Agricultural Drought Monitoring Using Satellite-

Based Products in Romania

Gheorghe Stancalie

, Argentina Nertan

, and Florin Serban

National Meteorological Administration, 97 Sos. Bucuresti-Ploiesti, Bucharest, Romania

Advanced Studies and Research Center, 4, Verii street, Bucharest, Romania

{gheorghe.stancalie, argentina.nertan}@meteoromania.ro, florin.serban@asrc.ro

Keywords: agricultural drought, satellite, vegetation indices.

Abstract: In Romania, the complex agricultural drought is a climatic hazard inducing the worst consequences ever

occurred in agriculture. The paper presents the results of recent studies developed in the National

Meteorological Administration, in the framework of national and European R&D projects, regarding the use

of satellite-derived products for agricultural drought monitoring. In this respect, different vegetation indices,

biophysical parameters and physically-based vegetation state indicators have been used and tested in study

areas over Romania, in order to monitor and assess the drought impact on crops, at different phenological

dates. The main sources of satellite data and related products were provided by TERRA/AQUA-Modis,

SPOT-Vegetation and Landsat TM/ETM+. By examining spatial and temporal patterns of satellite-derived

products and comparing/correlating with the field conditions measured on site, it was determined that the

NDVI, NDWI and NDDI vegetation indices, the leaf area index (LAI) and the fraction of absorbed

photosynthetical active radiation (fAPAR) proved to be good indicators of the vegetation condition and

relevant for the settlement, duration and intensity of the agricultural drought.

1 INTRODUCTION

Among the problems Europe is facing at the

beginning of the third millennium, the reduction of

water resources, their degrading quality and the

occurrence of ever more severe and frequent

droughts are of crucial importance.

In Romania, the complex agricultural drought is

a climatic hazard phenomenon inducing the worst

consequences ever occurred in agriculture.

The most frequent, the agricultural areas in

Romania are affected by drought (7 mil. ha), water

erosion and landslides (6.4 mil. ha), temporary water

excess (4 mil. ha.) and compaction (2.8 mil. ha).

Drought is the limiting factor affecting the widest

surface as regards the crops. The area subjected to

desertification, characterized by an arid, semi-arid or

dry sub humid climate is around 30% of the total

surface of Romania, being mostly situated in the

South-Eastern (Dobrogea), Eastern (Moldavia),

Southern parts of the Romanian Plain and in the

Western Plain. These areas are prevailingly used for

agriculture (around 80% of the total, 60% of which

is arable land) (Romanian Ministry of Agriculture and

Rural Development, 2008).

In the extremely droughty years the drought

phenomenon may engulf the whole Romania's

territory, as it happened more recently in 2000,

2003 and 2007. Large precipitation deficits were

recorded in 1907, 1924, 1928, 1934, 1945, 1946,

1948, 1953, 1982 and 1983, then in 1992 and 1993

and more recently in 2000, 2001, 2003, 2007, 2009

and 2011.

In Romania, the use of remote sensing data

in agriculture is a quickly developing and

promising trend. For a better operative surveillance

of the agricultural areas, starting with 2005, the

Romanian National Meteorological Administration

has implemented a dedicated service based on

satellite-derived products. The satellite

geoinformation products, elaborated by the Remote

Sensing and GIS Lab, are included and analysed in

the weekly Agrometeorological Bulletin issued by

the Agrometeorological Laboratory.

100

Stancalie G., Nertan A. and Serban F.

Agricultural Drought Monitoring Using Satellite - Based Products in Romania.

DOI: 10.5220/0005421901000106

In Proceedings of the Third International Conference on Telecommunications and Remote Sensing (ICTRS 2014), pages 100-106

ISBN: 978-989-758-033-8

2 SATELLITE DATA USED

The main sources of satellite data which have been

used are TERRA/AQUA-Modis, SPOT-Vegetation

and Landsat TM/ETM+. Due to cloudiness, most of

the satellite products are averaged in time, producing

composites based on data from periods of 8-16 days.

The MODIS instrument is one a key instruments

onboard the US satellites of the EOS series (Terra

and Aqua). The bands most applicable for rangeland

studies are bands 1-7 that gather data in the visible

and infrared range at a 250 m and 500 m spatial

resolution.

The SPOT VEGETATION S10 product (ten

days synthesis) with 1 km- resolution is composed

by merging atmospherically corrected segments

(data strips) acquired over a ten days interval. All

the segments of this period (decade) are compared

again pixel by pixel to pick out the 'best' ground

reflectance values.

PROBA-V is a new ESA satellite mission,

launched in May 2013, with the main task of

mapping land cover and vegetation growth across

the Earth every two days. This mission is extending

the data set of the long-established SPOT

Vegetation, but with an improved 350m spatial

resolution.

The LANDSAT TM/ETM+ imagery is a unique

resource for global change research and applications

in agriculture. The main ETM+ image features are: a

panchromatic band with a 15 m-spatial resolution

(band 8); visible (reflected light) bands in the

spectrum of blue, green, red, near-infrared (NIR),

and mid-infrared (MIR) with a 30 m-spatial

resolution (bands 1-5, 7); a thermal infrared channel

with a 60 m-spatial resolution (band 6).

3 AGRICULTURAL DROUGHT

MONITORING USING

SATELLITE-BASED RODUCTS

3.1 The vegetation indices

The vegetation indices (VIs) are among the most

commonly used satellite data products for the

evaluation, monitoring, and measurement of

vegetation cover, condition, biophysical processes,

and change. They have been used for over last

decades in a broad variety of applications, including

monitoring the effects of drought over regional,

national, and even multinational areas (Basso et al.,

2004). The VIs are an important tool for drought

monitoring and evaluation because of the accurate

discrimination of vegetation and correlations with

biophysical parameters which determine the

vegetation state.

The most important VIs for vegetation

monitoring include the "broadband greenness"

category (e.g.: Normalized Difference Vegetation

Index - NDVI, Soil Adjusted Vegetative Index -

SAVI, Enhanced Vegetation Index - EVI, etc) and

the "canopy water content" category (e.g.:

Normalized Difference Water Index - NDWI,

Normalized Difference Drought Index – NDDI, etc)

(Gu et al., 2007; Huete, 1997; Penuelas, 1995).

3.1.1 The Normalized Difference Vegetation

Index (NDVI)

The NDVI is one of the most well known and most

frequently used vegetation indices, being considered

as a measure of the amount and vigour of vegetation.

The combination of its normalized difference

formulation and use of the highest absorption and

reflectance regions of chlorophyll make it robust

over a wide range of conditions (GU et Al., 2007;

Peters et al., 2002). The value of NDVI ranges from -

1 to 1. The common range for green vegetation is

0.2 to 0.8.

The NDVI values have been used in correlations

with various meteorological parameters. For

example, figure 1 shows a good correlation between

the SPOT Vegetation 10 days synthesis NDVI

values and the precipitation, over the study area

situated in the lower basin of the Mures River,

located in the Western part of Romania (Pecica

agricultural area). In this agricultural area the sun

flower and oats crops were identified on the Landsat

ETM+ satellite image (and validated by GPS ground

measurements); the precipitation values were

recorded at Arad weather station, the closest to the

study area. The analysis covers periods from March

to June 2011.

Figure 2 reveals that in the period 6.03 -

6.04.2003, the NDVI values were lower, compared

to the rest, mainly because of the lack of

precipitation in March which have caused a delay of

the vegetation season start. The NDVI time series

analysis is very important for the crop state

monitoring. Such a complex analysis was made

using MODIS/TERRA NDVI products (MOD13A1)

for the following years: 2000 and 2003 (as drought

years), 2005 (as normal year) and 2010 (as rainy

year) and for different vegetation phases. The year

2010, on the other hand, presents greater NDVI

values due to high amount of precipitation. The

Agricultural Drought Monitoring Using Satellite - Based Products in Romania

101

NDVI maps show a rather equal set of values

between the four years, with a slight grow in 2005

and 2010 compared to 2000 and 2003 for the

periods: from the 7

of April to the 8

of May and

from the 9

of May to the 9

of June.

Figure 1: The correlation between NDVI (extracted from

SPOT Vegetation) and precipitation (recorded at Arad

weather station) for the sun flower and oats crops, in the

study area situated in the lower basin of the Mures River,

in the Western part of Romania.

Only during the last vegetation phenological

phase a visible difference occurs between 2000 and

2003 on one hand, and between 2005 and 2010 on

the other hand. The analysis clearly shows the effect

of low precipitation and high temperatures in 2000

and 2003 (very droughty years) over the agricultural

areas.

The NDVI-based analysis for crop state

monitoring was also performed using high-

resolution satellite images, such as LANDSAT

TM/ETM+ data (Jackson, 2007). Figure 3 presents an

example of using LANDSAT TM/ETM+ data for

14.08.2003, 22.08.2006 and 17.08.2010, in the same

study area, located in the Western part of Romania.

The figure 3a shows a “hot-spot” area, associated

with very low NDVI values (pixels in orange and

red), in the central-eastern part of the image acquired

on 14.08.2003 (up left image) and normal NDVI

values in the other 2 images acquired on 22.08.2006

and on 17.08.2010, respectively. In order to isolate

only the parts affected by drought a “low-

vegetation” NDVI threshold was applied to highlight

only two classes (figure 3b).

a) b) c) d)

Figure 2: Spatial variation of average NDVI values: )a – in the period 6.03 - 6.04; b) – in the period 7.04 - 8.05; c) – in the

period 9.05 - 9.06; d) – in the period 10.06 - 28.08.

The threshold value able to separate the dry and

normal conditions was set up using the NDVI

histograms (figure 3c). For this study, an NDVI

value of 0.22 was used as “drought threshold“.

These two classes representation excludes the

“normal” NDVI values while keeping the low ones.

Areas represented in brown in figure 5b can be

therefore associated with dry areas.

Third International Conference on Telecommunications and Remote Sensing

102

3.1.2 The Normalized Difference Water

Index (NDWI)

The NDWI is a satellite-derived index from the

Near-Infrared (NIR) and Short Wave Infrared

(SWIR) reflectance channels (Gao, 1996). The

SWIR reflectance reflects changes in both the

vegetation water content and the spongy mesophyll

structure in vegetation canopies, while the NIR

reflectance is affected by leaf internal structure and

leaf dry matter content but not by water content.

NDWI holds considerable potential for drought

monitoring because the two spectral bands used for

its calculation are responsive to changes in the water

content (SWIR band). This index increases with

vegetation water content or from dry soil to free

water (Chen et al., 2005; Gu et al., 2007). The NDWI

value ranges from –1 to 1. The common range for

green vegetation is –0.1 to 0.4.

Figure 4 presents an example for NDWI maps

over Romania, obtained from MOD09A1 products

(8-day composite) for 2005 – rainy year (figure 4.a)

and for 2007 – droughty year (figure 4.b).

...a) b) c)

Figure 3: a) – The NDVI maps extracted from LANDSAT data; b) – The two classes NDVI maps obtained by applying a

“low-vegetation” NDVI threshold; c) – NDVI histograms.

a) b)

Figure 4: NDWI maps over Romania, obtained from MOD09A1 products (8-day composite): for 03-10.06.2005 (a) and

03-10.06.2007 (b).

Agricultural Drought Monitoring Using Satellite - Based Products in Romania

103

The figure clearly emphasized the large areas

affected by drought in 2007, in the Eastern, South-

eastern and Western agricultural regions of

Romania.

3.1.3 The Normalized Difference Drought

Index (NDDI)

The NDDI is a satellite-derived index defined by the

equation:

NDDI = (NDVI – NDWI) / (NDVI + NDWI)

(1)

The NDDI can offer an appropriate measure of the

dryness of a particular area, because it combines

information on both vegetation and water. The

NDDI has a stronger response to summer drought

conditions than a simple difference between NDVI

and NDWI, and is therefore, a more sensitive

indicator of drought. In case of common range of

values for vegetation moniotoring the NDDI values

vary between 0.33 to 3, a higher range indicating

more severe drought. This index can be an optimal

complement to in-situ based indicators or for other

indicators based on remote sensing data (Gu et al.,

2007).

The figure 5 shows the NDDI over Romania,

obtained from MODIS MOD09A1 products (8-day

composite) for 2005 (rainy year) and 2007 (droughty

year). The figure also highlights the areas affected

by drought in 2007, especially the Eastern, South-

eastern and Southern agricultural regions of

Romania.

a) b)

Figure 5: NDDI maps over Romania, obtained from MOD09A1 products (8-day composite): for 03-10.06.2005 (a) and 03-

10.06.2007 (b).

3.2 The biophysical parameters and

physically-based vegetation state

indicators

3.2.1 The Leaf Area Index (LAI)

The LAI, defined, as half the total leaf area per unit

ground surface area, is a key biophysical canopy

indicator, which play a major role in vegetation

physiological processes and ecosystem functioning.

Assessment of crop LAI and its spatial distribution

are of importance for crop growth monitoring,

vegetation stress, crop forecasting, yield predictions,

management practices, and climate simulations.

Along with the fAPAR, the LAI is a biophysical

variable describing canopy structure and are closely

related to the rate of energy consumed in the

functional processes and exchange mass. Drought

monitoring, corresponding to the state and dynamics

of vegetation, in a given time may be accounting for

LAI values derived from satellite data.

The algorithm for generating the MODIS LAI

products uses surface reflectance (MOD09) and land

cover (MOD12) products. The MODIS LAI

algorithm is based on the analysis of multispectral

and multidirectional surface reflectance signatures of

vegetation elements. The figures 6 (a, b) show the

spatial evolution of average LAI values, as well as

the deviation from the multi-annual average (2000 –

2009) for the years 2000, 2003 (dry years) and

2005, 2010 ( rainy ones), in the study area located in

the Western part of Romania.

4 CONCLUSIONS

Remote sensing techniques could enhance and

improve the crop vegetation state monitoring and the

drought analysis, especially considering the limited

Third International Conference on Telecommunications and Remote Sensing

104

availability of ground measured agrometeorological

data. The use of remote sensing data in

agrometeorology is a quickly developing and

promising trend.

The main sources of satellite data for crop

vegetation state studies and monitoring are the

TERRA/AQUA – MODIS, LANDSAT-TM/ETM+

and SPOT-Vegetation archives. The satellite-derived

vegetation indices data and biophysical parameters

prouved to be good indicators of vegetation

condition and relevant for the installation, duration

and intensity of the agricultural drought.

The MODIS imagery still represents one of the

most important type of satellite data available free of

charge and can be successfully used in determining

the vegetation status at one point or to predict the

changes that may appear in plants activity.

By examining the spatial and temporal patterns

of vegetation indices and comparing/correlating with

the field conditions measured on site, it was

determined that NDVI, NDWI and NDDI, are more

suitable for agricultural drought characteristics

monitoring.

a) b)

Figure 6: a - Spatial variation of average LAI values (from the 6th of March to the 28th of August); b - The LAI deviation

from the multi-annual average (2000 – 2009) (from 6 March to 28 August).

ACKNOWLEDGEMENTS

The presented work was done in the frame of

STAR 2012 (Space Technology and Advanced

Research Program), project DROMOSIS (Drought

monitoring based on space and in-situ data).

REFERENCES

Basso, B., Cammarano1, D., and De Vita, P., 2004.

Remotely sensed vegetation indices: theory and

applications for crop management, Rivista Italiana

di Agrometeorologia , 36-53 (1).

Chen, D., J. Huang, and Jackson, T. J., 2005. Vegetation

water content estimation for corn and soybeans using

spectral indices derived from MODIS near- and

short-wave infrared bands, Remote Sens. Environ.,

98, 225– 236.

Gao, B., 1996. NDWI—A normalized difference water

index for remote sensing of vegetation liquid water

from space, Remote Sens. Environ., vol., 58, 257–

266.

Gu, Y., J. F. Brown, J. P. Verdin, and Wardlow, B.,

2007. A five-year analysis of MODIS NDVI and

NDWI for grassland drought assessment over the

central Great Plains of the United States, Geophys.

Res. Lett., 34, L06407.

Huete, A.R., H. Liu, K. Batchily, and van Leeuwen, W.,

1997. A Comparison of Vegetation Indices Over a

Global Set of TM Images for EOS-MODIS. Remote

Sensing of Environment 59(3):440-451.

Jackson, J. T., D. Chen, M. Cosh, F. Li, M. Anderson, C.

Walthall, P. Doriaswamy, and Hunt, E. R., 2004.

Vegetation water content mapping using Landsat

data derived normalized difference water index for

Agricultural Drought Monitoring Using Satellite - Based Products in Romania

105

corn and soybeans, Remote Sens. Environ., 92, 475–

482 (2004).

Penuelas, J., I. Filella, C. Biel, L. Serrano, and Save, R.,

1995. Reflectance at the 950-970 Region as an

Indicator of Plant Water Status. Intern. Journal of

Remote Sensing 14:1887-1905.

Peters, A. J., E. A. Walter-Shea, J. Lei, A. Vina, M.

Hayes, and Svoboda M. R., 2002. Drought

monitoring with NDVI-based standardized

vegetation index, Photogramm. Eng. Remote Sens.,

68, 71–75.

Third International Conference on Telecommunications and Remote Sensing

106