Plant growth and development are controlled by internal regulators that are modified according to environmental conditions. Water is one of the several factors that influence the yield and productivity of agricultural crops. It provides soil moisture which is essential for plant growth, without water, nutrients cannot be taken up by the crop and the plants wilt. The geographical limits of foodgrain production are largely controlled by precipitation or temperature. Under natural conditions of cultural practices on dry land, precipitation normally limits production (Newman, 1976). Rainfall occurring over India during summer monsoon season (the major rainy season generally starts in June and ends in September) significantly affects the agricultural production of the country by providing water for both the two main crop growing seasons, kharif (summer) and rabi (winter). It has been reported that the seasonal rainfall accounts for > 50% of the foodgrain production in the country (Krishna Kumar et al., 2004). Variations in the monsoon rainfall affect the total foodgrain yield of India and also the country's economy, which largely depends on agriculture.
Many studies have been carried out for understanding the crop–weather relationship over Indian region (e.g. Parthasarathy and Pant, 1985; Parthasarathy et al., 1992; Kulshrestha, 2002; Selvaraju, 2003; Krishna Kumar et al., 2004) and also on the dependence of the country's economy on agriculture (Kumar and Parikh, 1998; Gadgil et al., 1999). Year-to-year fluctuations in summer monsoon rainfall over India have a strong impact on the variability of aggregate kharif foodgrain production (Parthasarathy et al., 1992; Gadgil, 1996; Webster et al., 1998). Years with deficient and excess monsoon rainfall are associated with low and high production of foodgrain, respectively. But, the negative impact of deficit rainfall is larger than the positive impact of good rainfall (Gadgil and Rupa Kumar, 2006). These studies were mainly focussed on the seasonal rainfall but not on the day to day variation of rainfall on which the growth of the crop mainly depends. Even though crop requires water in all stages of its growth, the occurrence of extreme rainfall events can cause damage to the crop and hence its yield. A reduction in rainfall can cause drought while an increase in the intensity of rainfall and frequent occurrence of heavy rainfall days can lead to flood, but both the drought and flood can cause damage to crop. Spatial variations of the seasonal intensities and frequencies based on various rainfall thresholds clearly show dominance during summer monsoon over the entire country (Revadekar, 2010). Recent studies reported an increase in heavy rainfall activity over central India (Goswami et al., 2006) and this makes it important to study the impact of heavy rainfall activity on the foodgrain yield.
Revadekar and Preethi (2012) attempted to understand the changes in foodgrain yield in response to the occurrence of extremes in daily rainfall. Their study reveals that large part of the country exhibits positive relationship between kharif foodgrain yield and the indices of extreme precipitation. Moreover, the strength and spatial extend of the positive (negative) relationship decreases (increases) with increase in the threshold values of extreme precipitation, indicating that very heavy daily rainfall days are less useful for the foodgrain crops than the moderate rainfall events. However, their analysis is based on some selected threshold values representing moderate, heavy and very heavy rainfall activities. The fact that the foodgrain yield is adversely affected due to reduction in rainfall and occurrence of heavy rainfall activity, point towards the existence of a non-linear relationship between precipitation and foodgrain yield, with different thresholds of rainfall having different impact on the yield. Most of the foodgrain crops are grown during kharif season and are dependent on the summer monsoon rainfall and the day to day variation of rainfall during the season affects the crop growth. Keeping this in view, an attempt is made in this study to examine the role of frequencies of rainy days exceeding all possible range of rainfall thresholds on kharif foodgrain yield over India. Analysis covers all possible values of precipitation thresholds, from 1 mm and above as well as 1st percentile to 99th percentile of all-India summer monsoon rainfall.
2. Data and method of analysis
This study utilizes high resolution (1° × 1° latitude–longitude) gridded daily rainfall dataset for Indian region for the period 1966–2003, developed by Rajeevan et al. (2005, 2006). This dataset is based on rainfall data from 1803 stations that have at least 90% availability and is available for a period of 53 years, from 1951 to 2003 (e.g. Rajeevan et al., 2006). This rainfall data is extensively used for research studies including the analysis of extreme precipitation (e.g. Goswami et al., 2006; Krishnamurthy and Shukla, 2008; Revadekar and Preethi, 2012).
Kharif foodgrain yield over the country for the agricultural years 1966–1967 to 2002–2003 (1 June to 31 May) has been used in the study. This dataset is collected from the publication, ‘Agricultural Statistics At A Glance 2006′ compiled by the Directorate of Economics & Statistics of the Department of Agriculture & Cooperation (DES, 2006), which is a mirror of progress in agriculture at the all-India level as well as across the states. Time series of the foodgrain yield exhibits an increasing trend, mainly due to development of crop varieties, fertilizers, expanded use of high-yielding varieties and changing cropping patterns and agricultural practices (Krishna Kumar et al., 2004). Therefore, the time series of kharif foodgrain yield is detrended linearly to see year to year natural variability in yield. However, the Indian summer monsoon rainfall show significant decreasing trend during recent five decades from 1960 to 2010 (Kripalani and Kulkarni, 2012).
2.2. Method of analysis
Seasonal frequencies and intensities of precipitation are computed using the rainfall dataset at each grid point of India for summer monsoon season (1 June to 30 September) of the period from 1966 to 2003. Frequencies are computed as seasonal count of days when rainfall of a day exceeds certain threshold value. Study covers all possible values of precipitation thresholds. Two types of thresholds are chosen in the study:
(1)Fixed threshold values of precipitation, 1 mm, 2 mm and so on. In this case, same threshold value is used for all the grid points of India to compute the frequencies.
(2)Percentile threshold values, 1st percentile to 99th percentile, at an interval of 1 percentile, where percentiles are computed based on the daily data for a 30 year period, 1961–1990. In this case, the threshold values vary from grid to grid depending on precipitation climatology of that grid box.
For intensity indices, 1-day and 5-day maximum precipitation values from the daily precipitation during the season are also picked up for each year. Further, all-India time series of frequencies and intensities of precipitation have been constructed by computing area average of all the land point grid values of each threshold and intensity indices.
The role of variation in frequencies and intensities of precipitation during summer monsoon season on kharif foodgrain yield over India has been examined using the Karl Pearson correlation method. The correlation analysis has been performed between the kharif foodgrain yield and precipitation indices, ie., frequency indices [based on fixed (1 mm and above) and percentile threshold (1st percentile to 99th percentile) values] and intensity indices (1-day and 5-day maximum precipitation).
3. Impact of summer monsoon precipitation on foodgrain yield
The impact of all-India seasonal rainfall and rainy days on kharif foodgrain yield has been assessed. All-India values of these thresholds are obtained by averaging the indices over all the land points of the subcontinent. Further, detailed analysis of the impact of daily rainfall on the kharif foodgrain yield has been carried out using all-India detrended kharif foodgrain yield and different frequency and intensity indices of rainfall (for all-India as well as indices over each grid point of the country) defined using all possible range of rainfall thresholds that can occur during a day.
3.1. Seasonal precipitation
The relationship between seasonal rainfall and detrended kharif foodgrain yield over India is provided in Figure 1. It can be clearly seen from the figure that all lower (higher) values of seasonal precipitation are mostly associated with negative (positive) anomalies in kharif foodgrain yield. The linear and cubic curve fitted indicate an existence of non-linear relationship between the foodgrain yield and seasonal rainfall. The foodgrain yield increases with increase in seasonal precipitation, however, a reduction in the rate of increase in yield is observed for higher values of seasonal precipitation. For seasonal rainfall between 850 and 1000 mm, the kharif foodgrain shows a mixed response by producing negative and positive anomalies in foodgrain yield. This indicates that during the normal monsoon season, the occurrence of prolonged breaks or heavy rainfall days can cause a reduction in kharif foodgrain yield. Hence, the impact of other factors like number of rainy days, low to heavy rainfall days, has been examined further.
3.2. Seasonal rainy days
Crops require certain minimum amount of water supply continuously during its life period. Frequent occurrence of rainy days with low/moderate intensity of rainfall is more favourable for the growth of foodgrains (Revadekar and Preethi, 2012). Therefore, variation of seasonal foodgrain yield with rainy days has also been examined (Figure 2). A day is considered to be a rainy day when the rainfall of that day exceeds 1 mm. The response of kharif foodgrain to the variations in rainy days appears to be similar to that of seasonal precipitation. All the monsoon seasons having less number of rainy days are associated with negative anomalies in kharif foodgrain yield and all the years with large number of rainy days are associated with positive anomalies in foodgrain yield. Cubic curve fitted for rainy days and kharif foodgrain yield show linear relationship except for the higher number of rainy days. Impact of rainy days on the foodgrain yield seems to be stable when the season gets > 62 rainy days. Thus, the analysis indicates that occurrence of more rainy days can cease the increase in foodgrain yield.
It is clear from the analysis that kharif foodgrain yield increases with increase in seasonal precipitation (rainy days) and the foodgrain yield does not increase rapidly when the seasonal rainfall (rainy days) exceeds certain values. However, during monsoon season, the intensity of rainfall varies considerably from one day to another. The occurrence of rainy days having different intensity of rainfall can have different impact on kharif foodgrain crop and can be reflected on its yield. Hence, from the daily summer monsoon precipitation, all possible range of rainfall thresholds that can occur during a day have been found out and a detailed analysis of the impact of these indices on the seasonal kharif foodgrain yield has been carried out.
3.3. Frequency indices of precipitation
Frequencies, the seasonal count of days when the rainfall of a day exceeds certain threshold value, vary from year to year. Total amount of precipitation received from such frequencies during monsoon season also show year to year variation. Even though there is large spatial variation in the precipitation received over the country, water requirement of a specific crop may not vary much with region to region. However, the response of the crop may vary with different threshold values. Considering the requirement of precipitation for the crop and the large spatial variation of rainfall over the country, two types of thresholds are chosen to compute the frequencies: (1) fixed threshold values of precipitation use same threshold value over all the grid points of India and (2) percentile threshold values, where the threshold values vary from grid to grid depending on precipitation climatology of that grid box.
3.3.1. Fixed thresholds
All integer values from 1 mm and above, at an interval of 1 mm, have been selected as fixed threshold values to understand the impact of frequencies as well as total precipitation associated with that rainfall threshold on kharif foodgrain yield. To obtain an idea of how the frequency of rain events exceeding different threshold values varies over the country, climatological variation (based on the rainfall data for the period from 1961 to 1990) in frequencies corresponding to different fixed threshold values has been computed at each grid point and is averaged over all-India to obtain all-India representative value and is presented in Figure 3. The figure shows that the seasonal count of days with very heavy rainfall events are occasional over Indian land mass. Rainy days with low values of rainfall are more frequent over the country and the frequency decreases exponentially with increase in threshold value.
To understand the impact of the individual precipitation threshold value on the foodgrain yield, time series of all-India frequencies defined as a seasonal count of days for each fixed threshold (1 mm, 2 mm, … is correlated with all-India detrended kharif foodgrain yield (Figure 4). Significant positive correlations (statistically significant at 1% level) are obtained for the seasonal count of days when rainfall of the day exceeds 1 mm. This strong relationship continues up to the frequency for threshold value corresponding to 50 mm rainfall. Further, the magnitudes of correlation coefficient (CC) decrease with increase in the threshold value. For frequencies corresponding to fixed thresholds of 50 to 60 mm, the relationship is weak, but statistically significant at 5% level. Rare, very high precipitation activities show negative correlation, indicating adverse effect of such events on crop yield. Total amount of rainfall received from the days exceeding each threshold also vary from year to year. Therefore, in Figure 4, variation in correlation between the total precipitation received from each frequency index and all-India detrended kharif foodgrain yield is also shown. Strong correlations are seen for amount of rainfall due to threshold values ranging from 1 to 40 mm (significant at 1% level). The correlation values show similar pattern as that of the frequency indices, however, the magnitudes of CC are lower than that obtained from frequency indices.
It is seen from the analysis (Figure 4) that variation in rainfall events when rainfall of the day exceeding 1 to 50 mm has significant positive impact on all-India kharif foodgrain yield. From the threshold value of 50 mm onwards, the impact is reducing. To understand the spatial variability of these impacts, the spatial patterns of CC between kharif foodgrain yield and seasonal rainfall (and frequency) corresponding to two distinct fixed threshold ranges (1) when rainfall of the day is in the range 1–50 mm and (2) when rainfall of the day is > 50 mm are examined (Figure 5). Positive impact of the variation in rainfall obtained from the rainy days of 1 to 50 mm rainfall is seen over large part of the country. Substantial reduction in magnitudes and spatial extent of CC values are seen for the events when daily rainfall is > 50 mm; this reduction is more pronounced over northeastern parts of the country, where the impact turns out to be negative, indicating adverse effect of heavy rainfall events on kharif foodgrain yield. The northeastern region is known for heavy precipitation activities over the country. The figure indicates that the increase in heavy rainfall event may not be useful for good yield of kharif foodgrain. Spatial patterns of correlation between the foodgrain yield and frequency (seasonal count of days) indices are almost similar to that obtained from total rainfall. However, spatial patterns of correlation for frequency indices for < 50 mm rainfall events show slightly higher magnitude and comparatively large spatial extent towards southern India.
3.3.2. Percentile thresholds
Indian region is known for large spatial variations in summer monsoon precipitation. Therefore, rainfall amount corresponding to an extreme rainfall event of a particular place can be a normal event for other regions. The percentile threshold values vary from place to place depending upon the rainfall climatology of that place. Hence, variations in kharif foodgrain yield in response to different percentile thresholds of rainfall have been studied for considering the spatial variation of rainfall over India. Climatological percentile values of rainfall from 1st percentile to 99th percentile for each grid have been computed based on the daily summer monsoon rainfall data from 1961 to 1990. Further, climatological values of all-India precipitation corresponding to each percentile thresholds have been computed and are provided in Figure 6. The figure clearly shows that higher percentile thresholds correspond to higher precipitation amount. Lower percentiles correspond to lower precipitation amount with higher frequencies. So, it is obvious that frequency of rain events decreases (increases) with increase (decrease) in percentile values.
Frequencies as seasonal count of days for each year have been computed as frequency of days exceeding the climatological percentile values from 1st percentile to 99th percentile for each grid and are averaged over the land points of the country to obtain all-India year to year variation in the indices. Impact of such events on kharif foodgrain yield, represented in terms of CCs between the frequency indices and kharif foodgrain yield, is presented in Figure 7. The CCs show strong positive relationship (statistically significant at 1% level) and is almost constant from 1st percentile to 75th percentile. Thereafter, the correlation decreases sharply but remains statistically significant up to 95th percentile. Correlation becomes weak for higher percentile values, above 95th percentile. The figure also shows that the magnitude of CCs is lower for the amount of rainfall received from each threshold as compared to that of corresponding frequencies.
Similar to the analysis based on fixed precipitation threshold values, spatial variation of correlation between total precipitation (and frequency) obtained from two ranges of percentile thresholds, i.e. (1) 1st percentile to 90th percentile; (2) above 90th percentile and kharif foodgrain yield is analysed and is presented in Figure 8. Large spatial extent of significant positive correlation values can be seen for the percentile threshold range from 1st percentile to 90th percentile, while a large reduction in the magnitudes and spatial extent of positive CCs are seen for the rainfall events above 90th percentile values. Northeastern parts of the country show negative values of CC for the rainfall events above 90th percentile. Thus an increase in low/moderate rainfall events (1 to 50 mm or 1st percentile to 90th percentile) appears to be useful for kharif foodgrain yield over large areas of the country, while heavy rainfall events (with > 50 mm or > 90th percentile threshold) can cause adverse effect on the foodgrain especially over northeastern part of the country.
3.4. Intense precipitation indices
Impact of intense rainfall activities in terms of 1-day maximum and 5-day maximum precipitation on kharif foodgrain yield is examined in this section. Figure 9 shows all-India precipitation intensity indices versus kharif foodgrain yield over India. In general, both the 1-day and 5-day maximum precipitation show increase in foodgrain yield with increase in intensity of rainfall. However, it can be noticed that all flood years have above normal yield and all drought years (except one) show below normal foodgrain yield irrespective of the rainfall amount received due to intense rainfall activity. Also, for normal years, both positive and negative anomalies in foodgrain yield are seen. This indicates that the impact of other climatic factors such as temperature, humidity etc. can have significant effect on kharif foodgrain yield. It is interesting to note that most of the normal years are associated with intense rainfall activities in terms of 1-day and 5-day maximum precipitation, comparable with flood years. However, no drought year is associated with intense precipitation activity. Spatial correlation of 1-day maximum and 5-day maximum precipitation with the foodgrain yield (Figure 10) shows that the occurrence of the intense rainfall events has adverse effect on the crops, especially over the northeastern parts of the country. The figure also reveals that the impact of intense rainfall activities increases when the activities prolonged for long time, over northeastern parts of the country, the negative impact increases, however, over other parts of the country, both the positive and negative impacts increase with continuous occurrence of intense precipitation activities.
For comprehensive analysis, variations in foodgrain yield associated with variations in indices of precipitation extremes for two distinct ranges (lower and higher ranges) and are provided in Table 1. The table clearly shows increase in the yield for lower range of indices and decrease (weak positive) for higher range of indices, indicating that low/moderate rainfall events are more useful for the crops than very heavy rainfall events.
Table 1. Variations in all-India kharif foodgrain yield with respect to various indices of precipitation extremes over India during summer monsoon season of the period 1966–2003
Indices of precipitation extremes
Variations in yield (kg/ha/index)
4. Summary and conclusion
Kharif (summer) is the main crop growing season of India. Foodgrain yield during kharif season is directly affected by day to day variations of summer monsoon (June through September) precipitation. An increase (decrease) in kharif foodgrain yield is generally associated with an increase (decrease) in seasonal summer monsoon rainfall and rainy days. A drought in summer monsoon generally leads to large reduction in foodgrain yield. However, the occurrence of heavy rainfall activity can also cause adverse effect on the crop growth, especially over the regions of heavy rainfall activity. Indian summer monsoon has large spatial and intraseasonal variability of rainfall, and the day to day variation of rainfall can have significant impact on the kharif foodgrain yield of the country. This study is therefore aimed at examining the role of all possible ranges of day to day variation in Indian summer monsoon precipitation on kharif foodgrain yield over India for the period 1966–2003.
To understand the impact of daily precipitation, two types of precipitation thresholds, fixed and percentile threshold, have been defined based on all possible values of rainfall that can occur in a day. Considering the fact that foodgrain crop requires same amount of rainfall irrespective of the rainfall occurrence of the place where they are grown, fixed rainfall thresholds have been used and is defined as seasonal count of days when rainfall of the day exceeds 1 mm, 2 mm and so on. These threshold values remain fixed all over India and the frequency of rainy days exceeding the fixed threshold values are averaged over the country to obtain all-India indices. In addition to fixed rainfall thresholds, thresholds based on percentile values of rainfall have been used for considering the spatial variation of rainfall over the country. The percentile value varies from place to place depending upon the climatology of that place. Frequency of rainy days exceeding the climatological rainfall values corresponding to 1st percentile to 99th percentile thresholds has been computed at each grid point of India and is averaged over the country for the monsoon season of each year to obtain its’ year to year variation. CC analysis using all-India averaged percentile (fixed) thresholds with kharif foodgrain yield suggests a strong constant positive relationship (statistically significant 1% level) for fixed (percentile) precipitation threshold exceeding 1 to 50 mm of rainfall (from 1st percentile to 75th percentile value). The magnitudes decreases sharply but remain statistically significant from 75th to 95th percentile. For higher thresholds, above 50 mm fixed (95th percentile) threshold, the magnitude of CC becomes very weak and decreases continuously with increase in threshold value.
Spatial variations in the correlations between kharif foodgrain yield and the precipitation indices have been analysed to obtain the regions which are more affected by day to day variations in rainfall. The analysis reveals that large parts of the country exhibit strong positive correlation with the fixed precipitation thresholds between 1 and 50 mm (percentile thresholds between 1st percentile and 90th percentile). However, the spatial extent of the positive correlations becomes small and insignificant for the fixed thresholds above 50 mm (percentile thresholds above 90th percentile). Large area of northeast India exhibits significant negative correlations for the fixed thresholds above 50 mm (percentile thresholds above 90th percentile), indicating that heavy rainfall events can have adverse effect on the foodgrain crop.
It is interesting to note that most of the normal years (and flood years) are associated with intense rainfall activities in terms of 1-day and 5-day maximum precipitation but drought years are not associated with intense precipitation activity. The study reveals that prolonged occurrence of intense rainfall activity increases their impact on the foodgrain yield. Over heavy rainfall regions of northeastern parts of India, the negative impact increases, however, over other parts of the country, both the positive and negative impacts increase with continuous occurrence of intense precipitation activities.
In summary, an increase (decrease) in the yield/index is generally associated with lower (higher) range of precipitation indices used in the study. The occurrence of positive and negative anomalies of kharif foodgrain yield related to normal (moderate) range of indices, indicates the influence of other factors in determining better yield. Duration of wet or dry spells during the monsoon season and also the role of other meteorological parameters like temperature, humidity etc. needs to be examined thoroughly. It is seen that deficient/drought condition associated with high temperature may lead to low yield; flood conditions associated with very low temperature may lead to moderate yield. Therefore, examination of the combined effect of climate variables may provide better insight on crop–weather relationship.
The authors are thankful to Prof B. N. Goswami, Director, IITM, Pune, for providing the necessary facilities to pursue this study.