plata Basin5000+ Water basins derived from DEM

The La Plata Basin

Credit: U. of Washington Lettenmeier Group