drought and desertification monitoring zhan yulin institute of remote sensing applications, cas jan....
TRANSCRIPT
Drought and Desertification
Monitoring
Drought and Desertification
Monitoring
Zhan Yulin
Institute of Remote Sensing Applications, CAS
Jan. 23, 2007
Contents
Background
Drought monitoring
Desertification monitoring
Background
UN ESCAP -- the integrated system of water and land
resources for arid areas (2001-2004)Objectives
Enhance the capabilities of water and land management in arid areas by
using space technology
Participating Countries
China, D. P.R. Korea, Mongolia, Kazakstan, Uzbekistan, Pakistan
Scientific issues
Drought Monitoring, Desertification Monitoring …...
Background
European Commission -- A Surveillance System for
Assessing and Monitoring of Desertification (2005-2010)
Objectives
Assessing desertification and land degradation status.
Forecasting of desertification under selected climatic and socio-
economic scenarios.
Monitoring of desertification and land degradation status over
large areas.
Background
European Commission -- A Surveillance System for
Assessing and Monitoring of Desertification (2005-2010)Partners
39 organisations representing 10 EU Member States and 6 Third Country
States
China, Spain, Sweden, France, Germany, Greece, Italy, Netherlands,
Portugal, United Kingdom, Belgium, Chile, Algeria, Morocco, Senegal,
Tunisia
Drought is …
Lack of precipitation
A period of water shortage
Drought is…
WaterDemand
WaterSupply
Methods for drought monitoring
Vegetation Condition Index (VCI)
Crop Water Stress Index (CWSI)
Soil Thermal Inertial Model
Normalized Difference Temperature Index
(NDTI )
Temperature Vegetation Dryness Index
( TVDI) based on the NDVI-Ts Space
NDVI-Ts Space
wet edge Ts_min
dry edge Ts_max = a +b* NDVI
NDVI
Ts
Bare
soil Partial
cover
Full cover
Sandholt et al., 2002
)]*()*[(
)]*([
1122
11
NDVIbaNDVIba
NDVIbaTTVDI s
NDVIbaT
NDVIbaT
s
s
*
*
22max_
11min_
The Sketch map for NDVI-Ts space
Ts
NDVI
Theoretic Base of TVDI
The agricultural climatic regions (central weather bureau of China, 1994) and the distribution of the topsoil moisture observation stations
ACR I: Eastern Monsoon Agricultural Climatic Region ACR II: North-Western Arid Agricultural Climatic RegionACR III: Qingzang tableland Cold Agricultural Climatic Region
Drought -- Study region and dataset
Drought -- Study region and dataset
• NOAA-AVHRR 10 days composite NDVI , thermal infrared (Ch4 and Ch5) dataset from March-May in 2000
• The measured topsoil moisture collected from observation stations
Drought -- Data process
first ten days in March first ten days in April first ten days in May
270
280
290
300
310
320
0. 2 0. 4 0. 6 0. 8
270
280
290
300
310
320
0. 2 0. 4 0. 6 0. 8200
220240
260
280300
320
-0. 1 0 0. 1 0. 2 0. 3
200
220
240
260
280
300
320
-0. 1 0. 1 0. 3 0. 5 0. 7 0. 9200
220
240
260
280
300
320
0 0. 1 0. 2 0. 3 0. 4 0. 5 0. 6
The Extracted maximum and minimum Ts from NDVI-Ts space for small intervals of NDVI for ARC I in 2000
Drought -- Result
Very wet Wet Balanced Dry Severe drought
first ten days in March first ten days in April first ten days in May
Drought -- Result
the first ten days in March
y = -1. 9968x + 0. 8371
R2 = 0. 4616
0. 0
0. 2
0. 4
0. 6
0. 8
1. 0
0 0. 1 0. 2 0. 3 0. 4 0. 5
TVDI
the fi rst ten days i n May
y = -1. 0517x + 0. 7598
R2 = 0. 21310. 0
0. 2
0. 4
0. 6
0. 8
1. 0
0. 0 0. 1 0. 2 0. 3 0. 4
θ
θSevere drought
the fi rst ten days i n Apri l
y = -0. 7681x + 0. 6728
R2 = 0. 2388
0. 0
0. 2
0. 4
0. 6
0. 8
1. 0
0 0. 1 0. 2 0. 3 0. 4 0. 5
θ θ
Relations between TVDI and soil moisture in China
Drought – Conclusion and problem
Compared with VCI, CWSI and NDTI, TVDI is a promising method in monitoring drought for large region
the sensor view angle which can has some effect on Ts and NDVI has not been taken into consider
Although China is divided as three ACR, but it still has varied climate types and topography features.
Higher time-spatial resolution image and a more suitable regional dividing method can produce a more promising result with TVDI.
Desertification is …
Productive Soil turned into
Non- productive Desert
Factors of Desertification
Natural factors
climate
• temperature
• precipitation soil
• type
• erosion vegetation
• species
• biomass
Over mining
Overgrazing
Population increasing
Cutting unregularly
Human factors
Factors of Desertification
Desertification Indexes
Modified Soil Adjusted Vegetation Index (MSAVI)
Albedo
Land Surface Temperature (LST)
Fractional Vegetation Cover (FVC)
Temperature Vegetation Dryness Index (TVDI)
Comparison between Different Indexes
(a) MSAVI
(b)MSAVI+Albedo
(c)MSAVI+Albedo+LST
(d)MSAVI+Albedo+LST+TVDI
NonSlightModerateHeavyVery Heavy
(e)MSAVI+Albedo+LST +TVDI+FVC
NOAA 1995In Kerchin
Accuracy of Different Indexes
Index MSAVI Albedo LST FVC TVDI
Accuracy 68.91% 56.24% 43.56% 66.84% 30.57%
IndexMSAVI+Albedo MSAVI+LST MSAVI+TVDIFVC+Albed
o FVC+LST
Accuracy 86.59% 84.63% 77.65% 85.31% 82.49%
Index MSAVI+Albedo+LST
MSAVI+TVDI+Albedo
MSAVI+LST+TVDI
Albedo+TVDI+FVC
Albedo+LST+ FVC
Accuracy 90.23% 89.18% 86.84% 88.52% 87.67%
Index MSAVI+Albedo+LST+TVDI FVC+Albedo+LST+TVDI
Accuracy 93.83% 92.05%
Index MSAVI+Albedo+LST+TVDI+FVC
Accuracy 95.21%
Desertification climate types
Thornthwaite method
Desertification indicator system
level MSAVI FVC Albedo LST TVDI
Non >2.2 >0.6 <200 <30 <0.35
slight 1.9-2.2 0.3-0.6 200-220 30-36 0.35-0.45
moderate 1.2-1.9 0.23-0.3 220-260 36-40 0.45-0.56
severe 0.8-1.2 0.15-0.23 260-300 40-46 0.56-0.65
Very severe <0.8 <0.15 >300 >46 >0.65
Desertification indicator system in semi arid area for NOAA
>0.72>37>250<0.1<0.60Very severe
0.69-0.7234-37230-2500.1-0.310.60-0.90severe
0.60-0.6932-34205-2300.31-0.430.90-1.20moderate
0.54-0.6029-32180-2050.43-0.61.20-1.50slight
<0.54<29<180>0.6>1.50Non
TVDI陆面温度反照率植被覆盖度MSAVIlevel
>0.72>37>250<0.1<0.60Very severe
0.69-0.7234-37230-2500.1-0.310.60-0.90severe
0.60-0.6932-34205-2300.31-0.430.90-1.20moderate
0.54-0.6029-32180-2050.43-0.61.20-1.50slight
<0.54<29<180>0.6>1.50Non
TVDI陆面温度反照率植被覆盖度MSAVIlevel
Desertification indicator system in semi arid area for MODIS
Desertification distribution in China
Desertification distribution map from NOAA-Avhrr, 1995
Non
Slight Hyper Arid
Moderate
Severe
Very Severe
Water
Desertification distribution in Central Asia
Desertification distribution map from NOAA-Avhrr, 1995
Non
Slight Hyper Arid
Moderate
Severe
Very Severe
Water
Desertification distribution in China
Desertification distribution map from MODIS, 2001
Non
Slight Hyper Arid
Moderate
Severe
Very Severe
Water
Desertification distribution in Central Asia
Desertification distribution map from MODIS, 2001
Non
Slight Hyper Arid
Moderate
Severe
Very Severe
Water
Identifying the boundary of Desert
Occurrence of Onset of green-up
NDVI increases for at least one and a half months;
NDVI reached maximum during the particular periods;
NDVI is greater than 0.05;
This event must occur in April to August.
Identifying the boundary of Desert
AVHRR NDVI 8KmDataset 1982–2000
Study areas including Kazakhstan, Uzbekistan, Turkmenistan, Kirghiz, Tajikistan, Mongolia and western China.
1982 1986
1990 1993
Vegetated areas
Non-vegetated areas
Vegetated areas
Non-vegetated areas
Vegetated areas
Non-vegetated areas
Vegetated areas
Non-vegetated areas
Occurrence of onset of green-up in different years in the study areas
1997
2000
Vegetated areas
Non-vegetated areas
Vegetated areas
Non-vegetated areas
Vegetated areas
Non-vegetated areas
Vegetated areas
Non-vegetated areas
1995
1999
Occurrence of onset of green-up in different years in the study areas
Frequency of Occurrence of Onset of green-up in central Asia and western China (1982-2000)
•the onset of green-up event is never detected in the core desert areas.
•From desert steppe to typical steppe, the Frequency increases from 2 to 18.
•In lower onset frequency zone, the desert steppe (the transitional zone between desert and typical steppe) displayed a typical steppe-like phonological pattern in the wet year and a desert-like pattern in a dry year.
Identifying the boundary of Desert
Dynamic change of desertification
Coefficient of variation (CoV) of the NDVI
AVHRR NDVI 8Km Dataset 1982–2000
Annual pixel-level CoV of NDVI
The slope of the NDVI CoV (1982-2000)
ijijij /cov
NDVI CoV slope map for central Asia and western China(1982-2000).
•Areas without changes (near-zero CoV slope) are mapped in green
•Areas of improving vegetation cover (positive CoV slope) in red
•Areas of declining vegetation cover (negative CoV slope) in blue.
Dynamic change of desertification
Desertification monitoring based on geostatistical texture
0 10km 20km
海南省
北
Desertification in east of Hainan Island result from ilmenites’ over mining.
Desertification in Hainan is difficult to identify directly from origin radiometric bands •Distinguishing between beach sandlot and inner desert •Desertification classification in Hainan different from West-North in china
Hainan Island, china
DATA:TM(1989) and ETM+ (2003)
Desertification monitoring based on geostatistical texture
Method
1893× 3窗口
h=1 窗口移动方向
光谱影像 纹理影像
101
64
52
50
99
98
57
52
54
99
63
53
46
65
83
61
47
51
85
94
54
63
101
96
89
spectral bands geostatistical texture
2)()(2
1xDNhxDNEh
Desertification monitoring based on geostatistical texture
(a) 89年4、3、2合成 (b) 89年纹理波段合成 (d) 03年纹理波段合成(c) 03年4、3、2合成(a) 4、3、2 bands(1989) (b) Texture bands(1989) (c) 4、3、2 bands(2003) (d) Texture bands(2003)
Desertification monitoring based on geostatistical texture
极重度沙漠化
重度沙漠化 其它
北
极重度沙漠化
北
重度荒漠化其它
0 5km 10km中度荒漠化 轻度荒漠化
0 5km 10km
(a) 1989年分类结果 (b) 2003年分类结果(a) The classification result for 1989 (b) The classification result for 2003
Accuracy:92.4% Accuracy:94.7%
Desertification -- Conclusion
An integrated desertification indexes are proposed
Desertification monitoring method based on geostatistical texture
Desert boundary can be identified by using the occurrence of onset of green-up
Utilizing NDVI CoV, the Dynamic change of desertification was analyzed