remote sensing of rainfall · advantages, disadvantages and limitations? advantages: • perhaps...

REMOTE SENSING OF RAINFALL

http://www-calipso.larc.nasa.gov/about/constellation.html

Amir AghaKouchak & Soroosh SorooshianCenter for Hydrometeorology & Remote Sensing

University of California Irvine


Hydrologic Cycle

Precipitation

Source of Fresh Water

Precipitation


Severe Floods

Precipitation


Severe Droughts

Precipitation

Mea

suri

ng

Pre

cip

itat

ion

WHYM

easu

rin

g P

reci

pit

atio

n

HOW

Mea

suri

ng

Pre

cip

itat

ion

HOWACCURATE

Mea

suri

ng

Pre

cip

itat

ion

WHYWater Resources Management

Disaster Preparedness

Short-Term Decision Making

Precipitation

Precipitation Measurement DevicesM

easu

rin

g P

reci

pit

atio

n

HOW - Rain Gauges

- Weather Radars

- Satellites

Rain Gauges

Recording Gage

(a) Tipping bucket

0.01 inch or 0.2 mm / tipping

(b) Weighting type

It can measure other forms of precipitation, including hail and snow. It is more expensive and require more maintenance than tipping bucket gauges.

(c) Float Recording

Rain Gauges

Wind-Induced Undercatch

Influencing Factors:

– Wind speed

– Temperature

– Gauge type

– Gauge height

– Windshield

– Exposure

Limitations: drops will stick to the sides or funnel of the collecting device.

Rain may fall on the funnel and freeze.

Snow can block the funnel

Nespor and Sevruk, 1999

Gauges Networks

Rain Gauge Networks:

Advantages, Disadvantages and Limitations?

Dingman, 2002

U.S. National Weather Service operates:

≈ 3,500 recording gagesand ≈ 11,000 non-recording gages

Suggested minimum number of gages

1. Flat regions: tropical and relatively uniform 235 ~ 350 mi2 / gage2. Arid and Polar regions 600 ~ 4,000 mi2 / gage3. Mountainous region 40 ~ 100 mi2 / gage4. Small mountainous islands: irregular 10 mi2 / gage

Gauges Networks

Rain Gauge Networks:


Advantages:

• Perhaps “True” measurement of rain

Disadvantages:

• No coverage over oceans or remote regions

• Point measurement not representative of area

• Wind and instrumental underestimates of rain

• Attempting to collect rain data during hurricane, forexample, can be nearly impossible or unreliable due toextreme winds (even if the equipments survive)

Weather Radars

After J.C. Nam and G.H. Ryu 2001

Weather Radars

RADAR (NEXRAD: Next Generation Weather Radar system )

Weather Surveillance Radar-1988 Doppler (WSR-88D)

Weather Radars

http://weather.noaa.gov/radar/national.html

Los Angeles

Santa Ana

San Diego


Weather Surveillance Radar-1988 Doppler (WSR-88D)

http://radar.weather.gov/radar.php?rid=SOX&product=NCR&overlay=11101111&loop=yes

Weather Radars


http://www.nws.noaa.gov/radar_tab.php

Weather Radars


Radar Coverage at 3km (msl) Radar Coverage at 5km (msl)

Source: www.cimms.ou.edu/~jzhang/radcov.html

Blockage (mountainous region)

Weather Radars

Weather Radars:


Advantages:

• Excellent space and time resolution

• Rainfall estimation in near real-time

Disadvantages:

• Poor coverage over oceans or remote regions

• blockage (mountainous regions

SATELLITE-BASED

PRECIPITATION ESTIMATION

Remote Sensing

Satellites


Types of Satellites:

- Geosynchronous Earth Orbiting (GEO)

- Low Earth Orbiting (LEO)

Satellites

Source: www.comet.ucar.edu Source: www. history.nasa.gov

Geosynchronous Earth Orbiting (GEO) Low Earth Orbiting (LEO)

Satellite Precipitation Data

Meteosat 7 (EUMETSAT)

TRMM PR (NASA/NASDA)

SSMI 85GHz (DMSP)

Geostationary IRCloud top data15-30 minute temporal resolution

Passive Microwave (SSM/I)Some characterisation of rainfall~2 overpasses per day per spacecraft, moving to 3-hour return time (GPM)

TRMM precipitation RADAR3D imaging of rainfall 1-2 days between overpasses( S-35°N-35 °)

PERSIANN Algorithm

Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks (PERSIANN)

The algorithm utilizes a neural network classification and approximation approach to derive precipitation estimates based on IR data calibrated with microwave estimates.

GEO (VIS/IR):- Less accurate estimates- Good global areal coverage with high temporal sampling

LEO (PMW): - More accurate and less frequent estimates- Areal Coverage 3 hour accumulation (Regional gaps)

PERSIANN Algorithm

PERSIANN dataSpatial Resolution: 0.25o degreeTemporal Resolution: 3-hourSource: HyDIShttp://hydis8.eng.uci.edu/hydis-unesco/

PERSIANN-CCS dataSpatial Resolution: 0.04o degreeTemporal Resolution: 1-hourSource: GWADIhttp://hydis.eng.uci.edu/gwadi/

PERSIANN Algorithm

Satellites

http://hydis.eng.uci.edu/gwadi/

Why Satellites?

WSR-88D Radar Coverage Gauge Network

3 km Above Ground Level

Maddox, et. Al., 2002




Why Satellites?

Floods Among the worst Natural Disasters

Hydrologic Forecasting Needs: Flash Floods

Los Angeles (1955)

Application to Flood Forecasting

Hydrologic models are simplified, conceptualrepresentations of a part of the hydrologiccycle. They are primarily used for hydrologicprediction and for understanding hydrologicprocesses. An example of a conceptual modelthat represents a part of the natural system

High Resolution Data from Satellites

Radar Observation (2 km AGL) PERSIANN-CCS Estimates

4km x 4km, 3-hour accumulated precipitation

Study Area

Satellites:


Advantages:

• Global Coverage

• Relatively high resolution in space and time

Disadvantages:

• Still needs research on development of precipitationretrieval algorithms

Satellites

NASANOAA

NRLUC Irvine

Satellites

http://www.bom.gov.au/bmrc/SatRainVal/validation-intercomparison.html

Validation and inter-comparison of daily satellite Precipitation estimates - An IPWG project

Validation of Satellite Retrieval Algorithms

Quantile Probability of Detection (QPOD)

QPOD

Period of Analysis: 2005-2008

Reference data: Stage IV radar-based gauge adjusted data

Quantile False Alarm Ration (QFAR)

QFAR



Monthly Quantile Bias

Monthly Quantile Bias



Bias Adjustment

InputAdjustment Reference

Output

PERSIANN (0.25ox0.25o)

GPCP Monthly (2.5ox2.5o)

PERSIANN-MBA (0.25ox0.25o) - hourly


GPCP Daily(1.0ox1.0o)

PERSIANN-DBA(0.25ox0.25o) - hourly


GPCP Pentad (2.5ox2.5o)

PERSIANN-PBA(0.25ox0.25o) – hourly

PERSIANN-CCS (0.04ox0.04o)

GPCP Daily(1.0ox1.0o)

PERSIANN-CCS-DBA(0.04ox0.04o) – 30 min

Global IR

TRMM, DMSP, NOAA Satellites

ANN

ParameterAdjustment

Sate

llite

Data

High Temporal-Spatial Res.

Cloud Infrared Images

Fe

edback

Sampling

Instantaneous PMW Rain Estimates

PERSIANN Hourly Rainfall (0.25ox0.25o)

Downscaling

Adjusted Hourly Rainfall (0.25ox0.25o)

PMW-RRFill-in

PERSIANN-PMW filled Hourly (0.25o)

BiasAdjustment

Accumulation

PERSIANN Monthly Rainfall (2.5o)

Adjusted Monthly Rainfall (2.5o)

GPCP Monthly Precipitation (2.5ox2.5o)

Bias adjustment of PERSIANN rainfall:

Stage I: Fill-in missing PERSIANN rainfall by PMW-rainfall.

Stage II: Accumulation and Adjustment of PERSIANN rainfall based on GPCP monthly rainfall measurement.

Stage III: Spatial and temporal downscaling of PERSAINN bias estimates from monthly rainfall at 2.5 degree to hourly at 0.25 degree.

Downscaling of GPCP Rainfall to High Spatio-temporal Scale Using PERSIANN

Thank YouFor Your

Attention

Shundasht Fall, Shundasht, Iran

remote sensing of rainfall · advantages, disadvantages and limitations? advantages: • perhaps...

Documents