Policy oriented study on remote sensing agricultural drought monitoring methods

Activity 5.5.

Prof. János Tamás 3rd IDMP CEE Workshop Budapest, 2 & 4 October 2014

Partners

• Hungary (University of Debrecen and GWP HU):

– Applied hydrological remote sensing and GIS;

– Spatial Decision Supporting Systems

• Romania (University of Oradea):

– Geography and Integrated watershed management

• Slovakia (Institute of Hydrology of the Slovak Academy of

Sciences):

– Agricultural water management, Soil hydrology

Objective :

formulate concrete practical agricultural drought monitoring method and

intervention levels with calibrating for the important crops and fruits (wheat, corn

and apple)

Generals about Activity 5.5.

At what stage of the final output(s) are you at the moment?

2. RS tools for

vegetation indices

3. Agricultural

drought decision

support parameters

OUTPUT 1: An

analysis report on the

role of soil and crop

water content status in

waterbalance within

different agricultural,

landuse and water

management practices

at rain fed and irrigated

systems for the most

important crops and fruit

(wheat, corn and apple)

OUTPUT 2: Toolbox with

the concrete identification

of remote sensing and

GIS data tools for

agricultural drought

monitoring and forecast

OUTPUT 3: Report on

integration of RS and

GIS tools and

intervention levels into

drought monitoring

system

June 2013-Dec 2013

Sept 2013 – Jun 2014 May 2014 – Jan 2015

2. RS tools for

vegetation

indices

1. Analysis of

green and

brown water

status

3. Agricultural

drought decision

support

parameters

No changes in sceduling of the case study

What has been done since the 2nd IDMP CEE workshop till now (April 2014 – October 2014)?

Data acquisition and processing

Identification and calibration of drought risk level

Drought risk evaluation and mapping

Objectives of Output 2

• Allow the conversion of different purpose drought indices,

such as meteorological, agricultural and hydrological ones,

and more water-saving agricultural land use alternatives

• To develop a process, which can provide information for estimating relevant drought indexes and drought related agricultural yield losses more effectively from remote sensed spectral data.

• New calculation method, which provides early information on physical implementation of drought risk levels

five major steps were done in order to promote the MODIS NDVI

calibration:

1. Reprojection of MODIS satellite TSA data and masking region of

interest

2. Extraction of MODIS NDVI time series by masks

3. Normalization of extracted NDVI data matrix and yield data

4. Clustering and converting

5. Calculate actual or estimated yield loss

Data acquisition and processing

Data acquisition and processing

a certain crop-county mask

MODIS NDVI images of a certain year

extraction process

extracted MODIS NDVI image for a certain crop in arable lands

ArcGIS model for extraction process of a certain crop-county mask

from MODIS NDVI images

6 Y LONG TIME SERIES of WHEAT NDVI

DROUGHT IMPACT

Modis Terra/Aqua

Ground res. From 250 m

36 band, Cycle: 1 d

Identification and calibration of drought risk level

MODIS NDVI time series dataset, yield data is also available from 2000 – 2012.

Geographical position

Time

Time voxel

N-dimensional datacube

Identification and calibration of drought risk level

IWA 13th International Specialised Conference on Watershed and River Basin Management - San Francisco, CA, USA , 9-12 Sept 2014. 11

05 05 05 05 05

Dry

year

Wet

year

Modis NDVI values for arable land in Békés County, Hungary (June –

September)

Identification and calibration of drought risk level

IWA 13th International Specialised Conference on Watershed and River Basin Management - San Francisco, CA, USA , 9-12 Sept 2014. 12

Normalized yield changes of maize and wheat (2000-2012)

(Source: KSH and INSSE)

Identification and calibration of drought risk level

IWA 13th International Specialised Conference on Watershed and River Basin Management - San Francisco, CA, USA , 9-12 Sept 2014. 13

Yield changes of maize and wheat (2000-

2012) (Source: KSH and INSSE)

Green: optimal

(wet) years

Identification and calibration of drought risk level

IWA 13th International Specialised Conference on Watershed and River Basin Management - San Francisco, CA, USA , 9-12 Sept 2014. 14

Yield changes of maize and wheat (2000-

2012) (Source: KSH and INSSE)

Red: drought

affected

years

Identification and calibration of drought risk level

IWA 13th International Specialised Conference on Watershed and River Basin Management - San Francisco, CA, USA , 9-12 Sept 2014. 15

Yield changes of maize and wheat (2000-2012)

(Source: KSH and INSSE)

Blue: extreme

precipitation

Identification and calibration of drought risk level

IWA 13th International Specialised Conference on Watershed and River Basin Management - San Francisco, CA, USA , 9-12 Sept 2014. 16

Yield changes of maize and wheat (2000-2012)

(Source: KSH and INSSE)

Green: optimal

(wet) years

Red: drought

affected

years

Blue: extreme

precipitation

Reference spectral curves were generated in order to determine the Watch,

Early warning, Warning, Alert and Catastrophe levels of NDVI

Identification and calibration of drought risk level

Watch: When plant water stress is observed in sensitive phenological phases

Early Warning: When relevant plant water stress is observed. The available soil moisture is

close to critical, and it is suggested for farmers to start preparation of intervention.

Predicted potential yield loss is up to 10%.

Warning: When plant stress translates into significant biomass damage, and there is time to

start the intervention actions. Potential yield loss is up to 20%.

Alert: When farmers expect irreversible vegetation damage with real negative profit, and

they have to consider to give up additional cultivation actions in crop production in that actual

vegetation period. Potential yield loss is up to 30%.

Catastrophe: When serious damages and profit loss mitigation is necessary. Potential yield

loss is up to 40%.

Reference spectral curves Drought risk and signalling NDVI levels for maize and wheat

2003 (drought affected year) 2008 (average year)

Risk levels Tisza

catchment (T.c.)

Hungarian part of T.c.

Jász-Nagykun-Szolnok

Hajdú-Bihar

Tisza catchment

(T.c.) Hungarian part of T.c.

Jász-Nagykun-Szolnok

Hajdú-Bihar

Wheat (area %)

Catastrophe 38.44 45.38 59.08 34.80 11.99 13.42 15.73 4.77

Alert 8.25 8.27 7.84 7.70 5.05 5.23 5.40 3.30 Warning 8.29 7.96 6.99 8.03 0.45 4.41 4.62 3.04 Early Warning 10.59 9.75 7.50 10.69 8.30 8.09 8.75 6.13 Watch 7.18 6.42 4.51 7.82 10.14 9.69 10.57 9.06 No yield loss 27.25 22.22 14.07 30.96 64.07 59.15 54.92 73.70

Maize (area %)

Catastrophe 51.77 51.86 66.06 20.92 24.54 19.85 21.24 10.74

Alert 2.94 8.06 2.11 1.80 2.78 1.90 2.15 1.27 Warning 8.93 8.11 6.62 7.03 9.92 6.87 6.92 4.97 Early Warning 8.93 7.83 6.46 8.92 11.90 8.83 9.73 6.77 Watch 8.38 2.51 5.99 10.18 12.46 10.61 11.52 9.54 No yield loss 19.05 21.63 12.76 51.16 38.39 51.94 48.44 66.70

Percentage of affected areas with different drought risks

for wheat and maize (100% is the total area of investigated region)

MAIZE

significant correlation were

found between normalized

NDVI values and maize yield

from the middle of June, to the

end of August, including the

most drought sensitive

blooming period (July)

WHEAT

From May to mid June is

found to be reliable for yield

prediction and forecasting .

Wheat is more drought

tolerable than maize.

Drought risk evaluation and mapping

2008

Our plans for the final period (till 31st of January 2015.): Yield - Profit estimation (2008 normal year/2012 drought year

0100200300400500600

Dry/wet cereal prise %

Maize (T) price index (dry/wet year) Maize(T) price index (dry/wet year)

Catastrophe: When serious damages and

profit loss mitigation is necessary.

Potential yield loss is up to 40% but prise up to 300%

Market can not compensate yield loss,

but give extra large profit on irrigated areas

Source: Stock market price, AKI Wheat

• Agricultural drought characteristics were identified

• Monitoring and drought index conversion method was elaborated based on remote sensing data of Tisza watershed

• Decision support system will be developed assisting farmers in reducing drought risks

Summary

• Activity 2.1: Guideline for Drought Management Plan - MODIS NDVI based

agricultural drought mapping method can be one of the support guideline n

preparation of the Drought maps.

• Activity 5.1: Experimental field research on increasing of soil-water

holding capacity in agriculture and Activity 5.6: Upgrading agricultural

drought monitoring and forecasting: the case of Ukraine and Moldova -

Soil water capacity mapping, described in this Output can support the

identification of soil water holding capacity in regional scale.

• Activity 5.4: Drought Risk Management Scheme: a decision support

system - Yield calibrated drought risk levels can increase the feasibility of

other recently available of decision support system by the visualization and

communication the probability of occurrence of different phases of droughts.

• Activity 6.2: Capacity building trainings - Participation on capacity building

trainings with interpreting how to implement NDVI based yield calibrated

imagery in drought mapping and forecast.

• Activity 7.1: Development of the Compendium of Good Practices - Review

of remote sensing in agricultural drought monitoring and forecast for decision

support system.

Possible links for other activities

• Why fAPAR is not used?

– NDVI is worldwide available for free (http://earthexplorer.usgs.gov/) and

the most studied index on the field of vegetation analysis

• Area specific yield forecast?

– Forecasting of yield(loss) of a ROI (region of interest) sites with local

calibration data. Risk maps shows, how much the yieldloss (t/ha) could

be in the harvest period, if current conditions remains.

• Is the magnitude of potential yield losses applicable to other CEE countries?

– Yes, since NDVI is strongly correlates to biomass, and biomass with

with yield; thus low NDVI means low biomass, resulting low yield.

• CDI and threshold methods are missing

– Accepted, amended

• Two references without figures

– cleared

Referring to PRG comments

Thank you for your attention