Investigation of the hydrological drought behaviour of the Bharathapuzha river basin of Kerala, India, on account of the meteorological drought condition of the watershed was carried out in this study with the aid of four popular drought indices. Statistical relations were established between the indices using polynomial regression models to help in predictions. A study on the decadal spatial variation of hydrological drought behaviour over the basin was also conducted. The cross-correlation study was also performed to understand the propagation of hydrological drought in the event of meteorological drought. The results reveal a slight decrease in the spatial extent of hydrological drought over time. The correlation between meteorological and hydrological drought indices was found to be more assertive at grid locations G1 and G4. According to the findings, the strongest correlation of 0.8 in the propagation of meteorological to hydrological drought was observed at a lag of 3 months over a 12 -month timescale. Run theory reveals that the river basin is subjected to hydrological drought over 54–65% of the months in the historic period. Regression analysis suggests that the cubic model outperforms the other models.

  • Four distinct drought indices were used to identify drought characteristics.

  • Spatiotemporal drought patterns were analysed in a tropical river basin in Kerala, India.

  • The propagation of meteorological to hydrological drought was obtained through a cross-correlation study.

  • Establishes a statistical relation between meteorological and hydrological drought indices.

Graphical Abstract

Graphical Abstract
Graphical Abstract

Drought and water crisis are two of the most pressing, costlier, and damaging issues humanity has faced in recent years (Lindsay et al. 2017). Droughts are classified according to the cause of their occurrence, with meteorological, hydrological, agricultural, and socioeconomic droughts being the most common (Mishra & Singh 2010). The primary reason for the repeating extreme event of drought is reduced precipitation over a long period, ranging from several months to a year (Dai 2011), leading to deficiencies in surface and subsurface water, and affecting agricultural productivity and socio-economic stability. There are several definitions for each of the drought classifications, but there is a lack of unanimity on the definition of drought itself due to its varied nature (Van Loon 2015).

Hydrological droughts are significant among these categories due to the reliance of many anthropogenic activities on surface water resources (Vasiliades et al. 2011). They are also more noticeable than other types of droughts. According to Mishra et al. (2019), hydrological droughts are significant because they directly impact the water supply system, resulting in less agricultural productivity and may lead to famine in the worst-case scenario. As a result, it is critical to investigate the elements that influence the occurrence of hydrological droughts to appropriately handle drought-related calamities (Vidal et al. 2010).

Droughts are monitored using a variety of indicators to alleviate them (Eslamian et al. 2017). The popular meteorological and hydrological indices widely used in India include the Standardised Precipitation Index (SPI), Standardised Evapotranspiration Index (SPEI), Standardised Streamflow Index (SSI), and Streamflow Drought Index (SDI) (Wable et al. 2019). River flows, rainfalls, reservoir levels, soil moisture, groundwater levels, rate of evapotranspiration, and other parameters subsumed in these indices may be used to assess the severity of a drought (McCleskey et al. 2010). Compared to other meteorological indices, probability distribution functions are involved in the computation of SPI and SPEI, which are capable of displaying meteorological drought characteristics and are hence referred to as standard indices for analysing meteorological drought in various parts of the world (Lotfirad et al. 2022). Meteorological droughts occur in India frequently due to climate change (Mishra et al. 2020), and hydrological events are followed by meteorological drought episodes (Wilhite 2000). Drought propagation studies, which provide the time it takes for a meteorological to hydrological drought translation, are therefore critical for water planners (Apurv et al. 2017). Previous research in the subject area has shown an upward trend in temperature and a downward trend in precipitation over time (Rose & Chithra 2020). This complicates analysing hydrological droughts with streamflow as a function of precipitation, evaporation, and temperature (Mazrooei et al. 2015).

Past studies have explored different aspects of drought propagation and forecasting techniques using Cross wavelet analysis (Li et al. 2020a, 2020b), hydrological modeling (Ahmadalipour et al. 2017), and correlation among the drought characteristics between SPI value time series at various time intervals (Edossa et al. 2010). There are studies on building relationships among the various drought indices with a focus on drought characteristics. For instance, Salimi et al. (2021) established a relation between the duration and severity of drought indices using non-linear models.

Bharathapuzha is the largest river basin in Kerala, a once perennial and lifeline river which has experienced drought and water scarcity in recent decades. No other study in this regard on the tropical river basin is available. Also, this is a data-sparse location with considerable damming (11 dams) in a short 209 km river, and building hydrological models is time-consuming and likely to be erroneous. As a result, statistical research is the best alternative. The study was carried out assuming that the current river flow conditions and reservoir operation policies will continue in the future. In light of the preceding literature, the purpose of this research was (1) to understand the hydrological drought condition of the basin, (2) to investigate the drought propagation pattern of the river basin, and (3) to develop a statistical relationship between meteorological and hydrological drought indices. This research will help in the timely management of surface water, groundwater, agriculture, and livelihood. This study contributes to comprehensive drought studies, which help establish a statistical relation between meteorological and hydrological droughts.

Study area and data

This study is conducted in the Bharathapuzha river basin, which runs along India's Kerala-Tamil Nadu border. The river takes its origin at an altitude of 2,250 m above mean sea level in the Anamalai Hills in the state of Tamil Nadu and flows into the Arabian Sea in Ponanni in Kerala. Gayathripuzha, Chitturpuzha, Kalpathipuzha, and Thuthapuzha are the primary tributaries of the Bharathapuzha river. The Bharathapuzha river gets its name from the confluence of the Kalpathipuzha and the Chitturpuzha near Parali. The geographical coordinates of the river basin stretch between latitude 10°26′ to 11°13′ N and longitude 75°53′ to 77°13′ E. The overall length of the river from its source to its mouth is roughly 209 km, and it drains a total area of 6,186 km2, with 71% of it falling within Kerala and the rest in Tamil Nadu. The Bharathapuzha basin, located in the rain shed zone of the Western Ghats, receives a generous amount of rain during the southwest monsoon. The river basin receives an annual average rainfall of 1,132 mm each year. The basin's average annual streamflow is 45.87 cumecs (45.87 m3/s), while the minimum and maximum temperatures are 12.8 and 36 °C respectively. The temperature in the basin varies according to the season as it is located in a tropical region. Various literature describes that the watershed suffers a water-stressed condition (Drissia 2019). This information alludes to the need to investigate the basin's hydrological drought issues.

This study makes use of rainfall, temperature, and streamflow data for a historic period (1988–2005). The meteorological indices SPI and SPEI are calculated from gridded precipitation (0.5°×0.5°) and gridded temperature (1°×1°) data from the Indian Meteorological Department (IMD). The grids are denoted as G1, G2, G3, and G4. The streamflow data provided by the Central Water Commission (CWC) of India is freely available (https://indiawris.gov.in/wris/#/RiverMonitoring) and is used to generate the hydrological indices, SSI and SDI. Ambarampalayam (Amb), Pudur (Pud), Mankara (Man), Pulamanthole (Pul), and Kumbidi (Kum) are the hydrological stations. As the percentage of missing data in the time series is less than 3, a simple linear interpolation approach was used to fill in the gaps. The location of the study area along with the hydrological and meteorological locations in the study area is given in Figure 1. Information on hydrological and meteorological stations is presented in Tables 1 and 2.

Table 1

Information on the hydrological stations

Site nameCatchment area (sq.km)Latitude (N)Longitude (E)Elevation
Ambarampalayam 950 10°37′49″ 76°56′46″ 218 m 
Pudur 1,313 10°46′48″ 76°34′30″ 62 m 
Mankara 2,775 10°45′40″ 76°29′10″ 40 m 
Pulamanthole 940 10°53′57″ 76°11′50″ 8 m 
Kumbidi 5,755 10°51′00″ 76°01′12″ 2 m 
Site nameCatchment area (sq.km)Latitude (N)Longitude (E)Elevation
Ambarampalayam 950 10°37′49″ 76°56′46″ 218 m 
Pudur 1,313 10°46′48″ 76°34′30″ 62 m 
Mankara 2,775 10°45′40″ 76°29′10″ 40 m 
Pulamanthole 940 10°53′57″ 76°11′50″ 8 m 
Kumbidi 5,755 10°51′00″ 76°01′12″ 2 m 
Table 2

Information on the Meteorological stations

GridLatitude (N)Longitude (E)Elevation
G1 10°30′0″ 76°30′0″ 292 m 
G2 10°30′0″ 77°0′0′ 418 m 
G3 11°30′0″ 76°30′0″ 101 m 
G4 11°30′0″ 77°0′0' 397 m 
GridLatitude (N)Longitude (E)Elevation
G1 10°30′0″ 76°30′0″ 292 m 
G2 10°30′0″ 77°0′0′ 418 m 
G3 11°30′0″ 76°30′0″ 101 m 
G4 11°30′0″ 77°0′0' 397 m 
Figure 1

(a) Location map of the study area (b) Meteorological and hydrological locations in the study area.

Figure 1

(a) Location map of the study area (b) Meteorological and hydrological locations in the study area.

Close modal

Estimation of drought indices

McKee et al. (1993) and Vicente-Serrano et al. (2010) established SPI and SPEI, which are two frequently used multi-scalar drought indices. These indexes are primarily used to detect meteorological drought in a given area. Other factors, such as soil moisture at shorter durations (3 months) and hydrological conditions at longer timescales (12 months), can also be captured from these indices. In SPI computation, the probability of observed precipitation is turned into a simple index with precipitation as the single input parameter. The long-term precipitation time series is fitted mostly to a gamma distribution or Pearson type III distribution. It is then transformed to a normal distribution with mean zero and standard deviation of one. The normalised SPI values are seen as standard deviations from the median. Thus, wet and dry periods can be represented in the same way. The standard deviations higher than or less than the median precipitation explain the wet or dry condition respectively. SPEI is computed based on temperature and precipitation data, which assesses the effect of temperature variability in drought. SPEI is calculated using a climatic water balance equation, which takes into account the difference between precipitation and potential evapotranspiration (PET). The Thornthwaite equation is used to calculate PET in this investigation. Rose & Chithra (2020) explains a comparative examination of the meteorological indices for the study area and describes the comprehensive approach adopted.

The computation of both SDI developed by Nalbantis & Tsakiris (2009) and SSI introduced by Modarres (2007) requires only the streamflow values. Furthermore, the approach used by these indices is statistically similar to that used by SPI. SDI computes the cumulative flow rate of a river where, qi, j refers to the total streamflow volumes for the ith hydrological year and the kth timeframe; ‘j’ attributes the months, with values ranging from 1 for September to 12 for August. There are four timeframes: October–December, October–March, October–June, and October–September, with the numbers 1, 2, 3, and 4 respectively. Using the cumulative streamflow volume for each time frame of k in the ith hydrological year, SDI is calculated as follows:
(1)
where and are the mean and standard deviation, respectively, of the cumulative volume of the streamflow during the time frame of k and over a long period. Although SDI is similar to river standardised streamflow volume, this analysis uses mean river flow rate data. SSI has a similar theoretical background as that of SPI. The only difference is that instead of precipitation data, streamflow data are used. The monthly flow value aggregate is used to calculate SSI, whereas SDI is computed for each reference period as mentioned above. Following the actual definitions, SPEI uses a log-Logistic distribution by default, and SPI, SSI, and SDI use a Gamma distribution. However, a detailed analysis of the parameter distributions of the indices can also be investigated, as in the study of Malik et al. (2021). The drought indices used in this study share a common drought classification, as given in Table 3.
Table 3

Drought Classification (McKee et al. 1993)

Index ValueCategory
2.00 or greater Extremely wet 
1.50 to 2.00 Severely wet 
1.00 to 1.50 Moderately wet 
0 to 1.00 Near normal (mildly wet) 
0 to − 1.00 Near normal (mild drought) 
−1.00 to −1.50 Moderate drought 
−1.50 to −2.00 Severe drought 
−2.00 or lower Extreme drought 
Index ValueCategory
2.00 or greater Extremely wet 
1.50 to 2.00 Severely wet 
1.00 to 1.50 Moderately wet 
0 to 1.00 Near normal (mildly wet) 
0 to − 1.00 Near normal (mild drought) 
−1.00 to −1.50 Moderate drought 
−1.50 to −2.00 Severe drought 
−2.00 or lower Extreme drought 

Inverse distance weighting (IDW) interpolation

The rationale behind IDW interpolation is that points near together are more comparable than those farther away. It uses the known values surrounding the prediction point to compute the value of any point that is not quantified. The data points closest to the prediction point have a stronger influence than those further away. As a result, it is assumed that each data point exhibits a localised effect that increases in close proximity. The IDW approach gets its name because it gives more weight to locations closer to the forecast (Katipoğlu et al. 2021).

The equation is as follows:
(2)
(3)
(4)
where, is the prediction point, is the prediction value at , Z(xi) is the value of the sample point at xi, Wi is the IDW of the sample at xi comparison to x0, d is the proximity between the sample point and the prediction point, p is the power parameter, and n is the number of data points to be considered in the computation around the prediction point.

Statistical analysis

The study performs a Pearson correlation analysis and cross-correlation function over the indices and the efficiency of the mathematical relationship between the indices is established by the coefficient of determination (R2).

Cross-correlation function

The cross-correlation function determines how similar a time-series data is to lagged forms of another time-series data as a function of lag. Assume that the time series y1a and y2a have lags l, where l= 0, 1, 2,…. Cross-correlation may be calculated for data pairs (y11, y21), (y12, y22), (y1N, y2N) using sample standard deviations and the lag l cross-covariance as follows:
(5)
(6)
(7)

The lag l cross-covariance, sample standard deviations, and cross-correlation are represented by Cymyn, Sym, and rymyn, respectively.

Regression equations

Polynomial regression is a regression algorithm that models the relationship between the regressor x and the regressand y as an nth degree polynomial in x. Linear, quadratic and cubic regressions are commonly used polynomial regressions. Linear regression is a linear model that assumes a linear relationship between the input variable (x) and the output variable (y). It is of the form, y=a+bx+e, where a and b are regression coefficients and e is the error variable, an unobserved random variable that is assumed to be independently and identically distributed with zero mean and constant variance. Quadratic regression and cubic regression are of the form, y=a+bx+cx2+e, and y=a+bx+cx2+dx3+e, respectively, where a, b, c, and d are regression coefficients and e is the error variable.

Spatial variability of hydrological drought in the study area

As seen in Figure 2, the lighter shade denotes an extreme drought situation. Over the years, a slight decrease in the severity of hydrological drought is found both during the summer and monsoon months of the year. The dry eastern part of the basin gets wetter over the period. Compared to the monsoon months of the years, the summer period experiences a slightly higher drought severity, which explains the seasonal drought behavior of the river basin. Comparing the summer and monsoon months, it is evident that the downstream shows a wetter state during the monsoon, whereas it is relatively dry during the summer months. This demonstrates the impact of human interference, particularly the influence of a dam on the flow pattern. The overall root mean squared error (RMSE) statistic obtained from the cross-validation procedure is 0.61 for SDI and 0.39 for SSI.

Figure 2

Spatial plot of SSI and SDI with the decadal variation of indices in summer (March) and Monsoon (July) during 1988–2005.

Figure 2

Spatial plot of SSI and SDI with the decadal variation of indices in summer (March) and Monsoon (July) during 1988–2005.

Close modal

Meteorological and hydrological drought correlation

Figure 3 depicts the Pearson correlation (r) between SPI-SSI, SPI-SDI, SPEI-SSI, and SPEI-SDI throughout time steps of 1–12 months. A measure of strength and direction of association between continuous variables is obtained here. The value of r ranges from −1 to 1. A value of 1 indicates a perfect positive correlation stating that a hydrological drought will occur on the occurrence of a meteorological drought event, whereas a value of −1 indicates a perfect negative correlation among the indices. Positive and negative signs imply the direction of relation: positive values denote that both variables move in the same direction. Negative values explain that they move in the opposite direction (when one increases, the other decreases or vice versa). At G1, the correlation between all the meteorological and hydrological indices increases with the timescale, where the highest correlation is observed at the 12-month timescale. It is found that all the stations other than Ambarampalayam show an association of around 0.7 at G1. In G2, the association of SPI and SPEI with SDI decreases with timescale whereas a slight increase with SSI is observed. However, a correlation of below 0.6 with all the hydrological indices is observed. At G3, the correlation between meteorological and hydrological indices is low (r-value below 0.5) at all the hydrological stations and decreases with timescale. All the hydrological stations show a better correlation with G4, ranging from 0.61 to 0.73 at a 12-month timescale. Overall, G1 and G4 are highly associated with the hydrological stations.

Figure 3

1–12-month Pearson correlation coefficients between indices at (a) G1 (b) G2 (c) G3 (d) G4.

Figure 3

1–12-month Pearson correlation coefficients between indices at (a) G1 (b) G2 (c) G3 (d) G4.

Close modal

Analysis of drought propagation

The highest cross-correlation observed among the meteorological and hydrological drought indices in the river basin is depicted in Figure 4 with a lag of up to 7 months, which is summarised in Table 4. Based on the results, the highest correlation obtained at Ambarampalayam is at a 3-month lag with a value of 0.8 at a 12-month timescale between the SPI and SSI indices. This explains that a hydrological drought will occur at Ambarampalayam with a 3-month lag period after the occurrence of a meteorological event at G4. Pudur shows the highest association of 0.74 with G4 with a zero-lag period at a 12-month timescale. Mankara, Pulamanthole, and Kumbidi are highly associated with G1 with values of 0.73, 0.8, and 0.71 respectively at a zero-lag period in a 12-month timescale. A zero-lag period explains that the hydrological drought is reflected immediately after the occurrence of a meteorological event. The temporal variations of high cross-over correlated meteorological and hydrological indices can be seen in Figure 5.

Table 4

Cross-correlation among the indices (ccf = cross-correlation function)

Meteorological stationsHydrological Stations
AmbPudManPulKum
G1 Highest ccf 0.51 0.72 0.73 0.80 0.71 
Associated indices SPEI, SDI SPEI, SDI SPEI, SDI SPEI, SDI SPEI, SDI 
Lag (months) 
Timescale (month) 12 12 12 12 12 
G2 Highest ccf 0.17 0.32 0.37 0.51 0.48 
Associated indices SPEI, SDI SPEI, SDI SPEI, SDI SPEI, SDI SPI, SDI 
Lag (months) 
Timescale (month) 
G3 Highest ccf 0.2 0.3 0.3 0.4 0.5 
Associated indices SPEI, SDI SPEI, SDI SPEI, SDI SPEI, SDI SPEI, SDI 
Lag (months) 
Timescale (month) 12 
G4 Highest ccf 0.80 0.74 0.68 0.68 0.66 
Associated indices SPI, SSI SPEI, SSI SPEI, SSI SPEI, SSI SPEI, SSI 
Lag (months) 
Timescale (month) 12 12 12 12 12 
Meteorological stationsHydrological Stations
AmbPudManPulKum
G1 Highest ccf 0.51 0.72 0.73 0.80 0.71 
Associated indices SPEI, SDI SPEI, SDI SPEI, SDI SPEI, SDI SPEI, SDI 
Lag (months) 
Timescale (month) 12 12 12 12 12 
G2 Highest ccf 0.17 0.32 0.37 0.51 0.48 
Associated indices SPEI, SDI SPEI, SDI SPEI, SDI SPEI, SDI SPI, SDI 
Lag (months) 
Timescale (month) 
G3 Highest ccf 0.2 0.3 0.3 0.4 0.5 
Associated indices SPEI, SDI SPEI, SDI SPEI, SDI SPEI, SDI SPEI, SDI 
Lag (months) 
Timescale (month) 12 
G4 Highest ccf 0.80 0.74 0.68 0.68 0.66 
Associated indices SPI, SSI SPEI, SSI SPEI, SSI SPEI, SSI SPEI, SSI 
Lag (months) 
Timescale (month) 12 12 12 12 12 
Figure 4

Crossover correlation among the highly correlated combinations (a) G4 vs Ambarampalayam (b) G1 vs Pudur (c) G1 vs Mankara (d) G1 vs Pulamanthole (e) G1 vs Kumbidi.

Figure 4

Crossover correlation among the highly correlated combinations (a) G4 vs Ambarampalayam (b) G1 vs Pudur (c) G1 vs Mankara (d) G1 vs Pulamanthole (e) G1 vs Kumbidi.

Close modal
Figure 5

Temporal variation of indices with the highest cross-correlation (a) Ambarampalayam (b) Pudur (c) Mankara (d) Pulamanthole (e) Kumbidi.

Figure 5

Temporal variation of indices with the highest cross-correlation (a) Ambarampalayam (b) Pudur (c) Mankara (d) Pulamanthole (e) Kumbidi.

Close modal

Drought characteristics using run theory

Based on run theory, Table 5 explains the highest drought severity and corresponding duration (months) experienced in each meteorological and hydrological location. It is found that the meteorological drought at G1 and G4 with the highest severity was observed between 2000 and 2004, whereas those at G2 and G3 occurred between 1988 and 1992. The severity of these events in the basin ranges from 40.17 to 67.43 with a duration spanning between 37 and 68 months. It is also noted that the river basin experienced an overall meteorological drought event between 54 and 65% of the months during the historical period 1988–2005. The most severe hydrological droughts occurred between the years 1988–1992 and 1999–2005, with magnitudes ranging from 40.5 to 85.53 and lasting 50 to 70 months. In the basin, an overall hydrological drought month accounted for 53% to 65% of the time.

Table 5

Highest drought severity and corresponding duration (months) experienced in each meteorological and hydrological location based on run theory

LocationSeverity > 30periodCorresponding Duration (months)Total Drought duration (months)Drought months (%)
Amb SDI 12 41.48 Sep 1988–Oct 1992 50 135 64.90 
Amb SDI 12 40.50 Jul 2000–Jan 2005 55 135 64.90 
Pud SDI 12 85.53 Dec 1999–Aug 2005 69 124 59.62 
Man SDI 12 72.15 Oct 2000–Aug 2005 62 135 64.90 
Pul SDI 12 65.42 Jun 2000–Aug 2005 63 110 52.88 
Kum SDI 12 73.72 Sep 1999–Jun 2005 70 126 60.58 
G1 SPEI 12 67.43 May 2000–Dec 2005 68 116 55.77 
G2 SPEI 12 40.17 Sep 1988–Sep 1991 37 116 55.77 
G3 SPEI 12 63.60 Sep 1988–Jul 1992 47 135 64.90 
G4 SPEI 12 49.69 May 2000–Oct 2004 54 113 54.33 
LocationSeverity > 30periodCorresponding Duration (months)Total Drought duration (months)Drought months (%)
Amb SDI 12 41.48 Sep 1988–Oct 1992 50 135 64.90 
Amb SDI 12 40.50 Jul 2000–Jan 2005 55 135 64.90 
Pud SDI 12 85.53 Dec 1999–Aug 2005 69 124 59.62 
Man SDI 12 72.15 Oct 2000–Aug 2005 62 135 64.90 
Pul SDI 12 65.42 Jun 2000–Aug 2005 63 110 52.88 
Kum SDI 12 73.72 Sep 1999–Jun 2005 70 126 60.58 
G1 SPEI 12 67.43 May 2000–Dec 2005 68 116 55.77 
G2 SPEI 12 40.17 Sep 1988–Sep 1991 37 116 55.77 
G3 SPEI 12 63.60 Sep 1988–Jul 1992 47 135 64.90 
G4 SPEI 12 49.69 May 2000–Oct 2004 54 113 54.33 

Statistical interrelation of indices

In this study, linear, quadratic, and cubic polynomial regression models were fitted to pairs of indices (SPI-SSI, SPI-SDI, SPEI-SSI, SPEI-SDI) to establish statistical relations where the dependent variables - being the hydrological indices (y) - and the meteorological variables - being the independent variable (x) - were modeled as an nth degree polynomial in x. Better R2 values were obtained for the cubic model, followed by quadratic and linear models. The cubic and quadratic models performed significantly better compared to the linear model. R2 is calculated and tabulated in Table 6. All the R2 values greater than 0.5 are given in the table. It is found that the highest R2 value is obtained at Pulamanthole station against the G1 location between 12-month SPEI-SSI and SPEI-SDI and with a value of 0.65. Kumbidi, Pulamanthole, Mankara, and Pudur show an R2 value ranging between 0.52 and 0.65 with the G1 location, and a value ranging between 0.52 and 0.55 is obtained for G4 location with Ambarampalayam, Pudur, and Pulamanthole stations.

Table 6

Statistical relations between the indices

Meteorological stationHydrological stationIndexesRelationsR2
G1 Pud SPEI-12, SSI-12 SSI = −0.07 + 0.76 SPEI 0.521 
SSI = −0.12 + 0.77 SPEI + 0.05 SPEI2 0.523 
SSI = −0.12 + 0.79 SPEI + 0.05 SPEI2 − 0.01 SPEI3 0.523 
SPEI-12, SDI-12 SDI = −0.11 + 0.77 SPEI 0.519 
SDI = −0.15 + 0.77 SPI + 0.05 SPEI2 0.521 
SDI = −0.15 + 0.80 SPEI + 0.04 SPEI2 − 0.02 SPEI3 0.521 
Man SPI-12, SSI-12 SSI = −0.06 + 0.99 SPI + 0.07 SPI2 − 0.11 SPI3 0.508 
SPEI-12, SSI-12 SSI = 0.02 + 0.8 SPEI 0.53 
SSI = −0.04 + 0.80 SPEI + 0.07 SPEI2 0.534 
SSI = −0.03 + 0.92 SPEI + 0.06 SPEI2 − 0.06 SPEI3 0.538 
SPI-12, SDI-12 SDI = −0.1 + 0.99 SPI + 0.07 SPI2 − 0.11 SPI3 0.505 
SPEI-12, SDI-9 SDI = 0.8 SPEI 0.523 
SDI = −0.08 + 0.8 SPEI + 0.09 SPEI2 0.528 
SDI = −0.08 + 0.86 SPEI + 0.08 SPEI2 − 0.03 SPEI3 0.529 
SPEI-12, SDI-12 SDI = −0.02 + 0.8 SPEI 0.529 
SDI = −0.09 + 0.81 SPEI + 0.07 SPEI2 0.533 
SDI = −0.08 + 0.93 SPEI + 0.06 SPEI2 − 0.06 SPEI3 0.537 
Pul SPI-12, SSI-12 SSI = 0.04 + 0.66 SPI 0.568 
SSI = 0.08 + 0.65 SPI − 0.04 SPI2 0.57 
SSI = 0.1 + 0.8 SPI − 0.06 SPI2 − 0.07 SPI3 0.58 
SPEI-12, SSI-12 SSI = 0.06 + 0.71 SPEI 0.631 
SSI = 0.15 + 0.70 SPEI − 0.11 SPEI2 0.644 
SSI = 0.61 + 0.72 SPEI − 0.11 SPEI2 − 0.01 SPEI3 0.645 
SPI-6, SDI-6 SDI = 0.08 + 0.67 SPI − 0.11 SPI2 0.507 
SDI = 0.08 + 0.68 SPI − 0.11 SPI2−0.01 SPI3 0.508 
SPI-9, SDI-9 SDI = −0.01 + 0.68 SPI 0.533 
SDI = 0.10 + 0.68 SPI − 0.12 SPI2 0.553 
SDI = 0.11 + 0.61 SPI − 0.14 SPI2 + 0.03 SPI3 0.555 
SPI-12, SDI-12 SDI = −0.01 + 0.68 SPI 0.569 
SDI = 0.03 + 0.68 SPI − 0.04 SPI2 0.572 
SDI = 0.05 + 0.81 SPI − 0.06 SPI2 − 0.06 SPI3 0.579 
SPEI-6, SDI-6 SDI = 0.7 SPEI 0.547 
SDI = 0.14 + 0.7 SPEI − 0.16 SPEI2 0.572 
SDI = 0.13 + 0.59 SPEI − 0.15 SPEI2 − 0.06 SPEI3 0.576 
SPEI-9, SDI-9 SDI = 0.72 SPEI 0.585 
SDI = 0.15 + 0.71 SPEI − 0.16 SPEI2 0.611 
SDI = 0.16 + 0.48 SPEI − 0.18 SPEI2+0.13 SPEI3 0.631 
SPEI-12, SDI-12 SDI = 0.74 SPEI 0.634 
SDI = 0.1 + 0.73 SPEI − 0.11 SPEI2 0.645 
SDI = 0.1 + 0.74 SPEI − 0.11 SPEI2 − 0.01 SPEI3 0.645 
Kum SPEI-12, SSI-12 SSI = −0.06 + 0.61 SPEI 0.507 
SSI = 0.04 + 0.6 SPEI − 0.11 SPEI2 0.522 
SSI = 0.04 + 0.61 SPEI − 0.11 SPEI2 + 0.004 SPEI3 0.522 
SPEI-9, SDI-9 SDI = 0.03 + 0.44 SPEI − 0.16 SPEI2 + 0.1 SPEI3 0.512 
SPEI-12, SDI-12 SDI = −0.12 + 0.64 SPEI 0.508 
SDI = 0.62 − 0.12 SPEI + 0.62 SPEI2 0.523 
SDI = −0.02 + 0.62 SPEI − 0.12 SPEI2 + 0 SPEI3 0.523 
G4 Amb SPI-12, SSI-12 SSI = 0.08 + 0.89 SPI 0.524 
SSI = 0.07 + 0.88 SPI + 0.01 SPI2 0.524 
SSI = 0.05 + 0.96 SPI + 0.07 SPI2 − 0.03 SPI3 0.526 
SPEI-12, SSI-12 SSI = −0.07 + 0.87 SPEI + 0.17 SPEI2 0.519 
SSI = −0.08 + 0.92 SPEI + 0.18 SPEI2 − 0.02 SPEI3 0.519 
SPI-12, SDI-12 SDI = 0.05 + 0.88 SPI 0.528 
SDI = 0.04 + 0.87 SPI + 0.01 SPI2 0.528 
SDI = 0.02 + 0.95 SPI + 0.06 SPI2 − 0.03 SPI3 0.53 
SPEI-12, SDI-12 SDI = −0.10 + 0.87 SPEI + 0.17 SPEI2 0.522 
SDI = −0.10 + 0.91 SPEI + 0.17 SPEI2 − 0.02 SPEI3 0.523 
Pud SPI-12, SSI-12 SSI = −0.12 + 0.77 SPI 0.53 
SSI = −0.11 + 0.86 SPI − 0.11 SPI2 0.552 
SSI = −0.03 + 0.97 SPI − 0.04 SPI2 − 0.04 SPI3 0.556 
SPEI-12, SSI-12 SSI = −0.12 + 0.81 SPEI 0.554 
SSI = −0.1 + 0.81 SPEI − 0.02 SPEI2 0.554 
SSI = −0.1 + 0.83 SPEI − 0.01 SPEI2 − 0.01 SPEI3 0.555 
SPI-9, SDI-9 SDI = −0.1 + 0.97 SPI − 0.04 SPI2 − 0.77 SPI3 0.504 
SPI-12, SDI-12 SDI = −0.15 + 0.77 SPI 0.526 
SDI = −0.03 + 0.87 SPI − 0.12 SPI2 0.552 
SDI = −0.06 + 0.99 SPI − 0.04 SPI2 − 0.05 SPI3 0.556 
SPEI-9, SDI-9 SDI = −0.13 + 0.77 SPEI 0.503 
SDI = −0.14 + 0.77 SPEI + 0.01 SPEI2 0.503 
SDI = −0.14 + 0.83 SPEI + 0.02 SPEI2 − 0.03 SPEI3 0.504 
SPEI-12, SDI-12 SDI = −0.15 + 0.81 SPEI 0.553 
SDI = −0.13 + 0.82 SPEI − 0.02 SPEI2 0.554 
SDI = −0.13 + 0.86 SPEI − 0.02 SPEI2 − 0.02 SPEI3 0.555 
Pul SPEI-12, SSI-12 SSI = 0.13 + 0.31 SPEI − 0.18 SPEI2 − 0.14 SPEI3 0.52 
SDI = 0.08 + 0.35 SPEI − 0.18 SPEI2 + 0.14 SPEI3 0.515 
Meteorological stationHydrological stationIndexesRelationsR2
G1 Pud SPEI-12, SSI-12 SSI = −0.07 + 0.76 SPEI 0.521 
SSI = −0.12 + 0.77 SPEI + 0.05 SPEI2 0.523 
SSI = −0.12 + 0.79 SPEI + 0.05 SPEI2 − 0.01 SPEI3 0.523 
SPEI-12, SDI-12 SDI = −0.11 + 0.77 SPEI 0.519 
SDI = −0.15 + 0.77 SPI + 0.05 SPEI2 0.521 
SDI = −0.15 + 0.80 SPEI + 0.04 SPEI2 − 0.02 SPEI3 0.521 
Man SPI-12, SSI-12 SSI = −0.06 + 0.99 SPI + 0.07 SPI2 − 0.11 SPI3 0.508 
SPEI-12, SSI-12 SSI = 0.02 + 0.8 SPEI 0.53 
SSI = −0.04 + 0.80 SPEI + 0.07 SPEI2 0.534 
SSI = −0.03 + 0.92 SPEI + 0.06 SPEI2 − 0.06 SPEI3 0.538 
SPI-12, SDI-12 SDI = −0.1 + 0.99 SPI + 0.07 SPI2 − 0.11 SPI3 0.505 
SPEI-12, SDI-9 SDI = 0.8 SPEI 0.523 
SDI = −0.08 + 0.8 SPEI + 0.09 SPEI2 0.528 
SDI = −0.08 + 0.86 SPEI + 0.08 SPEI2 − 0.03 SPEI3 0.529 
SPEI-12, SDI-12 SDI = −0.02 + 0.8 SPEI 0.529 
SDI = −0.09 + 0.81 SPEI + 0.07 SPEI2 0.533 
SDI = −0.08 + 0.93 SPEI + 0.06 SPEI2 − 0.06 SPEI3 0.537 
Pul SPI-12, SSI-12 SSI = 0.04 + 0.66 SPI 0.568 
SSI = 0.08 + 0.65 SPI − 0.04 SPI2 0.57 
SSI = 0.1 + 0.8 SPI − 0.06 SPI2 − 0.07 SPI3 0.58 
SPEI-12, SSI-12 SSI = 0.06 + 0.71 SPEI 0.631 
SSI = 0.15 + 0.70 SPEI − 0.11 SPEI2 0.644 
SSI = 0.61 + 0.72 SPEI − 0.11 SPEI2 − 0.01 SPEI3 0.645 
SPI-6, SDI-6 SDI = 0.08 + 0.67 SPI − 0.11 SPI2 0.507 
SDI = 0.08 + 0.68 SPI − 0.11 SPI2−0.01 SPI3 0.508 
SPI-9, SDI-9 SDI = −0.01 + 0.68 SPI 0.533 
SDI = 0.10 + 0.68 SPI − 0.12 SPI2 0.553 
SDI = 0.11 + 0.61 SPI − 0.14 SPI2 + 0.03 SPI3 0.555 
SPI-12, SDI-12 SDI = −0.01 + 0.68 SPI 0.569 
SDI = 0.03 + 0.68 SPI − 0.04 SPI2 0.572 
SDI = 0.05 + 0.81 SPI − 0.06 SPI2 − 0.06 SPI3 0.579 
SPEI-6, SDI-6 SDI = 0.7 SPEI 0.547 
SDI = 0.14 + 0.7 SPEI − 0.16 SPEI2 0.572 
SDI = 0.13 + 0.59 SPEI − 0.15 SPEI2 − 0.06 SPEI3 0.576 
SPEI-9, SDI-9 SDI = 0.72 SPEI 0.585 
SDI = 0.15 + 0.71 SPEI − 0.16 SPEI2 0.611 
SDI = 0.16 + 0.48 SPEI − 0.18 SPEI2+0.13 SPEI3 0.631 
SPEI-12, SDI-12 SDI = 0.74 SPEI 0.634 
SDI = 0.1 + 0.73 SPEI − 0.11 SPEI2 0.645 
SDI = 0.1 + 0.74 SPEI − 0.11 SPEI2 − 0.01 SPEI3 0.645 
Kum SPEI-12, SSI-12 SSI = −0.06 + 0.61 SPEI 0.507 
SSI = 0.04 + 0.6 SPEI − 0.11 SPEI2 0.522 
SSI = 0.04 + 0.61 SPEI − 0.11 SPEI2 + 0.004 SPEI3 0.522 
SPEI-9, SDI-9 SDI = 0.03 + 0.44 SPEI − 0.16 SPEI2 + 0.1 SPEI3 0.512 
SPEI-12, SDI-12 SDI = −0.12 + 0.64 SPEI 0.508 
SDI = 0.62 − 0.12 SPEI + 0.62 SPEI2 0.523 
SDI = −0.02 + 0.62 SPEI − 0.12 SPEI2 + 0 SPEI3 0.523 
G4 Amb SPI-12, SSI-12 SSI = 0.08 + 0.89 SPI 0.524 
SSI = 0.07 + 0.88 SPI + 0.01 SPI2 0.524 
SSI = 0.05 + 0.96 SPI + 0.07 SPI2 − 0.03 SPI3 0.526 
SPEI-12, SSI-12 SSI = −0.07 + 0.87 SPEI + 0.17 SPEI2 0.519 
SSI = −0.08 + 0.92 SPEI + 0.18 SPEI2 − 0.02 SPEI3 0.519 
SPI-12, SDI-12 SDI = 0.05 + 0.88 SPI 0.528 
SDI = 0.04 + 0.87 SPI + 0.01 SPI2 0.528 
SDI = 0.02 + 0.95 SPI + 0.06 SPI2 − 0.03 SPI3 0.53 
SPEI-12, SDI-12 SDI = −0.10 + 0.87 SPEI + 0.17 SPEI2 0.522 
SDI = −0.10 + 0.91 SPEI + 0.17 SPEI2 − 0.02 SPEI3 0.523 
Pud SPI-12, SSI-12 SSI = −0.12 + 0.77 SPI 0.53 
SSI = −0.11 + 0.86 SPI − 0.11 SPI2 0.552 
SSI = −0.03 + 0.97 SPI − 0.04 SPI2 − 0.04 SPI3 0.556 
SPEI-12, SSI-12 SSI = −0.12 + 0.81 SPEI 0.554 
SSI = −0.1 + 0.81 SPEI − 0.02 SPEI2 0.554 
SSI = −0.1 + 0.83 SPEI − 0.01 SPEI2 − 0.01 SPEI3 0.555 
SPI-9, SDI-9 SDI = −0.1 + 0.97 SPI − 0.04 SPI2 − 0.77 SPI3 0.504 
SPI-12, SDI-12 SDI = −0.15 + 0.77 SPI 0.526 
SDI = −0.03 + 0.87 SPI − 0.12 SPI2 0.552 
SDI = −0.06 + 0.99 SPI − 0.04 SPI2 − 0.05 SPI3 0.556 
SPEI-9, SDI-9 SDI = −0.13 + 0.77 SPEI 0.503 
SDI = −0.14 + 0.77 SPEI + 0.01 SPEI2 0.503 
SDI = −0.14 + 0.83 SPEI + 0.02 SPEI2 − 0.03 SPEI3 0.504 
SPEI-12, SDI-12 SDI = −0.15 + 0.81 SPEI 0.553 
SDI = −0.13 + 0.82 SPEI − 0.02 SPEI2 0.554 
SDI = −0.13 + 0.86 SPEI − 0.02 SPEI2 − 0.02 SPEI3 0.555 
Pul SPEI-12, SSI-12 SSI = 0.13 + 0.31 SPEI − 0.18 SPEI2 − 0.14 SPEI3 0.52 
SDI = 0.08 + 0.35 SPEI − 0.18 SPEI2 + 0.14 SPEI3 0.515 

On determining the meteorological drought in the study area, the statistical relationship established among the hydro-meteorological indices helps anticipate a hydrological drought ahead of its occurrence. The insights found in this study could be an essential step in addressing drought susceptibility and help drought management measures for mitigation. Identifying regional vulnerabilities allows water planners and managers to formulate suitable mitigation strategies such as meticulous planning of water resources, changes in water-dependent industries, revamping of irrigation systems, and assisting decision-makers in considering drought from a hazard viewpoint and incorporating the notion of drought vulnerability into natural resource planning.

Once a perennial tropical river, the Bharathapuzha has become water-stressed in recent decades. This accentuates the necessity of monitoring for hydrological drought in the river basin. This study has utilised the capabilities of drought indices to understand the temporal and spatial extent of hydrological drought. The propagation of meteorological to hydrological droughts was also assessed in the study. The study assessed hydrological and meteorological drought events in the basin. During the summer and monsoon months of the year, decadal variation revealed a slight reduction in hydrological drought. According to the data analysed, droughts were most severe from 1988 to 1992 and 1999 to 2005. The highest severity observed was at Pudur, and the most extended drought duration observed was 70 months. The longest overall drought in the river basin lasted 64.90% of its observed period, encompassing mild to extreme drought occurrences. According to the cross-over correlation study, the meteorological stations that influence the hydrological stations are highly related on a 12-month timescale.

Furthermore, as per the literature, meteorological droughts are strongly linked to hydrological droughts over 12-month periods (Li et al. 2020a). A better response in the propagation of meteorological to hydrological drought was found within a lag period of zero to three months. The research shows that the SPI-SSI and SPEI-SSI results are nearly identical, with minor differences where SPEI values are relatively higher. A similar observation is made with SPI-SDI and SPEI-SDI. The inclusion of potential evapotranspiration in SPEI may have resulted in a slight rise in values. The relation between meteorological and hydrological drought was established using nonlinear polynomial models, of which cubic models gave the highest coefficient of determination, followed by quadratic and linear models. Integrating other drought triggering components of the hydrological cycle and sophisticated advanced techniques would give more insight into these research dimensions.

All relevant data are available from the streamflow data provided by the Central Water Commission (CWC) is freely available (see: https://indiawris.gov.in/wris/#/RiverMonitoring). The precipitation and temperature data provided by the Indian Meteorological Department, Pune, is also freely available (see: https://www.imdpune.gov.in/Clim_Pred_LRF_New/Grided_Data_Download.html).

Apurv
T.
,
Sivapalan
M.
&
Cai
X.
2017
Understanding the role of climate characteristics in drought propagation
.
Water Resources Research
53
(
11
),
9304
9329
.
Drissia
T. K.
2019
Spatial and temporal variation of water stress in Bharathapuzha River basin, Kerala, India
.
Journal of The Institution of Engineers (India): Series A
100
(
1
),
167
175
.
Edossa
D. C.
,
Babel
M. S.
&
Gupta
A. D.
2010
Drought analysis in the Awash River basin, Ethiopia
.
Water Resources Management
24
(
7
),
1441
1460
.
Eslamian
S.
,
Ostad-Ali-Askari
K.
,
Singh
V. P.
,
Dalezios
N. R.
,
Ghane
M.
,
Yihdego
Y.
&
Matouq
M.
2017
A review of drought indices
.
International Journal of Constructive Research in Civil Engineering
3
,
48
66
.
Li
R.
,
Chen
N.
,
Zhang
X.
,
Zeng
L.
,
Wang
X.
,
Tang
S.
&
Niyogi
D.
2020b
Quantitative analysis of agricultural drought propagation process in the Yangtze River Basin by using cross wavelet analysis and spatial autocorrelation
.
Agricultural and Forest Meteorology
280
,
107809
.
Lindsay
J.
,
Dean
A. J.
&
Supski
S.
2017
Responding to the Millennium drought: comparing domestic water cultures in three Australian cities
.
Regional Environmental Change
17
(
2
),
565
577
.
Lotfirad
M.
,
Esmaeili-Gisavandani
H.
&
Adib
A.
2022
Drought monitoring and prediction using SPI, SPEI, and random forest model in various climates of Iran
.
Journal of Water and Climate Change
13
(
2
),
383
406
.
Malik
A.
,
Kumar
A.
,
Salih
S. Q.
&
Yaseen
Z. M.
2021
Hydrological drought investigation using streamflow drought index
. In:
Intelligent Data Analytics for Decision-Support Systems in Hazard Mitigation
(Deo, R., Samui, P., Kisi, O. & Yaseen, Z. (eds))
,
Springer
,
Singapore
, pp.
63
88
. See:
https://doi.org/10.1007/978-981-15-5772-9_4
.
Mazrooei
A.
,
Sinha
T.
,
Sankarasubramanian
A.
,
Kumar
S.
&
Peters-Lidard
C. D.
2015
Decomposition of sources of errors in seasonal streamflow forecasting over the US Sunbelt
.
Journal of Geophysical Research: Atmospheres
120
(
23
),
11
809
.
McCleskey
R. B.
,
Nordstrom
D. K.
,
Susong
D. D.
,
Ball
J. W.
&
Holloway
J. M.
2010
Source and fate of inorganic solutes in the Gibbon River, Yellowstone National Park, Wyoming, USA: I. Low-flow discharge and major solute chemistry
.
Journal of Volcanology and Geothermal Research
193
(
3–4
),
189
202
.
Mckee
T. B.
,
Doesken
N. J.
&
Leist
J.
1993
The relationship of drought frequency and duration to time scales
. In:
Preprints 8th Conference on Applied Climatology
. Vol.
17
, pp.
179
184
.
Mishra
A. K.
&
Singh
V. P.
2010
A review of drought concepts
.
Journal of Hydrology
391
(
1–2
),
202
216
.
Mishra
V.
,
Tiwari
A. D.
,
Aadhar
S.
,
Shah
R.
,
Xiao
M.
,
Pai
D. S.
&
Lettenmaier
D.
2019
Drought and famine in India, 1870–2016
.
Geophysical Research Letters
46
(
4
),
2075
2083
.
Mishra
V.
,
Thirumalai
K.
,
Singh
D.
&
Aadhar
S.
2020
Future exacerbation of hot and dry summer monsoon extremes in India
.
npj Climate and Atmospheric Science
3
(
1
),
1
9
.
Modarres
R.
2007
Streamflow drought time series forecasting
.
Stochastic Environmental Research and Risk Assessment
21
(
3
),
223
233
.
doi:10.1007/s00477-006-0058-1
.
Nalbantis
I.
&
Tsakiris
G.
2009
Assessment of hydrological drought revisited
.
Water Resources Management
23
(
5
),
881
897
.
Rose
J.
&
Chithra
N. R.
2020
Evaluation of temporal drought variation and projection in a tropical river basin of Kerala
.
Journal of Water and Climate Change
11
(
S1
),
115
132
.
Salimi
H.
,
Asadi
E.
&
Darbandi
S.
2021
Meteorological and hydrological drought monitoring using several drought indices
.
Applied Water Science
11
(
2
),
1
10
.
Van Loon
A. F.
2015
Hydrological drought explained
.
Wiley Interdisciplinary Reviews: Water
2
(
4
),
359
392
.
Vasiliades
L.
,
Loukas
A.
&
Liberis
N.
2011
A water balance derived drought index for Pinios River Basin, Greece
.
Water Resources Management
25
(
4
),
1087
1101
.
Vicente-Serrano
S. M.
,
Beguería
S.
&
López-Moreno
J. I.
2010
A multiscalar drought index sensitive to global warming: the standardized precipitation evapotranspiration index
.
Journal of Climate
23
(
7
),
1696
1718
.
Vidal
J. P.
,
Martin
E.
,
Franchistéguy
L.
,
Habets
F.
,
Soubeyroux
J. M.
,
Blanchard
M.
&
Baillon
M.
2010
Multilevel and multiscale drought reanalysis over France with the Safran-Isba-Modcou hydrometeorological suite
.
Hydrology and Earth System Sciences
14
(
3
),
459
478
.
Wable
P. S.
,
Jha
M. K.
&
Shekhar
A.
2019
Comparison of drought indices in a semi-arid river basin of India
.
Water Resources Management
33
(
1
),
75
102
.
Wilhite
D. A.
2000
Drought as a natural hazard: concepts and definitions
. In:
Drought: A Global Assessment
,
Wilhite
D. A.
(ed.), Vol.
I
,
chap. 1
.
Routledge
,
London
, pp.
3
18
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/).