Sunteți pe pagina 1din 35

Monitoring Drought & Impact from Space

Felix Kogan

NOAA/NESDIS Center for Satellite Applications and Research

CREST 2008

Outline
Drought as Natural Disaster
Data and Background Products Applications

GVI-x Data Set

Drought as Natural Disaster


Drought (D) is a part of earths climate D. occurs every year D. does not recognize borders, political & economic differences D. affects the largest number of people D. unique features
Start unnoticeably Build-up slowly Develop cumulatively Impact cumulative & not immediately observable When damage is evident its too late to mitigate the consequences Drought type: Meteorological, Agricultural, Hydrological, Socioeconomic
3

AVHRR Data for Land Use


Sensor:
Satellites:

Advanced Very High Resolution Radiometer (AVHRR)


NOAA-7, 9, 11, 14, 16, 18 (afternoon.), 17 (morn.),

Data Resolution: Spatial - 4 km (GAC), 8 & 16 km;


Temporal - 7-day composit

Period:
Coverage:
Channels:

1981-2008

World (75 N to 55 S)
VIS (ch1), NIR (ch2), Thermal (ch4, ch5)
4

Typical Vegetation Reflectance


Theory

VIS reflectance depends on CHLOROPHYLL

NIR reflectance depends on WATER CONTENT CELL STRUCTURE

CAROTENOID

AVHRR VIS/NIR bands

NDVI & Reflectance


Theory
Cover Type Vegetation: Dense Ch1 .050 Medium .080 Ch2 NDVI

.150 0.500 .110 0.140

Light
Bare Soil Clouds Water

.100
.269 .227 .022

.120 0.090 .283 0.025


.228 0.002 .013 -0.26
6

NDVI= (Ch2-Ch1)/(Ch2+Ch1)

NDVImax in July

NDVI shows VEGETATION DISTRIBUTION

NDVI & Rainfall (% of mean),

SUDAN

NDVI sensitive to RAINFALL

115% 73% 95%

51%

NDVI in Wet & Dry Ecosystems, CENTRAL USA

* NDVImax indicates HEALTHY vegetation


* The same NDVI indicates HEALTHY vegetation in DRY and UNHEALTHY in WET ecosystems 9

dY, VCI & TCI: 1985-2005

Trend

10

AVHRR-based VH Indices
Vegetation condition index (VCI), values 0 - 100
VCI=(NDVI-NDVImin)/(NDVImax-NDVImin) NDVImax, and NDVImin climatology (1981-2000 maximum and minimum NDVI for a pixel;

MOISTURE

Temperature condition index (TCI), values 0 - 100


TCI=(BTmax-BTmin)/(BTmax-BTmin) NDVImax, and NDVImin climatology (1981-2000 maximum and minimum NDVI for a pixel

THERMAL

Vegetation Health Index (VHI), values 0 100


VHI=a*VCI+(1-a)*TCI indicates extreme stress

VEG. HEALTH
11

NDVI in Wet & Dry Ecosystems, CENTRAL USA

* NDVImax indicates HEALTHY vegetation


* The same NDVI indicates HEALTHY vegetation in DRY and UNHEALTHY in WET ecosystems 12

Drought 1988

Severe Moisture and Thermal Vegetation Stress

13

Percent of USA with rainfall < 50% and VCI < 10

14

Applications: Drought
VH & In Situ Data USA,1988, Week 27

15

Summer

Major US Droughts

Spr/Sum
Sum/Fall Spring Spring Summer
16

Spring

Percent of a state with extreme & exceptional drought

17

Vegetation Health Dynamics

18

7-Year Drought in the Horn of Africa Mar Feb Jan

2006
2005 2004 2003 2002 2001
19 2000

Corn Yield vs CRDs VCI Midwest, USA

Corn Yield normalized for Midwestern CRD

Yield Reduction

20

Yield & Trend Kansas 1980-2005

21

Corn dY vs VHI Correlation KANSAS & CRD20


KANSAS State CRD20

22

dY vs VCI & TCI Correlation Haskell CO, KANSAS, Corn

23

Corn dY vs VHI28
Illinois, Ohio, Mr. Fordham farm (41.55N, 89.47W)

24

dY vs VCI Correlation, KANSAS

25

Independent Model Validation

KANSAS

26

Winter Wheat Cross Validation

27

Yield Predicted (from VHI) vs Observed

RUSSIA, Saratov, SW INDIA, Rajasthan, WW

MOROCCO

ARGENTINA, W

28

2007 Corn, USA


Analogue Method
Major states producing corn: Illinois, Indiana, Iowa, Ohio, Nebraska, Minnesota Figure shows dynamics of Vegetation health (VH): moisture and thermal conditions Analogue of 2007 for US corn was 2003 VH During silk (critical period) VH was inside normal range limit

USA CORN YIELD is expected 1-3% above long-term trend (LTT=9.1 t/ha) or 9.2-9.4

t/ha

KENYA: MALARIA & Vegetation Health


(Kisii District Hospital) Comparison of Malaria Cases & VHI shows (1) Correlation of the number of malaria cases and VH Malaria Cases

(2) In 1998, the peak of malaria shifted to spring similarly to VH


(3) High VH occurs one-two months prior to the peak of malaria (4) VH can be used as malaria

Vegetation Health

Improving

Deterioration

epidemic
predictor

30

ENSO: SST Anomaly in Tropical Pacific (3.4 area)

31

SST anomaly & Vegetation Health El Nino & La Nina December

32

GVI-x Data Set: NDVI

33

Global Long-Term Land Data Sets


GVI Period Resolution: Spatial Temporal Projection Parameters available 19812005 16 km 7-day Lat/Long 7 NDVI C1, 2, 4, 5, SZA, SCA Pathfinder 1981-2000 ISLSCP/ Fasier 1981-2002 GIMMS 1981-2002 GVI-x

GVI-x

1981-2005

8 km 10-day

111 km 30-day

1 NDVI

1 NDVI

8 km 15/16-day Albers equal area 1 NDVI

4 km 7-day Lat/Long 14 NDVI C1, 2, 4, 5 A1,A2,BT4 SZA,SCA, RAZ Pix.Jul. Day Pix. Time, Cloud mask 8 smn, smt VHI,VCI, TCI,Climatol ogy, Drought Fire risk 2-byte (HDF) No NOAA 2005

Products available

Data Precision Validation Producer Year produced

8 smn, smt VHI,VCI TCI,Cli matology Drought Fire risk 1-byte Well done NOAA 1985

1-byte Partial NOAA/NASA 1991

1-byte No Wales Univ. 2003

4-byte (tiff) Partial NASA/Maryl and Univ. 2003

34

References
Salazar, L., F. Kogan and L. Roytman, 2008. Using Vegetation health indices and partial least squuares method for estimation of corn yield. Int. J. Rem. Sens., (Vol 29, No 1-2, 175-189 ) Salazar, L., F. Kogan and L. Roytman, 2007. Use of remote sensing data for estimation of winter wheat yield in the United States. Int. J. Rem. Sens., (Vol 25, No 1, 227-236 ) F. Kogan, 2002: World Droughts in the New Millennium from AVHRR-based Vegetation Health Indices. Eos, Trans. of Amer. Geophys. Union, 83, No 48, 26 November, 557-564. Kogan, F.N., 1997: Global Drought Watch from Space. Bull. Amer. Meteor. Soc. 78, 621-636.

WEB: http://www.orbit.nesdis.noaa.gov/smcd/emb/vci/VH/ind ex.html

35