Comparing water-vegetative indices for rice (Oryza sativa L.)–wheat (Triticum aestivum L.) drought assessment

Highlights ► As rainfall increases from 500 mm, NDVI increases and reaches up to a range of 0.55 at 1100 mm; however beyond 1100 mm, no significant increase. ► June dif-NDVI contributes more to rice productivity followed by July. ► Combined effect of June, July and August, explains 15% of the variation of KPRI. ► The KPRI variation for most of the States is <15%, the multiple regression equation can be used as an indicator of regional rice production. ► As far as wheat is concerned, statistically significant relation was found between WPI and dif-NDVI during December to March. Abstract Drought indices (DI) are an useful tool for assessing different sectarian droughts. Standardized Precipitation Index (SPI) has been used worldwide to assess/monitor the onset, active phase, cessation and severity of drought. Normalized Difference Vegetation Index (NDVI) provides a comprehensive vegetation dynamics, which directly linked with rainfall received in a particular region. Indo-Gangetic Region (IGR), providing employment and livelihood to tens of millions of rural families directly or indirectly and rice ( Oryza sativa L.)–wheat ( Triticum aestivum L.) (RW) system of the Indo-Gangetic Plains (IGP) contributes 80% of the total cereal production and is critical to food security of the region. This study tries to verify the applicability of water-vegetative indices viz., SPI, Rainfall Index (RI) and NDVI for drought assessment of rice–wheat system productivity over IGR-India. The relationship between monsoon rainfall and NDVI shows that at around 1100 mm rainfall, the NDVI reached saturation point and no further significant increase in NDVI with increase of rainfall is noticed. Even though, there was a positive correlation of seasonal monsoon rainfall and average NDVI, conflicting results are noticed between monthly distribution of rainfall and monthly anomaly of NDVI over IGR States. It is noticed that June dif NDVI (actual NDVI–mean NDVI) contributes more to rice productivity followed by July. However, the combined effect of June, July and August, explains 15% of the variation of Kharif Rice Productivity Index (KRPI). As far as wheat is concerned, statistically significant relation was found between Wheat Productivity Index (WPI) and anomaly NDVI during December–March. This explains 35% of the variability in WPI. Keywords Normalized Difference Vegetation Index Productivity Drought assessment Rainfall Index Drought index Anomaly NDVI 1 Introduction Drought is the most complex and least understood of all natural hazards, affecting more people than any other hazard. Drought affects virtually all climatic regions ( Wilhite, 2000 ) and more than one half of the earth is susceptible to drought each year ( Kogan, 1997 ). According to Hewitt (1997) drought ranks first among natural disasters in terms of persons directly affected. Drought is a “creeping phenomenon” ( Gillette, 1950 ). Every year one or other part of India experiences droughts of varying intensity during different time scales. In India, the climate and weather are dominated by the largest seasonal mode of precipitation in the world, due to the summer monsoon circulation. Over and above this seasonal mode, the precipitation variability has predominant inter-annual and intra-seasonal components, giving rise to extremes in seasonal anomalies resulting in large-scale droughts and floods, and also short-period precipitation extremes in the form of heavy rainstorms or prolonged breaks on a synoptic scale. There are four major reasons for droughts in India-delay in the onset of monsoon/failure of monsoon, variability of monsoon rainfall, long break in monsoon and areal difference in the persistence of monsoon. The most affected community is the marginal farmers, as mostly they are dependent on rainfed agriculture. The success of drought preparedness and mitigation depends upon timely and accurate information on drought onset, progress, extent and its end. These types of information can be obtained through drought indices, which provide decision makers to bring out drought contingency plans. Many drought indices such as the Palmer Drought Severity Index ( Palmer, 1965 ), the decile index ( Gibbs and Maher, 1967 ), Bhalme-Mooley Index ( Bhalme and Mooley, 1979 ), the Surface Water Supply Index ( Shafer and Dezman, 1982 ), the China-Z index (CZI) ( Wu et al., 2001 ) are widely used while the Standardized Precipitation Index (SPI) ( McKee et al., 1993 ) has achieved world popularity. Most of these indices are normally continuous functions of rainfall and/or temperature. These indices were tested and modified in different parts of the world ( Nguyen and In-Na N Bobee, 1989; Wu et al., 2001; Oza et al., 2002; Ntale and Gan, 2003; Ansari, 2003; Morid et al., 2006; Patel et al., 2007 ). Generally in India, meteorological drought is defined when rainfall in a month or a season is less than 75% of its long-term mean, if the rainfall is 50–74% of the mean, a moderate drought event is assumed to occur, and when rainfall is less than 50% of its mean a severe drought occurs ( Smakhtin and Hughes, 2004 ). The geospatial techniques comprising remove sensing, satellite imaging, geographical information systems (GIS) and geographical positioning system (GPS) can be put to effective use in forecasting and monitoring drought. Satellite data can also be used to detect the onset of drought, its duration and magnitude ( Thiruvengadachari and Gopalkrishna, 1993 ). Crop yields can be predicted 5–13 weeks prior to harvest using remote sensing techniques ( Ungani and Kogan, 1998 ). Normalised Difference Vegetation Index (NDVI), Vegetation Condition Index (VCI), Temperature Condition Index (TCI) and Vegetation Health Index (VHI) have been employed to assess vegetative drought ( Kogan, 2000; Kogan, 2001; Liu and Kogan, 1996; Singh et al., 2003 ). The Normalized Difference Vegetation Index (NDVI) is the most commonly used vegetation index ( Jensen, 1996 ). NDVI (a) uses only two bands and is not very sensitive to influences of soil background reflectance at low vegetation cover, and (b) has a lagged response to drought ( Reed, 1993; Rundquist and Harrington, 2000; Wang et al., 2001 ) because of a lagged vegetation response to developing rainfall deficits due to residual moisture stored in the soil. Previous studies have shown that NDVI lags behind antecedent precipitation by up to 3 months ( Justice et al., 1986; Farrar et al., 1994; Wang, 2000; Wang et al., 2001 ). The lag time is dependent on whether the region is purely rainfed, fully irrigated, or partially irrigated. But the severity of a drought (or the extent of wetness, on the other end of the spectrum) may be defined as NDVI deviation from its long-term mean ( Johnson et al., 1993; Bastiaanssen and Ali, 2003; Thenkabail et al., 2004; Kastens et al., 2005; Bhuiyan et al., 2006 ). Thus, climatic indices and vegetative indices have their own advantages and disadvantages. Since almost all the climate based drought indicators calculated based on point data, it lacks spatial representation. Moreover, meteorologically driven indices are dependent on data collected at weather stations and some areas have very sparsely distributed stations and this affects the reliability of drought indices ( Brown et al., 2002 ). Various researchers ( Singh et al., 2003; Bhuiyan et al., 2006 ) compared various climatology based indices and vegetative indices for drought monitoring and for yield estimation ( Prasad et al., 2005, 2007; Hayes and Decker, 1996 ). But use of these indices with major cropping system of a particular region is lacking. The main objective of this study is to investigate the applicability of climate based index and vegetative based index for drought assessment of rice–wheat system productivity over IGR-India States. 2 Background information about study region Indo-Gangetic Region (IGR), the “food basket” and “food bowl” of South Asia, providing employment and livelihoods to tens of millions of rural families directly or indirectly, extends over four countries – Bangladesh, Nepal, India and Pakistan. RW system of the Indo-Gangetic Plains (IGP) contributes 80% of the total cereal production and is critical to food security in the region. Rosegrant et al. (2001) estimated that the demand for these two cereals in South Asia is expected to grow at 2.02% and 2.49% per year, respectively. With the advent of “Green revolution” in the 60s and 70s brought the productivity to higher level tremendously in a short span of time thanks to expansion of irrigation, adoption of high yielding varieties (HYV) and improvement of infrastructural facilities ( Rai, 2000 ). But during the past decade, yields have stagnated or possibly declined, and there are large gaps between potential yields, experimental yields and farmers’ yields ( Gill, 1999; Ladha et al., 2003 ). The IGR primarily comprises five States: West Bengal (88,752 sq.km), Bihar (undivided) (1,73,877 sq.km), Uttar Pradesh (undivided) (2,94,411 sq.km), Haryana (44,212 sq.km) and Punjab (50,362 sq.km) ( Fig. 1 ). The area, production and yield of rice and wheat during 2006–2007 over IGR-States are given in Table 1 . The symptoms of degradation of the resource base include declining soil organic matter content and nutrient availability, and increasing soil salinisation and weed, pathogen and pest populations and declining of ground water and increasing air and water pollutions have been perceived in this part of the region ( Humphreys et al., 2004 ). The situation will be further complicated by the projected climate change scenario and the Intergovernmental Panel on Climate Change (IPCC) in its fourth assessment report (AR4) affirms with very high confidence (90% probability of being correct) that human activities, since industrialization have caused the planet to warm by about 1 °C. With the doubling of carbon dioxide content in the atmosphere, this trend is projected to cause average global warming of around 3 °C compared to the pre-industrial level. Climate variability is also projected to increase, leading to uncertain onsets of monsoons and more frequent extremes of weather, such as more severe droughts and floods. Therefore the sustainability of RW systems of the IGR and the ability to increase production in pace with population growth are major concerns. 3 Data and methodology 3.1 Data 3.1.1 Rainfall and rice–wheat productivity The monthly rainfall data series during 1906–2005 of 5 States available from the website of Indian Institute of Tropical Meteorology ( www.tropmet.ac.in ) was used in this study. They have considered 94 rain gauge stations well distributed over the region for preparing this data series, one from each of the districts which is the small administrative area and area-weighted mean monthly rainfall of all the meteorological sub-divisions as well as for the whole country by assigning the district area as the weight for each representative rain gauge station. The area, production and productivity of kharif rice over IGR States from 1974–1975 to 2005–2006 was taken from the Directorate of Rice, Ministry of Agriculture, Govt. of India and is available on-line at http://www.dacnet.nic.in . The area, production and productivity of wheat over IGR States from 1966–2007 to 2005–2006 was taken from the Directorate of Wheat Development, Ministry of Agriculture, Govt. of India and is available on-line at http://www.dacnet.in/dwd/ . Normally, in this part of the region, rice season (June–July to October–November) refers to “Kharif” while wheat season (November–December to March) refers to “Rabi”. 3.1.2 Moderate Resolution Imaging Spectroradiometer (MODIS) 1 km NDVI data A monthly composite time series of MODIS 1 km NDVI data (MOD13A3) spanning 2000–2006 for the study region were collected from Land Processes Distributed Active Archive Center (LPDAAC) maintained by US Geological Survey (USGS), NASA at Earth Resources Observation and Science Center (EROS). 3.2 Drought indices 3.2.1 Standardized Precipitation Index (SPI) The SPI has been used by several researchers to quantify the drought ( McKee et al., 1993, 1995; Edwards and McKee, 1997 ) and detailed methodology for SPI computation was described in Subash (2010) . 3.2.2 Rainfall Index (RI) The monthly rainfall during monsoon season was indexed by taking the monthly rainfall in terms of percentage deviation from its mean. The Rainfall Index for any month is expressed as RI i = ( R i - R ) ∗ 100 R where RI i is the monthly Rainfall Index for the ith year, R i is the monthly rainfall for the ith year and R is the mean monthly rainfall. 3.2.3 Kharif Rice Productivity Index (KRPI) and Wheat Productivity Index (WPI) The production of rice depends on type of soil, seeds used, crop area, availability of irrigation facilities, fertilizers, pesticides and also on the government incentives to the farming sector during a year as well as on the meteorological parameters such as rainfall, temperature, relative humidity and solar energy. The non-meteorological parameters i.e., the total technological inputs to the farming sector have been growing steadily and are difficult to quantify. Therefore, to know the pattern of trends and to quantify the growth rate of total technological inputs to the agricultural sector the actual productivity was fitted into an linear as well as any other best fit. To normalize the productivity, another index, the Kharif Rice Productivity Index (KPAI) was taken as the percentage of the technological trend productivity (TP) to the actual productivity. The normalized KRPI for the ith year is KRPI i = ( P i - TP i ) 100 TP i WPI i = ( P i - TP i ) 100 TP i where KRPI i and WPI i are the kharif rice productivity index and wheat productivity indices for the ith year, P i is the actual productivity for the ith year and TP i is the technological trend productivity for the ith year. 3.3 MODIS data processing MODIS data granules for the years 2000–2006 of 12 months for the study area were collected and imported from HDF format into image format and further reprojected using Geographic projection with the aid of Erdas Imagine 8.7 software. All the contiguous data sets were integrated through mosaicking processes in order to get the single file for entire study area for each month of the year. The image containing additional area other than present study has been eliminated using image subset option with reference to the boundary of study area. To obtain optimal classification accuracy, the NDVI data were filtered to correct pixels containing low quality or erroneous values: pixels containing water were removed from the image stack using majority filtering algorithm. The MODIS NDVI values equal to or below zero were assumed to be typically caused by water bodies/snow, and thus, excluded from the data sets. Finally statistical parameters such as mean, median, mode, maximum, minimum, range, standard deviation and CV have been computed for all the five States with the help of zonal statistics module available in Arc GIS 9.1 software. The software was also used to produce thematic maps for each months showing NDVI status. Anomaly NDVI is obtained by subtracting the individual value from the mean of the period 2000–2006. 4 Results and discussion 4.1 Total monsoon seasonal rainfall and average NDVI Considering the total monsoon season rainfall and NDVI patterns for the different States of the IGR for the period 2000–2005, it can be seen ( Fig. 2 ) that their exist increasing correlation between average NDVI and monsoonal rainfall. As rainfall increases from 500 mm, NDVI also increases and reaches up to a range of 0.55 at 1100 mm. However, it can be analyzed that beyond 1100 mm of rainfall NDVI does not show further significant increase. Since most of the rainfall occurs between July and September with a maximum in July, therefore averaging NDVI data for these months fairly represents the growing season for the region ( Anyamba and Tucker, 2005 ). 4.2 Relationship between monthly rainfall and NDVI for different States The monthly distribution of rainfall during monsoon season and corresponding NDVI over different States for the period 2000–2005 is given in Fig. 3 . During the 6 year period, considerable year-to-year variation was noticed between rainfall and NDVI. For Bihar, out of six years, two years (2002 and 2005) received below normal rainfall. No clear trend of any pattern of NDVI and monthly distribution has been noticed for Bihar. In 2005 due to 47.7% deficit of rainfall during June influenced the vegetation conditions during June as well as July even after a surplus rainfall of 4.9% received during July; this may be the reason for negative NDVI anomaly during July month. Similarly, during the year 2000, the continuous deficit of 15.4% and 39.5%, respectively for July and August may be the reason for the negative anomaly of NDVI during July, August and September months even though September received 33.8% surplus rainfall. No clear picture emerged between monthly rainfall and NDVI over West Bengal due to higher rainfall zone of IGR. For Haryana, out of six years, two years received deficit rainfall during June and these two years NDVI values were below normal. In Uttar Pradesh, during the year 2005, due to 31.8% deficit of June rainfall influenced the vegetation conditions during June and July, even though there is a 20% surplus rainfall received during July. But this surplus rainfall helps the vegetation to adjust the deficit rainfall of 35.7% during August. This may be reason for positive anomaly of NDVI during August and September. In Punjab, in the year 2002 all the months except September received deficit rainfall and a negative NDVI anomaly was noticed for these months. But during 2004, even though all the monsoon months received deficit rainfall, only September shows negative NDVI anomaly. Thus even though there was a positive correlation of seasonal monsoon rainfall and average NDVI, conflicting results were noticed in the monthly distribution pattern of rainfall and monthly NDVI anomaly for different States of IGR, indicating a lag time of one month between NDVI and monthly rainfall. These results are in agreement with those reported by Ferrar et al. (1994) for Africa, Li et al. (2004) for China and Chopra (2006) for Gujarat, India. As far as wheat season is concerned, different States responded differently to seasonal monsoon rainfall and NDVI during wheat season (December–March). For Bihar, during the year 2000, due to high rainfall received in September and a little rainfall during October delayed the sowing of wheat and thereby negative NDVI anomaly was noticed in all the months between December and March. But during 2005, even after the monsoon seasonal rainfall deficit, the NDVI showed higher values in all the months. No clear cut relationship between monthly and monsoon season rainfall and NDVI during wheat season has been noticed for Haryana. During the year 2005, a surplus rainfall of 102 per cent during September delayed the sowing of wheat and thereby negative NDVI anomaly was noticed in all the months during wheat season. But during the year 2004, even after deficit rainfall of 70.4% and 84.9% during July and September, respectively, due to good rainfall spells during October, January to March provides sufficient soil moisture led to positive NDVI anomaly for all the months. The same has been noticed for Punjab during 2004. In Punjab, during 2000, a high deficit of 56.8% and 52.4% during August and September, respectively and insufficient rainfall for the rest of the period affected the vegetation conditions and negative NDVI anomalies were noticed in all the months during wheat season. For Uttar Pradesh, ever after a good spell of monsoon, negative NDVI anomalies were noticed in all the months during the year 2000. But during 2005 deficit rainfall (35.7%) during August and absence of any significant rainfall during October to March reduced the vegetation growth and negative NDVI anomalies have been noticed in all the wheat season months except December. In the case of West Bengal during the year 2000, even after good spell of monsoon months, negative NDVI anomaly was noticed during wheat season. During the year 2005, all the monsoon months received deficit rainfall and negative NDVI anomalies have been noticed in all the months of entire wheat season. 4.3 Relationship between SPI and NDVI Relation between monthly SPI during monsoon season and NDVI anomaly of the corresponding month, following month and after two months indicated that NDVI anomaly and SPI shares no correlation ( Fig. 4 ). Large variability in crop physiological conditions exists among the IGR States: for example in June Punjab and Haryana have different rice pheno-phases compared to Bihar and Uttar Pradesh. This may be the reason for the absence of any relationship between SPI and NDVI over IGR. When we analyzed the NDVI and SPI for each State separately ( Fig. 5 ), it was found that there is a positive correlation between NDVI and SPI for Punjab, Haryana and a negative correlation between NDVI and SPI for West Bengal, though these correlations are not statistically significant. But as far as Bihar and Uttar Pradesh are concerned, it is found that there is no relationship between NDVI and SPI. 4.4 Relationship between NDVI anomaly and rice and wheat indices The State-wise Kharif Rice Productivity Indices (KRPI) and monthly NDVI anomaly (dif-NDVI) for monsoon months during the period 2000–2005 are given in Table 2 . It is observed that for Bihar, out of six years, two years (2004 and 2005) fall under deficit rice productivity and in these two years at least one of the monsoon months had established negative NDVI anomaly. But even after three continuous negative NDVI anomalies during July to September, positive KRPI was observed in the year 2000. Similarly, no definite pattern has been observed over the state West Bengal. Three continuous negative NDVI anomalies affected the rice productivity during the years 2000 and 2002. At least two negative NDVI anomaly months were noticed in all the three deficit rice productivity years (2000, 2002 and 2005). As far as Uttar Pradesh is concerned, out of six years, four years (2000, 2002, 2004 and 2005) fall under the category of deficit rice productivity years and in these years except the year 2004 at least one negative NDVI anomaly month has been observed. During 2004, all the months had positive NDVI anomaly. Over Punjab and Haryana, the variation of kharif rice productivity indices never reached ±10. In the case of Punjab, during deficit rice productivity years, at least two of the months had experienced negative NDVI anomaly. At all the IGR States, deficit wheat productivity indices were noticed continuously from the year 2002 to 2005 ( Table 3 ). It is clear that the variability of wheat productivity is more in Bihar compared to all other States. To quantify the relation between NDVI and rice productivity, regression analysis of different combinations of monthly NDVI anomalies during the rice season and KRPI was performed. The same analysis was also done for wheat productivity indices (WPI) and monthly NDVI anomalies during December to March. The regression coefficients are given in Tables 4 and 5 . It is clear that even though none of the combinations is statistically significant, June dif-NDVI contributes more to rice productivity followed by July ( Fig. 6 ). However, the combined effect of June, July and August, explains 15% variation of KPRI. Since the KPRI variation for all the States during the study period is <15%, except 2004 and 2005 for Bihar, this equation can be used as an indicator of regional rice production. As far as wheat is concerned, statistically significant relation was found between WPI and dif NDVI during December–March. This explains 35% of the variability in WPI and in addition to that, all the IGR States during the period 2000–2005 experienced the range of WPI within ±35%. Thus, this equation can be used for predicting/forecasting the wheat productivity over IGR. Among the months, impact of December is more on WPI comparing to all other months ( Fig. 7 ). 4.5 Comparison of SPI, dif-NDVI and KRPI and WPI – a case study during all-India drought year 2002 4.5.1 The salient features of 2002 monsoon over IGR The 2002 monsoon set in on time in most parts of the country except Punjab, Haryana and Western Uttar Pradesh of IGR. The monthly rainfall received during 2002 June to 2003 March over IGR States are given in Table 6 . All the IGR States received less rainfall compared to the normal during June: Uttar Pradesh received −41.9% deviation from the normal followed by Bihar (−33.5%). In July, all the IGR States except West Bengal received lower rainfall compared to normal. Haryana received a rainfall of 30.2 mm only, which is −81.6% deviation from the normal followed by Uttar Pradesh −74.5% deviation from the normal rainfall. Probably, it is for the first time probably in the last 100 years that the month of July received such a low rainfall ( Samra and Singh, 2002 ). Even in the month of August, the situation is not different. Punjab received a deficit rainfall of −35.8% followed by Bihar (−29%). The continuous three months of deficit rainfall caused heavy damage to the transplanted rice crop and withering happened in almost all the States, except West Bengal and Bihar. 4.5.2 Comparison of SPI, MRI, dif NDVI and KRPI – a case study during all-India drought year 2002 The monthly spatial NDVI between June and November during all-India drought year 2002 over IGR is given in Fig. 8 . It indicated that very low NDVI values were observed over Haryana, southern parts of Uttar Pradesh during July, August and September. But normal values were observed during June–July over most parts of Punjab and a very high value of NDVI has been noticed in eastern and northern parts of Punjab. The map also indicates that Bihar and West Bengal under normal or high NDVI during June–July, whereas most parts of West Bengal exhibited very high NDVI during August and September. Except Bihar and West Bengal, the mean NDVI values of 2002 during the rice season were higher than the normal ( Fig. 9 ). The SPI, MRI, dif. NDVI during all-India drought 2002 over the IGR States and corresponding KRPI has been given in Table 7 . This year, except Bihar, all the IGR States fell under the category of deficit rice productivity status. It is noticed that SPI and MRI followed the same pattern for all the States during June–September. It is observed for Punjab and Haryana, during all the months, the NDVI anomaly is negative. Even after continuous deficit rainfall for three months (June–August) causing no effect over the NDVI by the respective months, but affected the NDVI of the month September and thereby a deficit of KRPI. 4.5.3 Comparison of dif. NDVI and WPI – a case study during all-India drought year 2002 Very low NDVI values were noticed over eastern part of West Bengal and western part of Bihar during February–March in the year 2003 and very high values were observed over most parts of Punjab and Haryana during the same period. The rest of the IGR showed normal NDVI ( Fig. 10 ). The NDVI map provides the extent and variability of drought severity within the IGR States. All the States recorded deficit wheat productivity during the drought year, but these values were below −10. Moreover, all the States except Haryana, at least one negative NDVI anomaly have been observed during December to March ( Table 8 ). Therefore, it can be concluded that except Uttar Pradesh during rice season, no other State in Indo-Gangetic region were affected with drought with respect to rice and wheat is concerned. 5 Conclusions The study has explored the possibility of SPI, RI and NDVI for drought assessment of rice–wheat system productivity over IGR-India, which is one of the fertile and densely populated regions of South Asia. As rainfall increases from 500 mm, NDVI increases and reaches up to a range of 0.55 at 1100 mm. However, it can be analyzed that beyond 1100 mm of rainfall NDVI does not show further significant increase. Relationship between monthly SPI during monsoon season and NDVI anomaly of the corresponding month, following month and after two months indicated that NDVI anomaly and SPI shares no correlation. Large variability in crop physiological conditions exists among the IGR States: for example in June Punjab and Haryana have different rice pheno-phases compared to Bihar and Uttar Pradesh. This may be the reason for the absence of any relation between SPI and NDVI over IGR. It is clear that June dif-NDVI contributes more to rice productivity followed by July. However, the combined effect of June, July and August, explains 15% of the variation of KPRI. Since the KPRI variation for all the States during the study period is <15%, except 2004 and 2005 for Bihar, the multiple regression equation can be used as an indicator of regional rice production. As far as wheat is concerned, statistically significant relationship has been observed between WPI and dif-NDVI during December–March. This explains 35% of the variability in WPI and more over all the IGR States during the period 2000–2005 experienced WPI within a range of ±35%. Thus, this equation can be used for predicting/forecasting the wheat productivity over IGR. Among the months, December contributed more control on WPI compared all other individual months. Acknowledgements We are grateful to Indian Council of Agricultural Research for providing sabbatical leave for doing research at Cochin University of Science and Technology as part of the first author’s doctoral study. Also grateful to Head, Department of Atmospheric Sciences for providing necessary computing and other facilities to conduct this study. We are thankful to the two anonymous reviewers and editor-in-chief who have made critical comments and suggestions to improve the manuscript. References Ansari, 2003 Ansari, H., 2003. Monitoring and Zoning of Drought using Fuzzy Logic and GIS. PhD dissertation, Tarbiat Modarres University. Anyamba and Tucker, 2005 A. Anyamba C.J. Tucker Analysis of Sahelian Vegetation Dynamics using NOAA AVHRR NDVI data from 1981 to 2003 J. Arid Environ. 63 3 2005 596 614 Bastiaanssen and Ali, 2003 W.G.M. Bastiaanssen S. Ali A new crop yield forecasting model based on satellite measurements applied across the Indus Basin Pakistan Agric. Eco. Environ. 94 2003 321 340 Bhalme and Mooley, 1979 Bhalme, H.N., Mooley, D.A., 1979. On the performance of modified Palmer index. In: Proceedings of International Symposium: Hydrological Aspects of Droughts, Vol. 1. New Delhi, India, pp. 373–383. Bhuiyan et al., 2006 C. Bhuiyan R.P. Singh F.N. Kogan Monitoring drought dynamics in Aravalli region (India) using different indices based on ground and remote sensing data Int. J. Appl. Earth Obser. Geoinfor. 8 4 2006 289 302 Brown et al., 2002 Brown, J.F., Reed, C.B., Hayes, M.J., Wilhie, D.A., Hubbard, K., 2002. A prototype drought monitoring system integrating climate and satellite data. In: Pecora 15/Land Satellite Information IV/ISPRS Commission I/FIEOS 2002 Conference Proceedings. Chopra, 2006 Chopra, P., 2006. Drought Risk Assessment using Remote Sensing and GIS. A case study of Gujarat. MSc. Dissertation (Unpublished). Edwards and Mckee, 1997 Edwards, D.C., Mckee, T.B., 1997. Characteristics of 20th Century Drought in the United States at Multiple Timescales. Colorado State University, Fort Collins. Climatology Report No. 92-7. Farrar et al., 1994 T.J. Farrar S.E. Nicholson A.R. Lare The influence of soil type on the relationships between NDVI, rainfall, and soil moisture in semiarid Botswana. II. NDVI response to soil moisture Remote Sens. Environ. 50 1994 121 133 Ferrar et al., 1994 T.J. Ferrar S.E. Nicholson A.R. Lare The influence of soil type on the relationships between NDVI, rainfall and soil moisture in semiarid Botswana. II. NDVI response to soil moisture Remote Sens. Environ. 50 1994 121 133 Gibbs and Maher, 1967 Gibbs, W.J., Maher, J.V., 1967. Rainfall Deciles as Drought Indicators, Bureau of Meteorology Bulletin No. 48. Commonwealth of Australia, Melbourne, p. 29. Gill, 1999 Gill, M.A., 1999. Promotion of Resource Conservation Tillage Technologies in South Asia – An Overview of Rice-Wheat Consortium Activities in the Indo-Gangetic Plains Directorate General Agriculture – Water Management, Lahore, Pakistan. Gillette, 1950 H.P. Gillette A creeping drought under way Water Sewage Works 104 1950 105 Hayes and Decker, 1996 M.J. Hayes W.L. Decker Using NOAA AVHRR data to estimate maize production in the United States corn belt Int. J. Remote Sens. 17 1996 3189 3200 Hewitt, 1997 K. Hewitt Regional at Risk. A Geographical Introduction to Disasters 1997 Addison Wesley Longman Limited England Humphreys et al., 2004 Humphreys, E., Thaman, S., Prashar, A., Gajri, P.R., Dhillon, S.S., Yadvinder-Singh., Nayyar, A., Timsina, J., Bijay-Singh., 2004. Productivity, Water Use Efficiency and Hydrology of Wheat on Beds and Flats in Punjab, India. CSIRO Land and Water Technical Report 03/04. Jensen, 1996 J.R. Jensen Introductory digital image processing: a remote sensing perspective 1996 Prentice Hall Upper Saddle River, New Jersey Johnson et al., 1993 G.E. Johnson V.R. Achutuni S. Thiruvengadachari F.N. Kogan The Role of NOAA Satellite Data in Drought Early Warning and Monitoring: Selected Case Studies. Drought Assessment, Management, and Planning: Theory and Case Studies 1993 Kluwer Academic Publishers Justice et al., 1986 C.O. Justice B.N. Holben M.D. Gwynne Monitoring East African vegetation using AVHRR data Int. J. Remote Sens. 7 1986 1453 1474 Kastens et al., 2005 J.H. Kastens T.L. Kastens D.L.A. Kastens K.P. Price E.A. Martinko R.Y. Lee Image masking for crop yield forecasting using AVHRR NDVI time series imagery Remote Sens. Environ. 99 2005 341 356 Kogan, 1997 F.N. Kogan Global drought watch from space Bull. Am. Meteorol. Soc. 78 1997 621 636 Kogan, 2000 Kogan, F.N., 2000. Contribution of Remote Sensing to Drought Early Warning. Early Warning Systems for Drought Preparedness and Drought Management, World Meteorological Organization, Geneva. Kogan, 2001 F.N. Kogan Operational space technology for global vegetation assessment Bull. Am. Meteorol. Soc. 82 9 2001 1949 1964 Ladha et al., 2003 Ladha, J.K., Pathak, H., Tirol-Padre, A., Dawe, D., Gupta, R.K., 2003. Productivity trends in intensive rice wheat cropping systems in Asia. Improving the Productivity and Sustainability of Rice-Wheat Systems: Issues and Impacts. ASA Special Publication 65. Li et al., 2004 J. Li J. Lewis J. Rowland G. Tappan L.L. Tieszen Evaluation of land performance in Senegal using multi-temporal NDVI and rainfall series J. Arid Environ. 59 2004 463 480 Liu and Kogan, 1996 W.T. Liu F.N. Kogan Monitoring regional drought using the vegetation condition index Int. J. Remote Sens. 17 4 1996 2761 2782 Mckee et al., 1993 Mckee, T.B., Doesken, N., Kleist, J., 1993. The relationship of drought frequency and duration to time scales. In: Proceeding of the 8th Conference on Applied Climatology. American Meteorological Society. Mckee et al., 1995 T.B. Mckee N.J. Doesken J. Kleist Drought monitoring with multiple time scales. Ninth Conference on Applied Climatology, Dallas 1995 American Meteorological Society TX Morid et al., 2006 S. Morid V. Smakhtin M. Moghaddasi Comparison of seven meteorological indices for drought monitoring in Iran Int. J. Climatol. 26 2006 971 985 Nguyen and In-Na N Bobee, 1989 V.T.V. Nguyen B. In-Na N Bobee New plotting position formula for Pearson type III distribution J. Hydrol. Eng. 115 6 1989 706 730 Ntale and Gan, 2003 H.K. Ntale T.Y. Gan Drought indices and their application to East Africa Int. J. Climatol. 23 2003 1335 1357 Oza et al., 2002 S.R. Oza J.S. Parihar K. Dadhwal Evaluating Use of Standardized Index for Drought Assessment in the Region of North-West India. TROPMET-2002, 11–12 Feb 2002 Bhubaneswar India Palmer, 1965 Palmer, W.C., 1965. Meteorological Drought, Research Paper No. 45, US Department of Commerce, Weather Bureau, Washington, DC. Patel et al., 2007 N.R. Patel R. Chopra V.K. Dadhwal Analysing spatial patterns of meteorological drought using standardized precipitation index Int. J. Climatol. 14 4 2007 329 336 Prasad et al., 2005 A.K. Prasad L. Chai R.P. Singh M. Kafatos Crop yield estimation model for Iowa using remote sensing and surface parameters Int. J. Appl. Earth Obser. Geoinfor. 8 2005 26 33 Prasad et al., 2007 A.K. Prasad R.P. Singh V. Tare M. Kafatos Use of vegetation index and meteorological parameters for the prediction of crop yield in India Int. J. Remote Sens. 28 23 2007 5207 5235 Rai, 2000 Rai, M., 2000. A Perspective for Developing Action Program for Sustaining Productivity in Rice–Wheat Systems of the Indo-Gangetic Plains. Rice- Wheat Consortium for the Indo-Gangetic Plains. In: Proceedings of the International Workshop on Developing an Action Program for Farm-level Impact in Rice–Wheat Systems of the Indo-Gangetic Plains, 25–27 September 2000, New Delhi, India. Rice–Wheat Consortium Paper Series 14, New Delhi, India: Rice–Wheat Consortium for the Indo-Gangetic Plains. Reed, 1993 B.C. Reed Using remote sensing and Geographic Information Systems for analyzing landscape/drought interaction Int. J. Remote Sens. 14 1993 3489 3503 Rosegrant et al., 2001 M.W. Rosegrant M.S. Paisner S. Meijer J. Witcover Emerging trends and alternative futures 2001 International Food and Policy Research Institute Washington DC, USA Rundquist and Harrington, 2000 B.C. Rundquist J.A. Harrington Jr. The effects of climatic factors on vegetation dynamics of tallgrass and shortgrass cover GeoCarto Int. 15 2000 31 36 Samra and Singh, 2002 J.S. Samra G. Singh Drought management strategies 2002 Indian Council of Agricultural Research New Delhi Shafer and Dezman, 1982 B.A. Shafer L.E. Dezman Development of a Surface Water Supply Index (SWSI) to assess the severity of drought conditions in snowpack runoff areas. In: Proceedings of the Western Snow Conference 1982 Colorado State University Fort Collins, CO pp. 164–175 Smakhtin and Hughes, 2004 V.U. Smakhtin D.A. Hughes Review, Automated Estimation and Analysis of Drought Indices in South Asia. Working paper-83 Drought Series Paper-1, IWMI 2004 Colombo Sri Lanka Singh et al., 2003 R.P. Singh S. Roy F.N. Kogan Vegetation and temperature condition indices from NOAA-AVHRR data for drought monitoring over India Int. J. Remote Sens. 24 22 2003 4393 4402 Subash et al., 2010 Subash, N., 2010. Drought climatology of Indo-Gangetic Region of India using remote sensing and crop growth simulation models. Ph.D Dissertation (Unpublished). Thenkabail et al., 2004 P.S. Thenkabail M.S.D.N. Gamage V.U. Smakhtin The use of remote sensing data for drought assessment and monitoring in Southwest Asia. Research Report 85, IWMI 2004 Colombo Sri Lanka Thiruvengadachari and Gopalkrishna, 1993 S. Thiruvengadachari H.R. Gopalkrishna An integrated PC environment for assessment of drought Int. J. Remote Sens. 14 1993 3201 3208 Ungani and Kogan, 1998 L.S. Ungani F.N. Kogan Drought monitoring and corn yield estimation in southern Africa from AVHRR data Remote Sens. Environ. 63 1998 219 232 Wang, 2000 Wang, J., 2000. Relations between productivity, climate, and Normalized Difference Vegetation Index in the central Great Plains. University of Kansas (Ph.D. Dissertation), Lawrence. Wang et al., 2001 J. Wang K.P. Price P.M. Rich Spatial patterns of NDVI in response to precipitation and temperature in the central Great Plains Int. J. Remote Sens. 22 2001 3827 3844 Wilhite, 2000 Wilhite, D.A., 2000. Drought: A Global Assessment, Hazards and Disasters: A series of Definitive Major Works, vol. 2. Routledge Publishers, London. Wu et al., 2001 H. Wu M.J. Hayes A. Welss Q. Hu An evaluation the standardized precipitation index, the china- z index and the statistical z -score Int. J. Climatol. 21 2001 745 758