Remote Sensing based indices for

agricultural drought assessment

Dr. C.S. Murthy

Head, Crop Monitoring & Assessment Division
Remote Sensing Applications Area
National Remote Sensing Centre, Hyderabad
 Understanding Drought and its management
Meteorological drought: reduced rainfall
Hydrological drought: reduced surface water
Agricultural drought: reduced soil moisture
Information needs for
Drought Management
Weather Short term Long term
Management Management
Soil
Different indices
Agricultural • Meteorological
drought  Monitoring & • Soil moisture
Crop • Surface water  Vulnerability
Assessment • Ground water maps
 Prediction & • Crop related  Risk maps
Early warning • Socio-economic
 Prioritization
 Agro-advisories
Management  Crop damage
 Impact
assessment monitoring
• Complex nonlinear interactions
• Slow process with multiple impact
• No single index
• Different states adopt different norms
 National Agricultural Drought Assessment and
Monitoring Systems (NADAMS)
Conceptualisation, development, operational services and institutionalisation of a remote sensing application project

Institutionalisation phase
2012+
Trajectory of NADAMS project- Development and operational services

• New approach for sowing-period drought assessment

2009+ • Geospatial products on soil mositure and rainfall NADAMS Project
• New Indices – Shortwave Angle Slope Index
A success story of
• Use of multiple indices Institutionalization
• IRS AWiFS based sub-district level assessment
2004+ • Decision rules for drought warning & declaration
• Enhanced content & frequency of reporting
• Institutional participation & Capacity building
• Institutional
• Supplementation of WiFS with MODIS Arrangement at Ministry
2002+ • Agricultural area monitoring of Agriculture, GOI
• Increased number of indices
• Creation of MNCFC
• IRS WiFS based district / subdistrict level assessment
1998+ • Participation of user departments
• Transfer of NADAMS
• Operational Services from NRSC project to MNCFC
• NOAA AVHRR April 2012
1989 +
• Regional/district level assessment
• Prevalence, intensity and persistence of agricultural • Enhanced end-use of
1988 • Conceptualization of NADAMS project at NRSC NADAMS project
• Development of Methodology
 Data from multiple satellites/sensors
NATIONAL STATE DISTRICT
NOAA AVHRR Terra Modis IRS-WiFS IRS P6 AWiFS

Wavelength range (µm) Spatial resolution Spatial resolution

0.62 – 0.67 (red) 188 metres 56 metres
0.841 – 0.876 (NIR) Wave lengths 3 bands Wave lengths 4 bands
Spatial Resolution
(green, red and NIR ) (green, red, NIR and MIR)
250m
Swath 2330 kms Swath : 700 kms. Swath : 700 kms.

Wavelength range (µm)

0.58 – 0.68 (red)
0.725 – 1.1 (NIR)
3.55 – 3.93 (MIR)
10.3 – 11.3 (TIR)
11.5 - 12.5 (TIR)
Spatial Resolution
1.1 Km
 Spectral response of vegetation
Leaf Cell Dominant factor
pigments structure Water content controlling leaf reflectance

80

70 Chlorophyll Water absorption

R absorption
E
F 60
L
E
C 50
T
A
N 40
C
E 30
(%)

20

10

0
0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6

WAVELENGTH (um)
Normalized Difference Vegetation Index Spectral response of vegetation
Red – more absorption due to chlorophyll
Near Infra red – more reflection due to leaf
structure
NDVI Normalized Difference Vegetation
Index (NDVI)
NIR – Red / NIR+Red
Ratio of difference and sum of surface reflectance
in NIR and red spectral bands Reflected radiation in Near infrared and red
bands.
NDVI ranges from -1 to +1
NIR can see roughly 8 layers Water – negative NDVI
Red – one layer Clouds – zero NDVI
Vegetation – positive NDVI represents density,
vigor
Most successful indicator for describing vegetation

Normalisation - reduces the effect of sensor

degradation
sensitive to changes in vegetation
easy to compute and interpret
available from most of the sensor systems
 NDVI

0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9

0
1
01-May-11
09-May-11
17-May-11
25-May-11
02-Jun-11
10-Jun-11
18-Jun-11
26-Jun-11
04-Jul-11
12-Jul-11
20-Jul-11
28-Jul-11
05-Aug-11
13-Aug-11
21-Aug-11
29-Aug-11
06-Sep-11
14-Sep-11
22-Sep-11
30-Sep-11
08-Oct-11
16-Oct-11
24-Oct-11
01-Nov-11
09-Nov-11
17-Nov-11
25-Nov-11
03-Dec-11
11-Dec-11
19-Dec-11
27-Dec-11
04-Jan-12
12-Jan-12
20-Jan-12
28-Jan-12
05-Feb-12
13-Feb-12
21-Feb-12
29-Feb-12
08-Mar-12
16-Mar-12
24-Mar-12
NDVI profile over agricultural area

01-Apr-12
09-Apr-12
17-Apr-12
25-Apr-12
03-May-12
11-May-12
19-May-12
 Some important VI data sets
Moderate resolution data
Spatial Temporal
Satellites Sensor resolution resolution Swath
Resourcesat 1 & 2 AWiFS 56 m 5 days 750 km
LISS III 23 m 26 days 140 km
LISS IV 6m 48 days 70 km
LANDSAT 8 OLI 30 m 16 days 185 km

Coarse resolution data

Satellite/ Sensor Indices Relevant Parameter

NOAA AVHRR (1km) NDVI Crop condition
Oceansat 2- OCM (360m) NDVI, ARVI Crop condition
Terra MODIS (500 m) SASI, NDWI Surface wetness/ sown area
discrimination
Terra AMSRE (25 km) Soil moisture Surface wetness/ sown area
discrimination
INSAT 3A CCD (1 km) NDVI Crop condition
 NDVI based agricultural drought assessment

Seasonal NDVI profiles for drought assessment

0.6
Integration with ground data
0.5
0.4
NDVI

0.3
0.2 .
Weekly deviations of rainfall
0.1 300
250
0
200
June July Aug. Sept. Oct. Nov.

% deviation
150
Month 100
50

Normal delayed season Drought 0

-50 .

-100
(1) relative deviation from normal, 12/6 19/6 6/6 3/710/7 17/7 24/7 31/7 7/8 14/8 21/8 28/8 4/9 11/9 18/9 25/9

(2) vegetation Condition Index,

.
(3) in season rate of transformation Weekly progression of sown areas
100
90
80
70

% of normal
60
50
40
30
20
10
0

5 Jun
12 Jun
19 Jun
26 Jun
3 Jul
10 Jul
17 Jul
24 Jul
31 Jul
7 Aug
14 Aug
21 Aug
31 Aug
11 Sep
18 Sep
25 Sep
30 Sep
 Normalized Difference Wetness Index/ NDWI/LSWI
Land Surface Wetness Index
Reflectance in 0.9 – 2.5 microns dominated by liquid water absorption

Sensitive to surface wetness/ vegetation moisture

Less affected by atmosphere

Reflectance 1.5-2.5 microns does not saturate till LAI reaches 8

SWIR availability – AWiFS, LISS-III, MODIS, INSAT 3A

MODIS – 3 SWIR bands – 1240 nm, 1640 nm, 2100 nm

LSWI – Land Surface Wetness Index – uses SWIR at 2100 nm – very sensitive to
wetness

NDWI/LSWI uses
Agriculture – crop stress detection, crop yield, classification of succulent crops,
surface moisture

Forest – Monitoring, burnt area detection

 Commonly used indices for drought assessment
Most commonly adopted index – NDVI
a) chlorophyll based index
b) plant vigour and density
c) easy to compute and interpret
d) robust index
e) Limitations – soil back ground,
saturation, time lag etc.

Recently popularized index – NDWI/LSWI

a) Plant moisture based index
b) NIR and SWIR based
c) No saturation issues
d) Immediate response
e) Sensitive to surface wetness during
sowing period

Combination of NDVI and NDWI

a) Overcomes limitations of either one
b) amplifies anomalies and
c) more responsive to ground situation
 AWiFS NDVI - Odisha – June to November 2016

Jun 01-15, 2016 Jun 16-30, 2016 Jul 01-15, 2016 Jul 16-31, 2016

Aug 01-15, 2016 Aug 16-31, 2016 Sep 01-15, 2016 Sep 16-30, 2016

Oct 01-15, 2016 Oct 16-30, 2016 Nov 01-15, 2016 Nov 16-30, 2016

Increasing NDVI (greenness) Non-crop area

Interpretation of NDVI changes
to assess Agricultural drought
NDVI anomaly
% dev. from normal

(actual NDVI-normal
NDVI)/normal NDVI*100
Selection of normal year –
average of recent past normal
years

NDVI is a conservative indicator

and hence anomalies are not
very high

Thumb rule:
> 20% reduction in NDVI – drought
conditions

>30% reduction indicate moderate

to severe drought conditions
Interpretation of NDVI changes
to assess Agricultural drought
Vegetation Condition Index (VCI)
• Derivative of NDVI
• Substitute to NDVI deviation

VCI = NDVI- NDVImin /(NDVImax –

NDVImin) *100

Range 0-100%

0-40 % drought
40-60 mild to moderate drought
60-100 good
 Vegetation Condition Index (VCI)
Computation VCI = NDVI- NDVImin /(NDVImax – NDVImin)
*100
September
year NDVI VCI
2000 0.5 8
2001 0.55 50
2002 0.6 92
2004 0.52 25
2005 0.55 50
2006 0.51 17
2007 0.49 0
2008 0.59 83
2009 0.61 100 Some critical issues
2010 0.54 42
2011 0.51 17 Time series data base
(at least 10-12 years)
Mean 0.54
Min 0.49 Differences due to cropping
Max 0.61 pattern, crop calendar to be
diff 0.12 normalised
 Interpretation of NDVI changes for drought assesment

Reduction in NDVI is caused by;

• Crop moisture stress

• Flooding/excess rainfall

• Crop and crop stage differences

between the two years under
comparison
Limitations of NDVI

Chlorophyll based index – saturates with LAI

(=3)
Limited capability to detect vegetation water
content
Over-estimation when the veg. density is less
Saturation at peak vegetative phases
Conservative index
Time lag
Despite these limitations, NDVI is still a successful indicator for agricultural drought assessment
 Block level crop condition – Anantpur district
Comparison between normal year (2005) and drought year (2002)
 2002 drought conditions
at mandal level in Anantpur district
Mudigubba mandal - Anantpur district
Tadipatri mandal - Anantpur (Groundnut)
groundnut

0.350
0.450
0.400 0.300
0.350 0.250
0.300
0.200
0.250
0.200 0.150

NDVI
NDVI

0.150 0.100
0.100
0.050
0.050
0.000 0.000
June July August Sep October June July August Sep October
-0.050 -0.050
-0.100
-0.100
2005 2002
2005 2002

Kanekal mandal - Anantpur district

Pamidi mandal - Anantpur (Paddy)
groundnut + paddy

0.450
0.400
0.400
0.350
0.350 0.300
0.300 0.250
0.250 0.200
NDVI
NDVI

0.200 0.150
0.150 0.100

0.100 0.050
0.000
0.050
-0.050 June July August Sep October
0.000
June July August Sep October -0.100
-0.050
2005 2002
2005 2002
 Crop area affected by drought in kharif 2015, West Bengal
Purulia district Bankura district West Midnapur district

Normal Normal Normal

Mild Mild Mild
Moderate Moderate Moderate
Severe Severe Severe

50 47 50
39

Agriltural area Affected (%)

50
Agriltural area Affected

40 40 31
Agriltural area Affected

40 29 40 30 23
30 26 20
17 30
20 6
(%)

8 10 1
(%)

5 0
10 1 20 15 0
0 10
10
< 10 % 10-20 20-30 30-40 40-50 < 50 % 2 0
% % % % 0
% of reduction in crop condition < 10 % < 50 % % of reduction in crop condition
% of reduction in crop condition
 AWiFS derived crop condition anomalies showing agricultural
drought situation in Andhra Pradesh, kharif 2011
NDVI anomaly NDWI anomaly
Sep 2011 Sep 2010 Oct 2011
Sep 2011

Assessment at
sub-district level
NDVI - Roddam mandal, Anantpur NDWI - Roddam mandal, Anantpur

0.35
0.50
0.30
0.40 0.25
0.30
NDWI

0.20
NDVI

0.20 0.15
0.10 0.10
0.05
0.00
0.00
FN1-SEP FN2-SEP FN1-OCT FN2-OCT Nov
F1SEP F2SEP F1OCT F2OCT F1NOV
2010 2011
2011 2010

394 mandals
164 mandals
562 mandals
 Kharif 2010 – Agricultural drought assessment, Bihar
AWiFS NDVI AWiFS NDVI
Oct. 2010 (Drought year) Oct. 2008 (Normal year)

Satellite derived Area Favourable for Crop

0 Sowing/Crop Sown Area (AFCS) , Lakh ha.
Bhagalp…
Jahana…

Aurang…
Samasti…
% deviation from

Jamui

Nalanda
Buxar

Patna
Gaya

Bhojpur

Kharif Crop areas affected by agricultural

-20 Nromal AFCS AFCS Unfavourable drought situation are showing lower NDVI
normal

Area Jul-10 Aug-10 area compared to normal, in kharif 2010 in

37 19 23 14 Bihar state.
-40
Agricultural drought assessment
(based on multiple indices; NDVI, NDWI, SASI) 1. Pas. Champaran 20. Khagaria
-60 Rainfall deficiency 2. Pur. Champaran 21. Katihar
3. Sheohar 22. Bhagalpur
4. Sitamarhi 23. Banka
5. Madhubani 24. Munger
6. Supal 25. Lackeesarai
7. Ararai 26. Sheikhpura
100 8. Kishanganj 27. Nalanda
% of normal

Crop sown area status 9. Purnia 28. Patna

50 10. Madhepura 29. Bhojpur
11. Saharsa 30. Buxar
12. Darbhanga 31. Bhabua
0 13. Muzaffarpur 32. Rhotas
Jahana…

Aurang…

Bhagal…
Samasti…
Jamui

Nalanda
Patna
Buxar
Gaya

Bhojpur

14. Gopalganj 33. Aurangabad

15. Siwan 34. Jahanabad
16. Saran 35. Gaya
17.Vaishali 36. Nawada
Moderate (103) Moderate (103)
18. Samastipur 37. Jamui
Severe (55) Severe (55)
Satellite based agricultural
drought assessment
kharif 2010, West Bengal
 Crop condition assessment
Rajkot district, Gujarat Kharif 2016

September October November

Deviation
+ 10%
10-20%
20-30%
30-40%
40-50%
< 50%
 NDVI Anomaly – September 2016
Kerala
2016
Drought

Normal
Normal
Mild drought
Mild – moderate
Moderate to severe
PERCENTAGE OF AREA in different drought classes - SEPTEMBER

S.No Districts C1 C2 C3 C4 C5

1 Alappuzha 19.57 0.04 0.10 0.05 80.23

2
Ernakulam 36.42 0.16 0.10 0.01 63.31
3
Idukki 17.23 0.34 0.17 0.00 82.25
4
Kannur 86.78 0.07 0.05 0.01 13.09
5 Kasaragod
41.25 0.22 0.14 0.00 58.38
6
Kollam 55.96 0.05 0.10 0.01 43.87
7 Kottayam
16.33 0.04 0.06 0.03 83.54
8 Kozhikode
72.66 0.16 0.07 0.00 27.11
9 Malappuram
87.52 0.29 0.08 0.00 12.11
10 Palakkad
63.08 0.19 0.12 0.02 36.59
11 Pattanamthitta
47.75 0.12 0.07 0.01 52.06
12 Thiruvananthapuram
37.84 0.20 0.04 0.01 61.91
13 Thrissur
59.60 0.17 0.15 0.01 40.07
14 Wayanad
76.12 0.07 0.05 0.00 23.75
 Shortwave Angle Slope Index (SASI)
Features SASI value Features SASI value
Dry soil highly positive Dry vegetation low negative
Wet soil low positive Moist veg. high negative

Response of SASI to crop sown area

0.20 500
0.15

Sown area 000’ ha

400
0.10
0.05 300

SASI
0.00

09 Jun

25 Jun

11 Jul

27 Jul

12 Aug

28 Aug

13 Sep

29 Sep
-0.05 200
-0.10
100
-0.15
-0.20 0
Weeks

βSWIR1 = cos-1 [ (a2 + b2 - c2) / SASI Sown area

(2*a*b) ]
Slope = (SWIR2 − NIR)
SASI = βSWIR1 * Slope (radians)
where a, b and c are Euclidian
distances between vertices NIR
and SWIR1, SWIR1 and SWIR2,
and NIR and SWIR2, respectively
Chronological synchronization between
(a) Decrease in SASI
(b) Increase in rainfall
(c) Increase in sown area
NADAMS project
Conceptually and computationally simple
procedures to discriminate the crop sowing
favorable areas at state level
 Seasonal dynamics of SASI

0.20 Before crop Commencemen

0.15 sowing t of crop sowing
0.10
0.05
SASI values

0.00
-0.05
-0.10
-0.15
-0.20
-0.25 Crop growing and
maturity
-0.30
11 July

19 July

27 July
09 June

17 June

25 June

05 Sep
13 Sep

21 Sep

29 Sep
3 July

04 Aug

12 Aug

20 Aug
28 Aug
2002 2006 2009
 SASI

-0.8
-0.6
-0.4
-0.2
0

-1
0.2
0.4
01-May-11
09-May-11
17-May-11
25-May-11
02-Jun-11
10-Jun-11
18-Jun-11
26-Jun-11
04-Jul-11
12-Jul-11
20-Jul-11
28-Jul-11
05-Aug-11
13-Aug-11
21-Aug-11
29-Aug-11
06-Sep-11
14-Sep-11
22-Sep-11
30-Sep-11
08-Oct-11
16-Oct-11
24-Oct-11
01-Nov-11
09-Nov-11
17-Nov-11
25-Nov-11
03-Dec-11
11-Dec-11
19-Dec-11
27-Dec-11
04-Jan-12
12-Jan-12
Seasonal SASI profile

20-Jan-12
28-Jan-12
05-Feb-12
13-Feb-12
21-Feb-12
29-Feb-12
08-Mar-12
16-Mar-12
24-Mar-12
01-Apr-12
09-Apr-12
17-Apr-12
25-Apr-12
03-May-12
11-May-12
19-May-12
 Geospatial product on Area Favourable for
Crop Sowing (AFCS) using multi-criteria approach

Input datasets
Modelled Soil
SASI soil moisture Texture

Kharif area Rice area

mask mask

Area Favourable for

Crop Sowing/Crop
Sowan Area (AFCS)
Area Favourable for Crop Sowing (AFCS) derived from SASI and
water balance methodology, Kharif 2012

June July
State Kharif AFCS M ha. Unfavorable
normal area
state June July Aug Sep
Andhra Pradesh 7.8 2.0 6.8 6.9 7.3 0.4
Bihar 3.7 0.7 3.6 3.7 3.7 0
Chhattisgarh 4.8 3.2 4.8 4.8 4.8 0
Gujarat 8.7 1.3 5.0 5.8 8.1 0.6
Haryana 2.8 0.6 1.6 2.8 2.8 0
Jharkhand 2.5 0.3 2.4 2.5 2.5 0
Karnataka 7.5 3.5 6.0 6.0 7.0 0.5

Madhya Pradesh 10.4 0.7 9.7 10.3 10.4 0

August September Maharashtra 14.0 5.5 13.2 13.8 13.8 0.2
Odisha 6.3 3.9 6.1 6.2 6.3 0
Rajasthan 14.3 0.2 4.4 11.7 13.6 0.8
Tamil nadu 2.4 1.1 1.8 2.0 2.0 0.4
Uttar Pradesh 9.3 2.8 8.7 9.2 9.3 0
Sub-Total 94.5 25.8 74.2 85.7 91.7 2.9
All India 108.6 34.2 87.0 97.7 105.5 3.1
 Tracking the early season drought conditions of
2014 using LPRM Soil Moisture datasets
Soil moisture deviations from normal in 2014
4_11 JUNE 12_18 JUNE 19_25 JUNE 26_2 JULY 3_9 JULY 10_16 JULY

17_23 JULY 24_30 JULY 17_23 JULY 31_06 AUGUST

0.40 Drought frequency in the

Soil moisture (m3/m3)

0.35
0.30 sowing period
0.25
0.20
0.15
0.10
0.05
0.00
05 Jun

23 Jul
03 Jul
07 Jul
11 Jul
15 Jul
19 Jul

27 Jul
31 Jul
09 Jun
13 Jun
17 Jun
21 Jun
25 Jun
29 Jun
01 jun

04 Aug
08 Aug
12 Aug

LPRM_SM 2014 LPRM_SM 2013

 Mobile app. from NRSC/ISRO
that allows users to share, Field Data Collection using Geo-ICT
access and upload natural Observation Transmission Information Decision Action
resources information on a near
real time basis, with Bhuvan
serving as the platform

 Crowd sourcing approach with

open source tools like Open
layers, PHP, Geoserver and
Mapserver, etc. for
visualization and uploading

 Immense use for agricultural

information collection/analysis

 Provision to upload the

information through internet or
customized mobile which will
be geo-tagged for
visualization through Bhuvan
Successful and On-going applications of FDC
Portal • Crop mapping
• Pest/disease surveillance
 Geo-tagged in-season field • Crop Insurance
data enables developing a • Crop damage assessment enumerations
repository of agriculture/crop • Disaster
related data • Drought impact enumeration.
 Conclusions

 Satellite indices have the potential to capture drought conditions

 Satellite data – free access, analysis/computations simplified

 Strengthen the drought monitoring and declaration systems

 Selection of suitable satellite data - AWiFS is preferable

 States – maintain NDVI, LSWI data, effective utilisation

 State RS Centres - capability and expertise, support to State Depts.

 National Centres – provide expertise, training etc.

 Need for institutional mechanism

Thank you