agriculture drought with remote sensing
DESCRIPTION
Presented in “China – Thailand Geo-Informatics Workshop Series I: Agricultural Applications” 30 June - 2 July 2010 at Pattaya, ThailandTRANSCRIPT
Agricultural Drought Monitoring with Agricultural Drought Monitoring with Remote SensingRemote Sensing
Dr Yan NanaInstitute of Remote Sensing Applications (IRSA)
Chinese Academy of Sciences (CAS)
Introduction
Drought is the most serious disaster damaging crop
Establish drought monitoring system at Ministry of water resources
The system consists of three components– Meteorological Model (6 indices)
(1996-– Hydrological model (2 indices)
(2005-– Remote sensing (4 indices) (2005-
Methods
VisibleVisible and Near infraredThermal infraredMicrowave
Albedo Vegetation Index Surface temperature Backscatter coefficient
Indices– Visible and Near Infrared : NDVI, VCI, BMVCI– Shortwave Infrared : NDWI– Thermal Infrared: CWSI , TCI , DSI– Visible 、 Near Infrared and Thermal
Infrared : Ts/NDVI , TVDI , VTCI– Microwave remote sensing: soil moisture
……
DroughtWatch System
Remote sensing data :AVHRR/MODISIndices: VCI\TCI\NDWI \VHI
Preprocessing– Geometric correction– Atmosphere correction– BRDF correction
Surface parameters calculation– Cloud removal, NDVI, surface
temperature, Albedo
NDVIj: NDVI of date j ; NDVImax: the maximum NDVI of all dataset ; NDVImin :the minimum NDVI of all dataset ;
%100II
I
mm
m
inax
inj
j NDVNDV
NDVNDVIVCI
The maximum and minimum NDVI calculation May,2003
VCI (Vegetation Condition Index)
Daily NDVI dataset (1991-2004)
Cloud removal
Crop NDVI range
The maximum and minimum NDVI dataset
Cloud channel
growth seasonNDVI (0.15~0.8)
NDVImax and NDVImin
Tmax and Tmin
TCI, Early May 2003
Tsj surface temperature of date j ; Tmax the maximum temperature of all dataset ; Tmin the minimum temperature of all dataset ;
TCI (Temperature Condition Index)
Tmax and Tmin
Cloud ID
The range of T0 based on the drought district
DEM
Ts Dataset(1991-2005)
Cloud removl
T thresh
Tmax and Tmin dataset
DEM correction
NDWI dataset
Cloud removal
May,2003
NDWI (Normalized Difference Water Index)
VHI= aVCI+bTCI –VHI: considering vegetation condition and the surface
temperature change simultaneously;
–a,b are the weight coefficient of indices.
a=R2vci/( R2
vci+ R2tci) ;b=R2tci/( R2
vci+ R2tci)
R2vci : R2 between VCI and relative soil moisture
VHI
Soil Water Holding Capacity
0-10cm田间持水量图 10-20cm田间持水量图
Station Map and soil type map 178 stations
Study area: Shanxi Data
Soil moisture data: 2002-
2005, seventy stations
Indices: 2002 -2005,
seventy stationsShanxi
The relation between VCI, TCI and relative soil moisture: relative soil moisturerelative soil moisture == soil water content/field capacitysoil water content/field capacity
Soil Moisture inversion
Soil Moisture
Daily TCI and Soil MoistureTCI - 10CM相关性 y = 0. 059x + 12. 712
R2 = 0. 5865
05
1015202530
0 50 100 150 200
TCI - 20CM相关性 y = 0. 0737x + 12. 832R2 = 0. 4832
05
1015202530
0 50 100 150 200
TCI - 10CM旬 相关性 y = 0. 0432x + 12. 853
R2 = 0. 513
05
1015202530
0 50 100 150 200 250
TCI - 20CM旬 相关性 y = 0. 0451x + 14. 193
R2 = 0. 3483
05
1015202530
0 50 100 150 200 250
10 days TCI and soil Moisture
Soil Moisture
VCI - 10CM相关性 y = 0. 0457x + 12. 693R2 = 0. 4536
05
1015202530
0 50 100 150 200
VCI - 20CM相关性 y = 0. 0513x + 13. 556R2 = 0. 2694
05
1015202530
0 50 100 150 200
VCI - 10CM旬 相关性 y = 0. 0573x + 12. 859R2 = 0. 5072
05
1015202530
0 50 100 150 200
VCI - 20CM旬 相关性 y = 0. 0336x + 15. 656R2 = 0. 1598
05
1015202530
0 50 100 150 200
Daily VCI and Soil Moisture
10 days VCI and soil Moisture
Soil Moisture
VHI - 10cm太谷实验区天 湿度相关性分析y = 0. 2439x + 50. 678
R2 = 0. 60
0
20
40
60
80
100
120
0 50 100 150 200
TCI
湿度
VHI - 20cm太谷实验区天 湿度相关性分析y = 0. 2594x + 49. 101
R2 = 0. 42
0
20
40
60
80
100
120
0 50 100 150 200
TCI
湿度
VHI - 10cm太谷实验区旬 湿度相关性分析y = 0. 2406x + 50. 084
R2 = 0. 56
0
20
40
60
80
100
120
0 50 100 150 200
TCI
湿度
VHI - 20cm太谷实验区旬 湿度相关性分析y = 0. 1863x + 55. 461
R2 = 0. 29
0
20
40
60
80
100
120
0 50 100 150 200
TCI
湿度
Daily VHI and Soil Moisture
10 days VHI and soil moisture
VHI= a TCI + b VCI
Soil Moisture
ndwi - 10CM相关性 y = - 0. 0382x + 19. 5
R2 = 0. 2137
05
101520
2530
0 50 100 150 200
ndwi - 20CM相关性 y = - 0. 0498x + 21. 441
R2 = 0. 1667
05
1015202530
0 50 100 150 200
Daily NDWI and Soil Moisture
Comparison results
The relation between indices and soil moisture VHI > TCI >VCI>NDWI R2 between VCI and soil moisture at 10cm and 20cm respectively is 0.45,0.26 R2 between TCI and soil moisture at 10cm and 20cm respectively is 0.58 , 0.48 ; R2 between VHI and soil moisture at 10cm and 20cm respectively is 0.6 , 0.42;R2 between NDWI and soil moisture at 10cm and 20cm respectively is 0.21 , 0.16;
Region Suitability
Drought zone
geomorpho+Soil
Cluster
Climate
Cluster
Overlay
全国旱情区划基本单元制作流程图
Soil Moisture Ground Measurement Points
DroughtIndex Suitability
不同旱情区划单元指数适应性表 : T 表示 TCI , V 表示 VCI , H 表示 VHI , \ 后的指数为次最佳指数
Validation of drought monitoring resultsValidation area: Taigu,Shanxi
Data required:
– Indices,2003~2005, day and ten-days– Day soil moisture, 2003-2005, two depth of 10CM and 20CM– Ten-days soil moisture,2003-2005
5km
Sample allocation Observation point
100m
40m
40m
10m
10m
40m 40m10m 10m
Depth TCI_R2 VCI_R2 NDWI_R2 VHI_R2
10cm Soil moisture0.93 0.66 0.08 0.97
20cm Soil moisture0.91 0.60 0.15 0.92
Validation of drought monitoring results
Pre-processingDroughtIndexDroughtGrade Statistics
DroughtWatchUpgrading: AVHRR to MODIS MODIS to MERSI
Operational Products
Percent of different drought grads different provinces of China (April, 2006)
Because consecutive days without rain and snow reached more than 80 days in the north plain, the area percentage of wheat drought between January 20 and February 16 were 49.5%, 36.9% and 30.5%, respectively.
Since Feb 16, the drought area had a large decrease because of precipitation and irrigation in large area of the north plain
Drought monitoring in North Plain In early spring, 2009
January 21-31 February 1st-10
February 11-16 February17-22
In august, drought area expanded rapidly in the north-west of Liaoning and the west of Jilin, and mostly are minor drought
The drought area percentage are 20.9 % and 15.2 in Liaoning and Jilin respectively
Drought monitoring in Dongbei from July to August,2009
On-going research
Incorporate HJ data, CCD to increase resolution for better monitoring
Microwave data for soil moisture
Develop new indices: combination of optical and microwave data
Scale issues
Soil moisture inversion based on active microwave radiometer
Soil moisture in Yucheng based on Radarsat2 , August 28,2008
)())sec(2exp()))sec(2exp(1(cos0svvV dMcbMbMMa =
Water Cloud Model
Mv is vegetation watercontent , Ms is soil water content , θ is incidence angle, a,b,c,d are coefficients
Fast drought monitoring in region scaleDrought monitoring in local scale
HJ-1 (CCD, IRS)
Vegetation IndexSurface Temperature
Max/Min VIMax/Min Ts
AVHRR/MODIS/FY3(vi sible,near-
infrared,thermal red)
Drought moinitoring in region scale(1km)
Drought moinitoring in local scale
(30m-100m)
Vegetation IndexSurface Temperature
Soil moisture data
cal ibrationval idation
Serial data set
Drought Indices(VCI/TCI/EF)
ETWatch
Drought model
Conclusion
RS is only way to monitor Drought Need ground data to calibrate the model High Resolution for disaster reduction Soil Moisture is not a good indicator for all
crop season, new indicator should be found
Thanks for Attention !