Issuu on Google+

International Journal of Agricultural Science and Research (IJASR) ISSN 2250-0057 Vol. 3, Issue 1, Jun 2013, 1-28 © TJPRC Pvt. Ltd.

GEOSPATIAL ASSESSMENT OF AGRICULTURAL DROUGHT (A CASE STUDY OF BANKURA DISTRICT, WEST BENGAL) SUMANTA DAS, MALINI ROY CHOUDHURY & SACHIKANTA NANDA Department of Civil Engineering, Faculty of Engineering and Technology, SRM University, SRM Nagar, Kattankulathur, Tamilnadu, India

ABSTRACT Geospatial techniques have played a key role in studying different types of hazards either natural or manmade. Temporal satellite data of three years 2000, 2005 and 2010 are used to monitor and assess the drought severity and the impact of agricultural drought on crop production. In Bankura District, agricultural drought and crop failure have been common and spatial variability of rainfall in cropping season with frequent and longer dry spells. This makes them vulnerable to the risk of agricultural drought. This study is conducting with the objective of assessing agricultural drought risk and it’s impacts on yield reduction using RS and GIS techniques. The digital indices using satellite data namely, NDVI (Normalized difference Vegetation index) and NDVI anomaly can be prepared from the long term mean values of maximum NDVI to assess the severity of drought. VCI(Vegetation condition Index), TCI(Temperature condition Index), SPI(Standardized Precipitation Index), MSI(Moisture Stress Index) and YVI(Yellowness vegetation Index) is very essential to Assess the agricultural drought. The impact of agricultural drought on crop production was measured through estimation of yield reduction(in %). Compared to other cropping seasons of the analysis period, yield reduction for the year 2005 was lesser than the highly drought years 2000 and 2010. Simple regression analyses has performed between Land surface temperature with NDVI, TCI with VCI and SPI,MSI, NDVI Anomaly, VCI with Yield reduction(%). Finally, a resultant drought vulnerability map was obtained by integrating NDVI Anomaly, MSI, SPI, VCI, TCI and YVI which indicates the area facing a combined drought. The combined vulnerability map shows that 6% area has no risk, 53 % area face moderate risk and 41 % area face high risk within the entire geographical area. Thus, this agricultural drought risk mapping can be useful to guide decision making process in drought monitoring and to reduce the risk of drought on agricultural production and productivity.

KEYWORDS: Agricultural Drought, Grain Yield and Correlation, MSI, NDVI, Remote Sensing, Risk Assessment, SPI, TCI

INTRODUCTION Agricultural drought produce a complex web of impacts that span many economic sectors. Among, agriculture is the primary economic sector affected by agricultural drought and particularly, short term agricultural drought at the critical growth stages has severe impacts on agriculture (Wu and Wilhite, 2004 cited in Mokhtari, 2005). Agriculture is the largest consumer of water and, therefore, the most sensitive to agricultural drought. Moisture deficit is often the most limiting factor for crop production. In Ethiopia, rainfall in main rainy season (Kiremt) is the most important for agricultural activities as nearly 95 percent of crop production is in this season (Workneh Degefu, 1987). Thus the occurrence of agricultural drought during the main rainy season has greater impact on country.s food production. This impact is largely prominent in dry land semiarid areas. Agricultural drought has either direct or indirect impact on agricultural activities. Direct impact includes reduced crop yield, rangeland and forest productivities. The consequence of these impacts result in reduction of income of farmers and agro based industries, increased price for food and other agricultural products such as forest products. Besides, losses in crop production, agricultural drought is associated with increases in insect infestation, plant disease and wind erosion (Mokhtari, 2005). Agricultural drought induced physiological stress increases a plant’s


2

Sumanta Das, Malini Roy Choudhury & Sachikanta Nanda

susceptibility to disease and insects, and reduces crop survival. Furthermore, the loss of soil organic matter can lessen cropland productivity and facilitate for wind erosion. On the other extreme, agricultural drought has also social impact particularly on farmers that drive the agricultural sectors. It involves food shortage and migration to urban areas. This makes drought migrants increase pressure on social infrastructure of the urban areas and leads to increased poverty. Geospatial techniques can assist for detection and mapping of the agricultural drought prone areas including monitoring appropriate site for specific mitigation actions. Thus, agricultural drought risk zone map will be produced from this study, can be useful for policy makers to take appropriate actions depending upon the risk level. On the other hand it is helpful for researchers to generate the information including selection of drought prone area as well crop management and soil moisture conservation practices. Moreover, it may be helpful for development agencies and Government Organization (NGO) for drought management.

STUDY AREA Bankura district is situated between 22° 38’ and 23° 38’ north latitude and between 86° 36’ and 87° 46’ east longitude(Figure.1). It has an area of 6,882 square kilometres (2,657 sq mi). On the north and north-east the district is bounded by Bardhaman district, from which it is separated mostly by the Damodar River. On the south-east it is bounded by Hooghly district, on the south by Paschim Medinipur district and on the west by Purulia district. Bankura district has been described as the “connecting link between the plains of Bengal on the east and Chota Nagpur plateau on the west.” The areas to the east and north-east are low lying alluvial plains, similar to predominating rice lands of Bengal. To the west the surface gradually rises, giving way to undulating country, interspersed with rocky hillocks. Much of the country is covered with jungles. The western part of the district has poor, ferruginous soil and hard beds of laterite with scrub jungles and sal woods. Long broken ridges with irregular patches of more recent alluvium have marks of seasonal cultivation. During the long dry season large extents of red soil with hardly any trees lend the country a scorched and dreary appearance. In the eastern part the eye constantly rests on wide expanses of rice fields, green in the rains but parched and dry in summer. The Gondwana system is represented in the northern portion of the district, south of the Damodar, between Mejia and Biharinath Hill. The beds covered with alluvium contains seams of coal belonging to the Raniganj system. The climate, especially in the upland tracts to the west, is much drier than in eastern or southern Bengal. From the beginning of March to early June, hot westerly winds prevail, the thermometer in the shade rising to around 45 °C (113 °F). The monsoon months, June to September, are comparatively pleasant. The total average rainfall is 1,400 millimetres (55 in), the bulk of the rain coming in the months of June to September. Winters are pleasant with temperatures dropping down to below 27 °C (81 °F) in December. The spatial features of the study area is represented through base map. It also indicates the spatial extent of the study area and it is prepared from toposheet. The base map(figure.2) is containing Administrative boundary, Location, major and minor roads, drainage and waterbodies, railway tracks and district headquarter.

OBJECTIVES The first and foremost objective of the study is to assess the spatiotemporal occurrence of agricultural drought and their impacts on agricultural production. For achieving the objective, the following analyses are required: •

Drought risk assessment using remotely sensed image based vegetation, precipitation, Temperature and soil moisture indices.

Impacts of drought on agricultural production


Geospatial Assessment of Agricultural Drought (A Case Study of Bankura District, West Bengal)

3

Drought risk zone map showing the severity of drought condition at various levels

MATERIALS USED Spatial Data • Satellite data •

LANDSAT ETM+ (2000)

LANDSAT TM (2005)

LANDSAT TM (2010)

Toposheet of the study area( 1: 50,000)

SRTM data(90m. spatial Resolution)

Non-Spatial Data •

Agricultural yield data

Soil data

Monthly average rainfall data(2000-2010)

The Satellite and Sensor characteristics of these imageries are shown in the Table 1. Softwares •

ArcGIS 9.3

ERDAS Imagine 9.1

TNT Mips

Microsoft Office Excel 2007

METHODOLOGY Data Collection The first phase of the methodology is data collection. The relevant data has been collected from different sources. Remote sensing data is having a immense role in agricultural drought study. In this study three years of satellite data has used between 2000-2010 time period with 5 years interval. All the satellite images in Figure. 4, (LANDSAT ETM+, 2000), (LANDSAT TM, 2005) and LANDSAT TM, 2010) has been downloaded from Global Land cover Facility(GLCF) and GLOVIS site and SRTM data is downloaded from SRTM site. Toposheet has been collected from Survey of India at 1:50000 scale. Toposheet is very essential for delineating the Area of Interest(AOI) and base map preparation. Seasonal rainfall data has collected from Indian Meteorological Dept.(IMD) in 2000-2010. And other ancillary data like Agricultural yield data has collected from WBSMB(West Bengal State Marketing Board). Image Pre-Processing Image pre-processing is very necessary to further use of the data for processing. Here geometric correction and radiometric correction has been done as image pre-processing techniques. In geometric correction, Toposheet has been georeferenced first in ERDAS Imagine 9.1 software and On the basis of georeferenced Toposheet all the satellite images has been georeferenced by using Image to Image georeferencing technique in Erdas Imagine 9.1 software and atmospheric correction of satellite data(Figure.5(b)) has done as radiometric correction. Image Processing In Image processing phase supervised Classification of the satellite image and digital indices method has followed. The supervised classification has done in ERDAS Imagine 9.1 software by using Maximum Likelihood algorithm to show the basic land use and land cover classes.


4

Sumanta Das, Malini Roy Choudhury & Sachikanta Nanda

Digital Indices for Agricultural Drought Assessment Different types of drought require different indices that can be used to quantify the moisture condition of a region and thereby detect the onset and measure the severity of drought events, and to quantify the spatial extent of a drought event thereby allowing a comparison of moisture supply condition between regions (Quiring and Papakryiakou, 2003; Beyene Ergogo, 2007). It has become clear that no single indicator or index is adequate for monitoring drought on regional scale. Instead, a combination of monitoring tools integrated together is preferable for producing regional or national maps (Martini et al., 2004). Thus, spatiotemporal patterns of seasonal drought can be detected using meteorological, vegetative as well as crop performance indices among others. Normalized Difference Vegetation Index NDVI is an index of vegetation health and density and computed from the satellite Image using spectral radiance in red and near infrared reflectance using the formula(eq.1): NDVI= (NIR-R) / (NIR+R) equation

(1)

Where, NIR= near infrared band, R= Red band; NDVI is a powerful indicator to monitor the vegetation cover of wide areas, and to detect the frequent occurrence and persistence of droughts (Thavorntam and Mongkolsawat, 2006). It provides a measure of the amount and vigor of vegetation at the land surface. The magnitude of NDVI is related to the level of photosynthetic activity in the observed vegetation. In general, higher values of NDVI indicate greater vigor and amounts of vegetation. Tucker first suggested NDVI in 1979 as an index of vegetation health and density (Thenkabail et al., 2004) and it has been considered as the most important index for mapping of agricultural drought (Vogt, 2000 cited in Mokhtari, 2005). NDVI is a nonlinear function that varies between -1 and +1 and values of NDVI for vegetated land generally range from about 0.1 to 0.7, with values greater than 0.5 indicating dense vegetation (FEWSnet). NDVI is good indicator of green biomass, leaf area index and patterns of production (Thenkabail et al., 2004). Furthermore, NDVI can be used not only for accurate description of vegetation vigor, vegetation classification and continental land cover but is also effective for monitoring rainfall and drought, estimating net primary production of vegetation, crop growth conditions and crop yields, detecting weather impacts and other events important for agriculture, ecology and economics (Ramesh et al., 2003). NDVI Anomaly NDVI can be used as an index to assess crop condition through analysis of NDVI anomaly (Murali et al., 2008). Vegetative drought index has been calculated using NDVI values. Maximum NDVI and long term mean maximum NDVI in the long term were computed in order to derive NDVI anomaly. NDVI anomaly percentage was then derived using the following formula(eq.2) for each grid cell in the study area: NDVI Anomaly = [(NDVI Max - Mean NDVI Max)/ (Mean NDVI Max)]*100 equation

(2)

Where, NDVI max =Maximum NDVI of the year and Mean NDVI max=long term mean maximum NDVI of the range of year. The resulting NDVI anomaly percentage assigned to respective grid cell was reclassified into five drought severity classes based on Table 2. Vegetation Condition Index Kogan (1990) developed Vegetation Condition Index (VCI) using the range of NDVI which is a good indicator for assessing the severity of agricultural drought. It is defined as(eq.3): VCI = (NDVI-NDVI Min) / (NDVI Max-NDVI Min)*100 equation

(3)

Where, NDVI is the actual value of NDVI , NDVI max and NDVI min are smoothed weekly NDVI absolute maximum and its minimum. The VCI values between 50 to 100 % indicate optimal or above-normal conditions. At the VCI value of 100% the NDVI value for this month (or week) is equal to NDVImax. Different degrees of a


Geospatial Assessment of Agricultural Drought (A Case Study of Bankura District, West Bengal)

5

drought severity are indicated by VCI values below 50%. The resulting VCI percentage assigned to respective grid cell was reclassified into five drought severity classes based on Table 3. Temperature Condition Index TCI is also suggested by Kogan(1997). It was developed to reflect vegetation response to temperature i.e. higher the temperature more extreme the drought. TCI is based on brightness temperature and represents the deviation of current month’s value from the recorded maximum. It is defined as(eq.4): TCI = (BT Max-BT) / (BT Max-BT Min)*100 equation

(4)

Where, BT, BT max and BT min are smoothed weekly brightness temperature absolute maximum and its minimum. The TCI value of 50% indicates the normal condition or no drought. At the TCI value of 100% represents the optimal/above normal condition and Different degrees of a drought severity are indicated by TCI values below 50%. The resulting TCI percentage assigned to respective grid cell was reclassified into five drought severity classes based on Table 4. Yellowness Index (YVI) The TM Tasselled Cap yellowness vegetation index (YVI) is a linear combination of the six reflecting wavebands. The YVI coefficients with the highest values are for the red (negatively loaded ) and the near-infrared (positively loaded) wavebands. The YVI is a more direct measure of vegetation abundance. It is calculated as(eq.5 & 6): YVI= - 0.16263 * band1 - 0.040639 * band2 - 0.85468 * band3 + 0.05493 * band4 + 0.24717 * band5 0.11749 * band7 equation

(5)

[for Landsat TM] YVI= - 0.16263 * band1 - 0.040639 * band2 - 0.85468 * band3 + 0.05493 * band4 + 0.24717 * band5 0.11749* band8 equation

(6)

[for Landsat ETM+] The resulting YVI values are assigned to the respective image was reclassified into five drought severity classes based on Table 5. Moisture Stress Index(MSI) Moisture Stress Index is Used to determine the soil moisture condition during drought. It is a good indicator of agricultural drought. It has been calculated by using MIR band and NIR band of Landsat data. MSI value range starts from 0 to 4. It is computed as(eq.7 & 8) MSI= Band 7/ Band 4 {for Landsat TM} equation

(7)

MSI= Band 8/ Band 4 {for Landsat ETM+} equation

(8)

Where, [band 4 is NIR and band 7 is MIR for Landsat TM] and [band 4 is NIR and band 9 is MIR for Landsat ETM+]. The resulting MSI values assigned to respective grid cell was reclassified into five drought severity classes based on Table 6. Standardized Precipitation Index Standardized Precipitation Index (SPI) is an index that was developed to quantify precipitation deficit at different time scales, and can also help assess drought severity. It is defined as(eq.9): SPI = {(Xij − Xim) /σ } equation

(9)

where, (Xij= is the seasonal precipitation and, Xim is its long-term seasonal mean and σ is it’s standard deviation.). SPI results computed from seasonal rainfall data were assigned to each grid cell of the study area, and reclassified based on drought severity classes as shown the Table 7.


6

Sumanta Das, Malini Roy Choudhury & Sachikanta Nanda

Computation of Yield Reduction The impact of agricultural drought on crop production can be largely expressed by yield reduction. The collected year wise(2000,2005 and 2010) agricultural yield data in individual blocks from WBSMB(West Bengal State Marketing Board) of the entire study area has computed in % and incorporated over the study area in ARC GIS 9.3 software and finally it was divided according to zone. The formula is shown in eq.10. Yr=((Expected Production of the Year-Actual Production of the Year)*100) equation

(10)

The major crops grown in the study area paddy, wheat, maize, arhar, oilseeds, teff , sorghum, finger millet, bean, chick pea, field pea, and lentil, were considered in computation of aggregated yield reduction. Regression Analysis of Grain Yield with Drought Indices Output Grain yield data and data derived from the drought indices were prepared for simple regression analysis. The average raster cell values of NDVI, SPI, NDVI anomaly, VCI, TCI and MSI images were extracted using ERDAS IMAGINE 9.1 and TNT Mips software and the data manually input in MS Excel for correlation and linear regression analysis. The relationship between NDVI anomaly, SPI, VCI, TCI and MSI result from each seasonal year with corresponding grain yield anomaly was computed to validate the derived indices. In order to understand the response of NDVI for rainfall event at different time interval , simple regression analysis between zero, one, two, three and four decade time or preceding rainfall. Besides, different information related to agricultural drought hazard and their impacts on agricultural activities as well as cropping practices were collected from Zonal agricultural and rural development, It was also used for the evaluation of the result obtained from satellite images. Agricultural Drought Vulnerability Map Agricultural drought risk map of the study area was produced from the output derived from satellite based vegetation, climate, soil moisture stress and temperature indices of each year by using multi criteria evaluation (MCE) techniques. In order to compute the frequency of drought occurrence, drought class image from each index was reclassified into Boolean image based on their threshold value and frequency maps were generated at each pixel level for each drought index in ARC GIS 9.3 software. The seasonal frequency maps derived from each drought indices were reclassified into common scale based on the frequency of drought occurrence. According to Lemma Gonfa (1996), the probability of drought occurrence in a given area can be classified into high, moderate and low drought probability zones when drought occurs in more than 50 percent, 30 to 50 percent and less than 30 percent of the years, respectively. Finally, maps from each drought indices were weighted according to the percentage of influence using ARC GIS software, and then combined using weighted overly analysis.

RESULTS AND DISCUSSIONS The chapter explains about how the analysis has been done taking into consideration different indices being computed as well as the statistical correlation and regression method applied between various digital indices and agricultural yield data to determine the yield reduction over the drought years. The final results achieved after the entire analyses are embedded in this chapter. Land Use and Land Cover Map As far as agricultural activity is concerned, Land use pattern is an important factor that influences agricultural production and productivity. The land use/ land cover map (Figure.6) has been prepared of Bankura district in Erdas Imagine 9.1 software through supervised classification technique by using maximum likelihood algorithm for understanding the basic land use land cover types over the study area. These include Agricultural land, Fallow land,


Geospatial Assessment of Agricultural Drought (A Case Study of Bankura District, West Bengal)

7

Wetland, Dense Vegetation, Mixed Vegetation, Degraded vegetation, Water bodies, Sandy area, Barren Land and Settlements area. NDVI (Normalized Difference Vegetation Index) LANDSAT ETM+ and TM data is used for computation of NDVI (Figure.7.(a),(b) & (c)) for three different years (2000, 2005 and 2010) that represents the healthy condition and distribution of vegetation. From NDVI values NDVI anomaly can also be prepared which is a good indicator of assessment of drought. NDVI value lies between -1 to +1. It has been observed that, in the year of 2000 the minimum NDVI value is -1 and the maximum value shows .9362. Similarly, in 2005 the minimum NDVI value is -1 and the maximum is .9921 and in 2010, the minimum NDVI value is -1 and the maximum value is .8593. From this NDVI value we can assess the vegetation healthiness. From this NDVI study it is clearly observed that, NDVI of 2010 year is indicating the maximum unhealthy condition of vegetation rather than 2000 and 2005. In the other side NDVI of 2005 is showing the maximum healthy condition between three years. NDVI Anomaly NDVI anomaly is one of agricultural drought index that shows the severity level. Based on this index, spatial pattern of agricultural drought for drought years (2000, 2005 and 2010) was computed for agricultural production area to determine the severity of agricultural drought (Figure. 8.(a),(b) & (c)). The result provided the spatial patterns of agricultural drought events, and the level of drought severity ranges from low to very severe in each 2000, 2005 and 2010 drought years. However, the extent of very severe drought covers small areas. The majority of the study area was stricken by moderate and severe agricultural drought. From Figure. 8.(a),(b) & (c) it is clearly observed that, the area covered by very severe drought is maximum in 2000 year, minimum in 2005 and moderate to maximum in 2010 during cropping season. Area covered by moderate and severe drought is maximum in 2010 years and area of no and low drought is maximum in 2005 between three years of study. Vegetation Condition Index(VCI) for Drought Assessment VCI is one of the most important parameter of agricultural drought. It has developed by taking range of NDVI values between three years of study. The result of VCI values(figure. 9(a),(b) & (c)) shows that, the maximum area within the entire study area is covered by severe and moderate drought in the year of 2000. In 2005, very severe, severe and moderate drought prone area is less visible, Maximum area is covered by no and low drought and moderate drought is covered maximum area in 2010 year. Land Surface Temperature(LST) Figure. 10(a),(b) & (c) shows the spatial distribution of surface temperature of Landsat ETM+ and TM data. The LST in 2000 year ranged from 18º to 42 °C, 14 º to 30 º C in 2005 and 17 º to 40 º C in 2010.It is observed from the result that west and south-west part exhibits high temperature mainly due to waste land, barren lands and rocky terrain. Some of the high temperature zones are also seen in the central part of the image mainly due to commercial/industrial land use. The temperature calibration of the thermal infrared band into the value of ground temperature has been done using eq. 11 and 12. L=Lmin+ ((Lmax-Lmin)/255)*Q equation

(11)

T=K2/ (ln (K1/L+1)) equation

(12)

Where, L: value of radiance in thermal infrared. T: ground temperature (k). Q: digital record. K1, K2: calibration coefficients.


8

Sumanta Das, Malini Roy Choudhury & Sachikanta Nanda

K1=666.09 watts / (meter squared * ster* ¾m) K2=1282.71 Kelvin Temperature Condition Index(TCI) for Agricultural Drought TCI is another most important parameter of agricultural drought like VCI. It has developed by taking range of LST values between three years of study. The result of TCI values(Figure. 11.(a),(b) & (c)) shows that, the maximum area within the entire study area is covered by Very severe and severe drought in the year of 2000, very few portion of area covered by very low and no drought have been found in southern and south-eastern part of Bankura district. In 2005, very severe and severe drought prone area is very less visible, Maximum area is covered by no, very low and moderate drought and Severe and moderate drought is covered maximum area within the entire area in 2010 year. The highly drought prone area has mainly found in western, south-western and central part of the study area due to barren lands and industrial/commercial area. Yellowness Index(YVI) From the results of YVI(Figure. 12.(a),(b) & (c)) it has observed that, severe and moderate drought have taken place most of the area within the study area in the year of 2000. Similarly in 2005, no and low drought have covered up maximum places and severe drought area has not identified. In 2010, moderate drought prone area is more visible within the total geographical area. Moisture Stress Index(MSI) MSI indicates the soil moisture condition and it is very useful parameter for assessing agricultural drought. In Figure. 13.(a),(b) & (c) it has been observed that, the MSI value ranged from 0 to 2.96, 0 to 2.97 and 0 to 3.93 for the year of 2000,2005 and 2010 respectively. The Western, South-Western and to some extent central part within the total area is under maximum soil moisture stress condition due to presence of barren lands, hard rock terrain and some industrial/commercial area. So, the 2010 year is identified maximum stressed condition between all the three years. Standardized Precipitation Index(SPI) SPI is computed for growing season of Bankura district. The analysis of SPI (Figure. 14.(a),(b) & (c)) of each three years of drought revealed that drought has occurred at different level of severity from 2000 to 2010 cropping season. The drought that happened in year of 2000 and 2010 was very severe rather than 2005 as explained by the SPI values that range from -1.98 to 2.00 and -1.71 to 2.00 respectively. The result indicates that during 2000 and 2010 both years, there was rainfall deficit in the growing season and it, therefore, was the worst dry seasons. On the other side, the SPI values in 2005 was ranged between -1.03 and 2.23 which indicates during this year there was relatively less precipitation deficit in the growing season. Correlative Study between the Indices The regression analysis performed between Land surface temperature and NDVI & TCI and VCI of each three years to show the relationship between the indices. The relationship between Land surface temperature and NDVI is shown on the Figure.15.(a),(b) & (c), which indicates, with the increasing of land surface temperature vegetation healthy condition and distribution of vegetation both decreases and it’s showing the negative correlation between Temperature and NDVI. Figure.16.(a),(b) & (c) is showing the relationship between TCI and VCI. With the increasing of TCI, VCI also increases and it indicates the positive correlation between them. Yield Reduction as an Impact of Agricultural Drought Agricultural drought decreases grain yield through the reduction of various yield components of a crop. Considering spatial pattern of yield reduction, analysis has been carried out in the study area from 2000 to 2010 cropping season. Among them, the highest yield reduction occurred in 2000 and 2010 cropping season and the spatial pattern of


Geospatial Assessment of Agricultural Drought (A Case Study of Bankura District, West Bengal)

9

yield reduction has been illustrated on the Figure.17.(a),(b) & (c). According to the result, 2000 cropping season, nearly all area was hit by agricultural drought and the agricultural yield reduction was reached upto 80 percent. During this season, Southern, Western/North-western and central parts of the Bankura district encountered 60 to 80 percent yield reduction while small areas in Western, Southern and Northern part along the river side encountered 0 to 20 percent yield reduction. Whereas, 20 to 60 percent yield reduction was encountered maximum area within the total area. In 2005 cropping season, the level of yield reduction has been reduced upto 60 percent. 0 to 20 percent yield reduction covers maximum area of the total area of Bankura district. 40 to 60 percent crop reduction has been found in Western, few portion of Northern and Southern part in this study area. In 2010 cropping season, the level of reduction again reached upto 80 percent. 60 to 80 percent yield reduction has occurred in Western and few portion of South-Western part. Maximum area is covered by 40 to 60 percent yield reduction and 0 to 40 percent yield reduction has been found total North-Eastern part along the Damodar and Darkeswar river side, and few portion of central and South-Western part within the total geographical area. According to Wilhelmi et al. (2002), agricultural drought is the leading cause of crop failure throughout the world. Thus, moisture deficit significantly influence the growth and development of crops and the ultimate yields. Agricultural production and agricultural area under different crops in 2000, 2005 & 2010 cropping season is graphically represented in Figure. 18 and Figure. 19 respectively. The two graphs indicate that impacts of drought on agricultural production over the years. Agricultural production was maximum in 2005 related to 2000 & 2010 due to less impact of drought in 2005 cropping season. In 2000, it is found the agricultural production and crop under agricultural area is very less due to maximum impacts of drought. Relationship between Indices and Yield Reduction In order to validate satellite derived output, grain yield of agricultural production is the main ground truth data. Therefore, it is crucial to analyze the relationship between Indices and grain yield reduction(%) to quantify the impact of agricultural drought on agricultural production. Figure.20.(a),(b),(c) & (d) is showing the relationship between NDVI anomaly and grain yield reduction(%), SPI and grain yield reduction(%), TCI and grain yield reduction(%) & MSI and grain yield reduction(%) respectively, were analyzed and from scatter plot. In Figure.20.(a), it can be observed that the two variables have established good correlation (r2= -0.275) and it indicates the negative correlation between NDVI Anomaly and YR(%). If NDVI Anomaly value increases, YR will be less and vice-versa. This means that 54 percent of yield variability can be explained by NDVI anomaly. Figure.20.(b) is showing the relationship between SPI and YR(%) and from this regression analysis it is observed that with the increasing of SPI values YR will be decreased and vice-versa. So, the trend line indicates negative correlation (r2= -0.379). In Figure.20.(c), it is shown the relationship between TCI and YR(%).The result revealed that the relationship established between the two variables is negative (r2= -0.195). With the increasing of TCI, YR will be decreasing and vice-versa. Figure. 20.(d) is showing the correlation between MSI and YR(%). The result is indicating the positive correlation (r2= 0.696) between them. MSI increases so do agricultural yield reduction and vice-versa. Thus the strength of the indices is used to explain the existence of agricultural drought through statistical regression analysis of agricultural production. Agricultural Drought Risk Zonation Maps The agricultural drought risk maps for each three years has been prepared by integrating all the drought frequency maps generated from the six drought indices, SPI, TCI, VCI, MSI, NDVI Anomaly and YVI for each year(Figure. 21.(a),(b) & (c)). The six layers representing drought indices were prioritized according to their degree of influence using pair-wise comparison and the drought condition maps were obtained by overlaying all the six indices in terms of weighted overlay methods using the spatial analysis tool in ArcGIS 9.3. During the weighted overlay analysis, the ranking has been


10

Sumanta Das, Malini Roy Choudhury & Sachikanta Nanda

given for each individual parameter of each indices and the weights were assigned according to the influence of the different parameters and was presented in Table 8. According to the result derived from the integration of all these parameters Bankura district is classified into No, Moderate, Severe and Very Severe agricultural risk zones (Figure.21.(a),(b) & (c)). From the obtained result of drought condition map of 2000(Figure.21.(a)) it has observed that, No, Moderate, Severe and very Severe drought is covered up 13, 29, 31 and 27 percent of the total geographical area of Bankura District, respectively. Drought condition map of 2005(Figure.21.(b)) is showing 37, 40, 19 and 4 percent area is covered by No, Moderate, severe and Very Severe drought within the entire study area. Similarly, drought condition map of 2010(Figure.21.(c)) shows that, 1, 33, 65 and 1 percent area is occupied by No, Moderate, Severe and Very severe drought. The percentage of area coverage by various levels of drought is represented graphically in Figure.22. From the result(Figure.22), it can be demonstrated that, the area under Very severe drought is maximum and No drought is minimum in 2000. Similarly, area under Moderate drought and No drought is maximum and Very Severe drought is minimum in 2005 and in 2010, No and very Severe drought is minimum and Severe drought covered maximum area of the entire geographical area. So, 2005 is less vulnerable and 2000 is more vulnerable of agricultural drought. Agricultural drought vulnerability index map(Figure.23) has been prepared by integrating drought condition and it’s frequency of 2000, 2005 & 2010 by weighted sum analysis in arcGIS 9.3 software and it is classified into Low, Moderate and High risk zone. Drought vulnerability was ranged from 0 to 100. According to the result(Figure.24), the value 24 to 50 indicates Low drought vulnerable, 50 to 64 is moderate vulnerable and 64 to 96 is high vulnerable. Mainly, North-Western, Western and South-Western part of Bankura district is showing high agricultural drought prone area due to dry hilly terrain, commercial/industrial land and lack of waterbodies/river flows. Some Eastern and central part is moderate agricultural drought prone and North-Eastern , Eastern and few South-eastern part along the both side of damodar, darkeswar and Kasai river is Low drought vulnerable zone. The probability of occurrence of agricultural drought ranges from 20 to 45 percent for slight severity level, 45 to 60 percent for moderate severity level and 60 to 100 percent for severe severity level. Thus, the western and South-Western part of Bankura district is categorized into Severe to Very Severe drought probability zone while most of North-Eastern, Eastern and central part is into Moderate to Low drought probability zone. Table.9 is showing percentage coverage of area under different agricultural drought risk levels.

CONCLUSIONS Agriculture remains by far the most vulnerable and sensitive sector that is seriously affected by the impacts of climate variability and climate change, which is usually manifested through rainfall variability and drought. Rainfall is one of the climatic variables that largely determine the occurrence of drought and also influences the growth and development of vegetations which is reflected by NDVI. In this study, the agricultural drought prone areas in the Bankura district were identified by using Remote Sensing and GIS technology and drought risk areas were to delineate by integration of satellite images, meteorological information and crop yield data. The role of satellite derived index for drought detection has been exemplified by integrating meteorological derived index called Standardized Precipitation Index. It is found that the temporal variations of NDVI anomaly, VCI, TCI and MSI are closely linked with SPI and a strong linear relationship exists between them. Satellite derived drought-monitoring indices have also been correlated with precipitation index to see how vegetation stress condition and consequently agricultural production yield is changing with the variability of rainfall. Moreover, a significant correlation has been observed between NDVI anomaly, SPI, TCI &MSI and crop yield anomaly for most of the blocks of Bankura district. The seasonal pattern of rainfall, temperature and NDVI, suggest that the Western and North-Western part of the Bankura district is a low rainfall and high temperature area, where SPI value is low


Geospatial Assessment of Agricultural Drought (A Case Study of Bankura District, West Bengal)

11

and the corresponding and NDVI values is also low. But MSI and LST is higher. Thus it can be said that, NDVI index and precipitation index & MSI index and LST shares a strong correlation where water is a major limiting factor for plant growth. Agricultural drought risk can be viewed as a product of both exposure to the climate hazards and the vulnerability of farming or cropping practices to drought conditions. In view of this, agricultural risk zone map produced by integrating all drought frequency maps derived from all drought indices indicates that Bankura district is classified into Low, Moderate and High agricultural drought risk zone, respectively. The results of this study are being used for the development of a regional drought monitoring system. Considering the spread and frequency of droughts in the region on the one hand, and the lack of ground climate observations and technical capacity in the countries of the region to deal with droughts on the other, such a system could play an invaluable role for drought preparedness, while identifying appropriate sites for specific adaptation and mitigation actions.

RECOMMENDATIONS Based on the findings of the study, the following recommendations are suggested: •

Prioritization and implementation of site specific adaptation and/or mitigation projects should be made based on such identification of risk levels of specific locations.

Since agricultural drought severity levels vary spatially, selection of agricultural technologies and information (drought tolerance crops, the type of crop variety and soil moisture conservation practices) should be made to fit in to the agricultural drought severity levels.

The methods as well as results of agricultural drought assessment and risk mapping are believed to be highly important for decision makers and stakeholders, who have a stake in the study area. However, it is recommended that future studies can build up on this work by including real time seasonal rainfall forecasts as model parameter so that stakeholders can get early warning information that helps to take necessary adaptation measures and reduce the impact of agricultural drought. Moreover, Government responses to drought is necessary through adhoc or crisis management. Some specific

tasks of agricultural drought management is expected to perform. These are, food production and security, employment generation, contingency crop plan, livestock management, drilling tube wells to provide drinking water, providing monitory relief and some social security and infrastructure facility schemes.

ACKNOWLEDGEMENTS We express our gratefulness to the Director, NRDMS, Bankura district, West Bengal and all concerned authorities of NRDMS, West Bengal, IMD, Kolkata and Dr. R. Annadurai (Professor and Head, Dept. Of Civil Engineering), Dr. R. Sivakumar(Associate Professor and our class-in-charge), Dr. M. Nagarajan (Assistant prof.) and Mr. V. Satya Ramesh Potti (Assistant prof.) of SRM University for their continuous support in this study.

REFERENCES 1.

Anonymous. 2004. Drought 2002: A report (Part.1) and states report (Part-II, 4 volumes). Department of Agriculture & Co-operation, Ministry of Agriculture, Government of India, Krishi Bhawan, New Delhi–110 001, India.

2.

Anyamba, A. and Tucker, C.J. (2005). Analysis of Sahelian vegetation dynamics using NOAA-AVHRR NDVI data from 1981-2003. Journal of Arid Environment, 63:596.614.

3.

Aslam, M., Khan, I. M., Saleem, A. and Ali, Z. (2006). Assessment of water stress tolerance in different maize accessions at germination and early growth stage. Pakistan Journal of Botany, 38: 1571- 1579

4.

Bastiaanssen, 1998, Remote Sensing in Water Resources Management: The state of the art, IWMI publication,


12

Sumanta Das, Malini Roy Choudhury & Sachikanta Nanda

p118. 5.

Ergogo, B. (2007). Drought assessment for Nile Basin using Meteosat Second Generation data with special emphasis on the upper Blue Nile Region. Msc. Thesis, ITC, Enschede.

6.

Chavez, P. S. Jr, 1996, Image-based atmospheric corrections – revisited and improved. Photogrammetric Engineering and Remote Sensing, 62, 1025–1036.

7.

Chavez, P. S. Jr, 1989, Radiometric calibration of Landsat Thematic Mapper multispectral images. Photogrammetric Engineering and Remote Sensing, 55, 1285–1294

8.

Chopra, P. (2006). Drought risk assessment using remote sensing and GIS: A case study of Gujarat. Msc. Thesis, ITC, Enschede.

9.

Comenetz, J. and Caviedes, C. (2002). Climate variability, political crises, and historical population displacements in Ethiopia. Environmental Hazards, 4: 113-127.

10. Dracup, J. A., Lee, K. S. and Paulson, E. G. (1980). On the definition of droughts. Water Resources Research, 16:297-302 11. FAO. (1979). Agrometeorological crop monitoring and forecasting. In: FAO Plant Production and Protection paper No. 17. FAO, Rome. p. 212. 12. Goldberg, I. Ed., 1972, Agro climatic atlas of the World, Hydrometizdat, p 133 13. Hayes, M. J. (1999). Drought Indices, NDMC – Drought Happens, Drought Indices. Accessed on 07th October 2009: http://www.drought.unl.edu/whatis/indices.htm. 14. Hayes, M. J. (2000). Revisiting the SPI: Clarifying the Process. Drought Network News (Newsletter of IDIC and NDMC), Vol. 12 (1): pp. 13–14. 15. Hayes, M. (2004). Drought Indices, National Drought Mitigation Centre. Accessed on 07th October 2009: http://www.drought.unl.edu/whatis/indices.htm. 16. Hayes, M., Svoboda, M., and Wilhite, D. A.,(2000). Monitoring Drought Using the Standardized Precipitation Index. In: D.A. Wilhite(ed.).Drought: A Global Assessment, Chapter 12,pp. 168–180.Natural Hazards and Disasters Series.Routledge Publishers, London. 17. Hayes, M. J., Wilhelmi, O. V. and Knutson, C. L., (2004). Reducing Drought Risk: Bridging Theory and Practice. Natural Hazards Review, Volume 5 (2), pp. 106-113. 18. Jeyaseelan, A.T., (2004). Drought and flood assessment and monitoring using remote sensing and GIS.URL: http://www.wamis.org/agm/pubs/agm8/Paper-14.pdf 19. Kogan,

F.N.,

(1997).

Contribution

of

Remote

Sensing

to

Drought

Early

Warning.URL:

http://drought.unl.edu/monitor/EWS/ch7_Kogan.pdf 20. Gonfa, L. (1996). Climate classification of Ethiopia. In: Meterological Reserch Report Series No.3. Addis Ababa, pp.1-8. 21. Liillsand & Kieffer, Chipman,2004, Remote Sensing & Image Interpretation. Page. 491-624. 22. Martini, M., Soumare, P.B., Ndione, J. and Touré, A. (2004). Crops and range land monitoring in Senegal using SPOT 4/5 vegetation data. In: Proceeding of the 2nd vegetation user conference, PP. 239-245,( F. Veroustraete, E. Bartholomé and W.W. Verstraeten, Eds), Office for Official Publications of the European Communities. 23. Murali, K., Ravikumar, G. and Krishnaveni, M.(2008). Remote sensing based agricultural drought assessment in Palar basin of Tamil Nadu State, India. Journal of the Indian Society of Remote Sensing, 37:9-20. 24. National Meterological Service Agency (NMSA). (1996). Assessment of drought in Ethiopia. In: Meterological research report series No.2.Addis Ababa.


Geospatial Assessment of Agricultural Drought (A Case Study of Bankura District, West Bengal)

13

25. National Drought Mitigation Center. http://www.drought.unl.edu 26. Quiring, S.M. and Papakryiakou, T.N.(2003). An evaluation of agricultural drought indices for the Canadian prairies. Agricultural and forest Meteorology, 118:49-62. 27. Ramesh, P., Singh, P., Roy, F. and Kogan, F. (2003). Vegetation and temperature condition indices from NOAA AVHRR data for drought monitoring over India. International Journal of Remote sensing, 24:4393-4402. 28. Thavorntam, W. and Mongkolsawat, C. (2006). Drought assessment and mitigation through GIS and Remote sensing. URL: 29. http://www.gisdevelopment.net/application/natural_hazards/drought/ma 30. Thenkabail P. S., Gamage M. S. D. N., Smakhtin, V. U. (2004). The Use of Remote Sensing data for drought assessment and monitoring in south west Asia. Colombo, Sri Lanka, International Water Management Institute: pp.1-23. 31. Thiruvengadachari, S., Prasad, T. S. and Harikishan, J. (1987). Satellite monitoring of agricultural drought in Anantapur district in Andhra Pradesh State. Report No.: RSAM - NRSA - DRM - TR - 03/87. India: Drought Mission Team, Department of Space, Government of India. pp. 35. 32. Thornthwaite, C.W., (1948). An approach toward a rational classification of climate, Geographical review, 21: 633-655. 33. Ungani, L.S., Kogan, F. N. (1998). Drought monitoring and corn yield estimation in southern Africa from AVHRR data. Remote Sensing of Environment 63:219–232. 34. Wilhelmi, O.V., Hubbard, K.G. and Wilhite, D.A.(2002). Spatial representation of agroclimatology in a study of Agricultural drought. International Journal of Climatology, 22:1399-1414. 35. Wilhelmi, V.O. and Wilhite, D.A., (2002). Assessing Vulnerability to Agricultural Drought: A Nebraska Case Study. Natural Hazards, Vol. 25: pp. 37-58.

APPENDICES

Figure 1: Location Map of the Study Area

Figure 2: Base Map of the Study Area


14

Sumanta Das, Malini Roy Choudhury & Sachikanta Nanda

Figure 3: Flow Chart of the Methodology

Figure 4: Collected Satellite Data of the Study Area at Various Time Scale

(a)

(b)

Figure 5: Figure Showing before Atmospherically Corrected Image (a) and after Atmospheric Correction(b)


Geospatial Assessment of Agricultural Drought (A Case Study of Bankura District, West Bengal)

Figure 6: Land Use and Land Cover Map of Bankura District

Figure 7(a): NDVI Map of Bankura District(2000)

Figure 7(b): NDVI Map of Bankura District(2005)

Figure 7(c): NDVI Map of the Bankura District(2010)

15


16

Sumanta Das, Malini Roy Choudhury & Sachikanta Nanda

Figure 8(a): NDVI Anomaly of Bankura District(2000)

Figure 8(b): NDVI Anomaly of Bankura District(2005)

Figure 8(c): NDVI Anomaly of Bankura District(2010)

Figure 9(a): Vegetation Condition Index Map of Bankura District(2000)


Geospatial Assessment of Agricultural Drought (A Case Study of Bankura District, West Bengal)

Figure 9(b): Vegetation Condition Index Map of Bankura District(2005)

Figure 9(c): Vegetation Condition Index Map of Bankura District(2010)

Figure 10(a): Land Surface Temperature of Bankura District(2000)

Figure 10(b): Land Surface Temperature of Bankura District(2005)

17


18

Sumanta Das, Malini Roy Choudhury & Sachikanta Nanda

Figure 10(c): Land Surface Temperature of Bankura District(2010)

Figure 11(a): Temperature Condition Index of Bankura District(2000)

Figure 11(b): Temperature Condition Index of Bankura District (2005)

Figure 11(c): Temperature Condition Index of Bankura District (2010)


Geospatial Assessment of Agricultural Drought (A Case Study of Bankura District, West Bengal)

Figure 12(a): Yellowness Index of Bankura District(2000)

Figure 12(b): Yellowness Index of Bankura District(2005)

Figure 12(c): Yellowness Index of Bankura District(2010)

Figure 13(a): Moisture Stress Index of Bankura District(2000)

19


20

Sumanta Das, Malini Roy Choudhury & Sachikanta Nanda

Figure 13(b): Moisture Stress Index of Bankura District(2005)

Figure 13(c): Moisture Stress Index of Bankura District(2010)

Figure 14(a): Standardized Precipitation Index of Bankura District(2000)

Figure 14(b): Standardized Precipitation Index of Bankura District(2005)


Geospatial Assessment of Agricultural Drought (A Case Study of Bankura District, West Bengal)

Figure 14(c): Standardized Precipitation Index of Bankura District(2010)

(a)

(b)

(c) Figure 15: Correlation between the LST and NDVI (a) 2000, (b) 2005 & (c) 2010

(a)

(b)

(c) Figure 16: Correlation between TCI and VCI (a) 2000, (b) 2005 & (c) 2010

21


22

Sumanta Das, Malini Roy Choudhury & Sachikanta Nanda

Figure 17(a): Spatial Pattern of Yield Reduction in 2000 Cropping Seasons

Figure 17(b): Spatial Pattern of Yield Reduction in 2005 Cropping Seasons

Figure 17(c): Spatial Pattern of Yield Reduction in 2010 Cropping Seasons


Geospatial Assessment of Agricultural Drought (A Case Study of Bankura District, West Bengal)

Figure 18: Line Graph Showing the Fluctuation of Agricultural Production over the Years

Figure 19: Bar Graph Showing the Status of Agricultural Area (hect.) under Crops over the Years

(a)

(b)

(c)

(d)

Figure 20: Relationship between (a) NDVI Anomaly and YR(%), (b) SPI and YR(%), (c) TCI and YR(%) & (d) MSI and YR(%)

23


24

Sumanta Das, Malini Roy Choudhury & Sachikanta Nanda

Figure 21(a): Drought Condition Map of Bankura District(2000)

Figure 21(b): Drought Condition Map of Bankura District (2005)

Figure 21(c): Drought Condition Map of Bankura District (2010)


25

Geospatial Assessment of Agricultural Drought (A Case Study of Bankura District, West Bengal)

Figure 22: Graphs Showing the Status of Drought Condition over the Years

Figure 23: Agricultural Drought Vulnerability Index Map of Bankura District Table 1: Characteristics of Satellite Imageries Used in this Study Satellite and Sensor Year No. of Bands Spatial Resolution(M) Projection Landsat ETM+ 2000-03-29 9 30 UTM45 Landsat TM 2005-11-14 7 30 UTM45 Landsat TM 2010-04-02 7 30 UTM45

Table 2: NDVI Anomaly Based Drought Severity Class NDVI Anomaly Above 0 0 to -40 -40 to -80 -80 to -120 <-120

Drought Severity Class No Drought Low Drought Moderate Drought High Drought Very High Drought

Table 3: VCI (%) Based Drought Severity Class VCI(%) 0 to 20 20 to 40 40 to 60 60 to 80 80 to 100

Drought Severity Class Very Severe Drought Severe Drought Moderate Drought Very Low Drought No Drought

Path 139 139 139

Row 044 044 044


26

Sumanta Das, Malini Roy Choudhury & Sachikanta Nanda

Table 4: TCI(%) Based Drought Severity Class TCI(%) 0 to 20 20 to 40 40 to 60 60 to 80 80 to 100

Drought Severity Class Very Severe Drought Severe Drought Moderate Drought Very Low Drought No Drought

Table 5: YVI Based Drought Severity Class YVI 0 to 40 40 to 80 80 to 120 120 to 160 > 160

Drought Severity Class No Drought Low Drought Moderately Drought Severe Drought Very Severe drought

Table 6: MSI Based Drought Severity Class MSI 0 to 0.8 0.8 to1.60 1.60 to 2.40 2.40 to 3.20 3.20 to 4.00

Drought Severity Class No Stressed Low Stressed Moderately Stressed Severe Stressed Extremely Stressed \

Table 7: SPI Based Drought Severity Class SPI 1.20 to 2.00 .40 to1.20 -.60 to .40 -1.80 to -.60 < -1.80

Drought Severity Class Extremely Wet Moderately Wet Normal Moderately Dry Extremely Dry

Table 8: Rank and Weights of Different Parameters for Drought Condition Maps Sl No. Criteria Classes Rank Influence(%) Extremely dry 5 1

SPI

2

TCI

3

VCI

Moderately dry Normal Moderately wet Extremely wet Very severe severe Moderate Very Low No Very severe severe Moderate Very Low No

4 3 2 1 5 4 3 2 1 5 4 3 2 1

25

25

20


27

Geospatial Assessment of Agricultural Drought (A Case Study of Bankura District, West Bengal)

4

MSI

5

NDVI ANOMALY

6

YVI

Table 8: Contd., Extremely stressed Severe stressed Moderately stressed Low stressed No stressed Very High High

5 4 3 2 1 5 4

Moderate

3

Low No Very severe Severe Moderate Low No

2 1 5 4 3 2 1

20

6

4

Table 9: Area(%) Affected by Various Agricultural Drought Risk Levels Agricultural Drought Risk levels High Moderate Low

Probability of Occurrence(%) 64-96 50-64 24-50

Area(sq.km)

Area(%)

2839.78 3571.74 455.56

41 53 6



1.Geospatial assessment.full