Introduction

Vegetation is an important indicator for analyzing and evaluating the impacts of changing climatic variables. The seasonal variation in the climatic variables such as precipitation and temperature and hydrological parameters like soil moisture, terrestrial water storage, and groundwater levels due to climate change may affect the phenology and growth patterns of vegetation. Many studies performed on different continents revealed that the phenological changes are directly connected to climate change (Fuller and Prince 1996; Chmielewski and Rötzer T 2001; Cramer et al. 2001; Bothale and Katpatal 2014; Rishma and Katpatal 2016). Therefore, understanding the relationship between the duration and density of photosynthetically active vegetation cover with climatic parameters is fundamental in predicting responses of the regional ecosystem to climate change at the regional scale (Nicholson et al. 1990). Previous studies have recognized interactive and restraining effects of variation in climate factors on NDVI (Emmett et al. 2019; Balasmeh et al. 2020; Kumar and Roshni 2019; Chu et al. 2019; Ghebrezgabher et al. 2020). Emmett et al. (2019) showed that the rise in precipitation has the strongest positive influence on NDVI in the Great Yellow stone ecosystem. Balasmeh and Karmaker et al. (2020) conducted studies at four sites in various parts of India from 1998 to 2007 and revealed that the local hydrology controls the vegetation dynamics within a very short period.

The study by Kumar and Thendiyath (2019) on the Sone River command in Bihar, India has shown a strong correlation between monthly NDVI and rainfall during the Kahrif season. Chu et al. (2019) analyzed the vegetation dynamics and its response to climate change. This study analyses the NDVI variation in the growing season and three seasons i.e. spring, summer, and autumn. The study has shown that the NDVI is having a positive correlation with temperature and a negative correlation with precipitation during the growing season in some of the regions while semiarid grassland and cropland have shown a positive correlation with precipitation and weekly positive correlation with temperature during the growing season. The study also revealed that vegetation during spring was very sensitive to temperature variations and negatively correlated with precipitation during autumn. Ghebrezgabher et al. (2020) studied NDVI variation in the Horn of Africa by considering the land use and land cover of the area. The results concluded that the NDVI in grassland, savanna, and shrubland is more sensitive to climate change than the NDVI of forest and woodland. This study also analyzed the effect of climate variables and elevation on NDVI within the study area which reflects that temperature decreases with NDVI and precipitation increase with the increase in elevation. Muradyana el al. (2019) studied the NDVI variation and its relationship with precipitation and temperature in mountain ecosystem by considering various classes of vegetation like xerophytic/steep vegetation, meadow vegetation, etc. The study reveals that a negative correlation was observed between NDVI and temperature for xerophytic/steep vegetation and a positive correlation was observed between NDVI and temperature in meadow vegetation. Also, correlations between precipitation and NDVI for all vegetation types were positive.

Remote sensing satellite data has become important for vegetation monitoring. Normalized Difference Vegetation Index (NDVI) is the ratio of reflectance in the red wavelength (600–700 nm) to the reflectance in near-infrared (720–1300 nm) wavelength due to chlorophyll concentration.

NDVI using remotely sensed satellite images is calculated by using (1).

$${\text{NDVI}} = \frac{{{\text{NIR}} - {\text{Red}}}}{{{\text{NIR}} + {\text{Red}}}}$$
(1)

Several vegetation indices are available to study vegetation dynamics. Among them, NDVI and enhanced vegetation index are widely used for vegetation growth patterns (Crist and Cicone 1984).

Moderate Resolution Imaging Spectroradiometer (MODIS) on board the Earth Observing System Terra is a highly advanced platform that has improved sensor characteristics and atmospheric correction. It obtains the signal of vegetation more accurately than an Advanced Very High Resolution Radiometer (AVHRR) (Huete et al. 1999, 2002). MODIS data provides higher spatial resolution of 250 m to 1000 m and spectral resolution of 36 bands. MODIS data has narrower bands in red and near-infrared wavelengths compared to AVHRR. Therefore, MODIS-derived NDVI data is more sensitive to chlorophyll concentration and less sensitive to atmospheric vapor conditions (Huete et al. 2000).

The present study has been conducted in the Upper Bhima sub-basin which has a semi-arid ecosystem with a typical setup of five agro-climatic zones (Fig. 1) with contrasting precipitation and temperatures. Studies in the Upper Bhima sub-basin have been conducted to identify variations in the evapotranspiration and crop coefficients without considering the agro-climatic zones (Londhe and Katpatal 2020).

Fig. 1
figure 1

Location map of Upper Bhima Sub-basin with Agro-climatic zones and Groundwater level observation wells

Full size image

The present study analyses the spatial variation of precipitation, temperature, soil moisture, terrestrial water storage, and groundwater levels from 2003 to 2013 and the impact of this variation on NDVI over the five agro-climatic zones viz. Western Ghat Zone (WGZ), Transition Zone-I (TR-I), Transition Zone-II (TR-II), Water Scarcity Zone (WSZ), and Assured Rainfall Zone (ARZ) in Upper Bhima Sub-basin (Fig. 1). The assessment of the correlations of NDVI with precipitation, temperature, soil moisture, terrestrial water storage, groundwater levels is done by using statistical analysis for correlation coefficient over study area for pre-monsoon, monsoon and post-monsoon period. Generally, NDVI and precipitation have a positive correlation (Kumar and Roshni 2019; Ghebrezgabher et al. 2020) but in the present study, a negative correlation was observed in the WGZ.

Study area and data used

Study area

The Upper Bhima sub-basin is located in Maharashtra state, India (Fig. 1). Bhima River is the major tributary of the Krishna River which is 4th biggest river in terms of water inflow and river basin area in India. The study area lies between latitude 17° 10′ 30″ N to 19° 14′ 30″ and longitude 73° 12’ E to 76° 9’ E (Upper Bhima Sub basin). The Maharashtra state is divided into nine agro-climatic zones based on geographical location, soil types, rainfall patterns, and cropping patterns out of which Upper Bhima Sub-basin covers five zones viz. Western Ghat Zone (WGZ), Transition Zone-I (TR-I), Transition Zone-II (TR-II), Water Scarcity Zone (WSZ) and Assured Rainfall Zone (ARZ) (Government of Maharashtra).

The total area of the Upper Bhima sub-basin is approximately 46,401 km2. The elevation in the sub-basin varies between 160 and 1472 m above mean sea level. The WGZ has the highest elevation between 520 and 1472 m. This causes the formation of a shadow zone area for precipitation in the eastern region of the watershed. The climate of the Upper Bhima sub-basin is very diverse due to the interaction between the monsoon and the Western Ghat mountain range.

Before investigating the dynamics of the vegetation, it is necessary to understand the geological formations within study area. The majority of the study area is covered by the Sahyadri formation’s Deccan Volcanic Basalts, which are composed of various lava flows. These basaltic flows can be divided into two categories: simple flows and compound flows. While compound flows exist at lower heights, simple flows are only present at an elevation of 680 m. Simple basaltic flows are compact, unweathered with lower porosity (CGWB Report on Aquifer Maps and Ground Water Management Plan). Detailed information of climatic conditions, geographic location, cropping pattern and soil type of agro-climatic zones (Agroclimatic Zones in Maharashtra) in Upper Bhima Sub-basin is given in Table 1.

Table 1 Detailed information climatic conditions, geographic location, cropping pattern and soil type of agro-climatic zones
Full size table

Data used

The datasets used in the analysis are explained in this section. The detailed information about the datasets used is illustrated in Table 2.

Table 2 Detailed information about the datasets used in the analysis
Full size table

Precipitation and temperature data

Daily temperature and precipitation data were obtained from Climate Forecast System Reanalysis (CFSR). The CFSR dataset is a 3rd generation reanalysis data product developed by the United States National Centers for Environmental Prediction (NCEP). The CFSR weather data is global high resolution data from coupled atmosphere–ocean-land surface-sea ice system developed by the National Oceanic and Atmospheric Administration (NOAA-NCEP) to provide the best estimate of weather data. The spatial resolution of CFSR weather data is 0.313° which is nearly 38 km. CFSR data gives the maximum and minimum temperatures, precipitation, wind velocity, humidity, and solar radiation data from January 1979 to July 2014 (Saha et al. 2014; Dile and Srinivasan 2014). In this study, the daily data of 38 grid stations covering the study area were collected from January 1, 2003, to December 31, 2013, as this is the common period of availability of the datasets analyzed in this study.

MODIS NDVI data

MODIS is a built-in remote sensing device on Terra and Aqua satellites. The NDVI is an index that is used to measure the health and density of vegetation growth. NDVI value is ranging from − 1 to + 1, where lower values indicate no or sparse vegetation while higher values represent denser or healthier vegetation. NDVI value is estimated based on measured reflectance from the red band (R band) and near-infrared bands (NIR Band) and is defined by (1).

MODIS NDVI data has been used in the analysis which is a standard product and has also been validated by NDVI derived from Standard False Color Composite images of IRS LISS III and LANDSAT 8 data (Ke et al. 2015; Palchaudhuri and Biswas 2020).

In the present study, MOD13Q1 NDVI data product is acquired from the MODIS Terra sensor. This dataset is a Level 3 MODIS product that has 16 days of temporal resolution and 250 m spatial resolution available from February 2000 to the present.

GLDAS soil moisture and terrestrial water storage data

GLDAS is a joint project of NASA’s Goddard Space Flight Center and NOAA. GLDAS is a global terrestrial modeling system that combines satellite data and field station data to generate uniform land surface models to predict real-time data (Rodell et al. 2004). In this study, GLDAS monthly data of the NOAH model for the period from 2003 to 2013 with a spatial resolution of 0.25° is used (Mitchell 2005). GLDAS-NOAH dataset contains a time series of land surface data, which is used to investigate water storage. Consequently, the GLDAS-NOAH model data was used to sort the terrestrial water storage changes in soil moisture to understand the variability of hydrological fluxes which affect the terrestrial water storage.

Groundwater level data

Groundwater data for the study area were collected from the India WRIS WebGIS website. India WRIS Web GIS contains comprehensive and consistent data which gives information about India’s water resources including natural resources in a standardized GIS framework. It has tools to search, access, visualize, understand and analyze the data for assessment, monitoring, planning, development, and finally Integrated Water Resources Management (IWRM). The project has been jointly undertaken by Central Water Commission (CWC), the Ministry of Jal Shakti (MoJS) and National Remote Sensing Centre (NRSC), the Indian Space Research Organization (ISRO), and the Department of Space (DoS) since 2009. In the present study, groundwater level data of 51 observation wells within the watershed is used for January, May, August, and November months for each year during 2003 to 2013.

Digital elevation model (DEM)

The Digital Elevation Model (DEM) is used to analyze the effect of climate variables and elevation on NDVI within the study area. The Shuttle Radar Topography Mission (SRTM) DEM (Mission 2013) with 30 m spatial resolution is used for the analysis. The highest elevation of 1472 m is located in Western Ghat Mountain range and lowest elevation of 160 m is located in the Eastern part.

Data preparation for each agro-climatic zone

The Upper Bhima sub-basin contains 38 weather stations for which the weather data is available from January 1979 to July 2014. For the analysis, it was necessary to sort the datasets for each agro-climatic zone in the watershed. The availability of all the datasets is for different period and hence to avoid the uncertainty in the results, the data for common time period, 2003 to 2013 have been used in the analysis. The details of the total area, number of weather station grid points, and a number of groundwater observation wells in each zone are given in Table 3.

Table 3 Details of agro-climatic zones
Full size table

Materials and methods

Pre-processing of MODIS NDVI data

MOD13Q1 data from February 2000 to December 2013 were downloaded from the Earth data website. Each year comprises 23 granules available for a single tile. The NDVI data layer was extracted from the downloaded MODIS data and re-projected in the ArcGIS environment.

In the data pre-processing stage, NDVI data from the downloaded NDVI images were extracted for the watershed boundary and were then stacked in a single raster file. Level 3 data product provides the data in which the effects of atmospheric elements like cloud cover are minimized. Map algebra statistical tool in ArcGIS has been used for the raster calculation to get the averaged NDVI raster file for the whole watershed.

Statistical analysis

In the study, statistical analysis was carried out for 11 years (2003–2013). Correlation coefficients are calculated for the time series data of NDVI with precipitation, temperature, soil moisture, terrestrial water storage, and groundwater levels.

Pearson correlation coefficients

The linear relationship between the NDVI and selected parameters is calculated using the Pearson correlation coefficient (r) (Kantharao and Rakesh 2018) using (2).

$$r\left( {X, Y} \right) = \frac{{\mathop \sum \nolimits_{i = 1}^{N} (X - \overline{X})(Y - \overline{Y})}}{{\sqrt {\mathop \sum \nolimits_{i = 1}^{N} (X - \overline{X})^{2} \mathop \sum \nolimits_{i = 1}^{N} (Y - \overline{Y})^{2} } }}$$
(2)

The Pearson correlation coefficient ranges from − 1 to + 1 and the positive correlation value indicates that the change in NDVI is in the same direction and vice versa. The correlation coefficient r is dependent on the sample size N and hence it is necessary to perform the hypothesis testing to decide whether the correlation is significant at a 95% significance level. Here, the p value method is used to test the hypothesis. If the p value is less than the significance level of 0.05, then the correlation coefficient is significant at a 95% significance level. The standard range of correlation values has been adopted in the study to indicate correlation between the variables as 0 to 0.3 indicate very weak, 0.3 to 0.5 as weak, 0.5 to 0.7 as moderate and 0.7 to 0.9 as strong (Akoglu 2018).

Geographical detector model (GDM)

The geographical detector model (GDM) is a novel tool that can evaluate the relationship between two variables Y which is the response variable an X which is the environmental factor (Meng et al. 2020). It can also investigate the interaction relationship between two environmental factors to the response variable according to spatial distributions of these datasets without assuming linearity. The GDM gives statistical information regarding the contribution of environmental factors to a response variable. If the spatial distribution between these two variables shows similarity, then that similarity can be measured by the power of determinant (q value) or q statistic expressed in (3).

$$q = 1 - \frac{{\mathop \sum \nolimits_{h = 1}^{L} N_{h} \sigma^{2}_{h} }}{{N\sigma^{2} }}$$
(3)

where N and σ2 are number and the variance of Y values.

The q value ranges between 0 and 1. If the q value equals 0, there is no relation between X and Y, whereas if q equals 1 then Y is strongly connected with X.

The factor detector is used to measure the driving force of each environmental factor X on response variable Y. In this study, response variable (Y) is NDVI and environmental factors (X) are precipitation and terrestrial water storage. If X is numerical, then the first step is to discretize X into different classes. There are many discretization methods such as the equal interval method and the natural breaks method. In this study, manual classification of the environmental factors has been done by considering a maximum and minimum range of a particular parameter. In this study, single-factor influence detection has been carried out to detect the contribution of the explanatory variables to a response variable.

Spatial interpolation

Various algorithms are available to interpolate spatially distributed hydrological and climatic data. The inverse Distance Weighted (IDW) method, which is best for dense and evenly spaced samples, is used for spatial interpolation of precipitation, temperature, and groundwater levels. IDW infers that each station has a local influence which decreases with distance.

Taylor diagram

Taylor diagrams are used to visually show which of several approximations (or models) of a system, process, or event is the most accurate (Taylor 2001). This diagram was invented by Karl E. Taylor in 2001 and it facilitates researchers to compare the models or parameters within the process or system. In this study, the correlation values with respective standard deviation of climatic and hydrological parameters with NDVI in WGZ and WSZ are represented using Taylor Diagrams (Fig. 6) during pre-monsoon, monsoon and post-monsoon period.

Results and discussion

NDVI and its connection with climatic and hydrological parameters

The relationship between average monthly NDVI and precipitation for each zone in the study area is shown as scatterplots in Fig. 2. Monthly precipitation and NDVI data are considered for analyzing the vegetation growth with precipitation.

Fig. 2
figure 2

Response of monthly NDVI with monthly precipitation in each zone in the Upper Bhima Sub-basin

Full size image

NDVI in TR-I, WSZ, and ARZ responds positively to precipitation with R2 values of 0.1736, 0.1871, and 0.1779, respectively, i.e. NDVI is increasing with the increase in precipitation. NDVI in TR-I is not varying with precipitation which has R2 value of 0.0002 i.e. in this zone NDVI does not indicate any connection with precipitation. On the other hand, in WGZ, NDVI values are decreasing with an increase in the precipitation which shows NDVI and precipitation are negatively correlated with an R2 value of 0.2837 (Fig. 2).

Study performed by Bianchi et al. (2020) shows that seasonal variations in NDVI and precipitation are closely connected in Monte biomes which is located in relatively warmer and drier region. On the contrary, in the arid and semiarid biomes east of the Andes, both precipitation and NDVI show weaker annual cycles and less consistent relationships.

The relationship between average monthly NDVI with temperature is shown in Fig. 3. In this study, temperatures data is considered for analyzing the impact of temperature on vegetation growth. NDVI in TR-I, TR-II, WSZ, and ARZ have a negative correlation with temperature having R2 values of 0.1866, 0.3335, 0.3816, and 0.4601, respectively. It is observed that temperature has no effect on NDVI in WGZ with a very negligible R2 value of 9E-06 while in other zones; NDVI shows a negative correlation with rising temperature.

Fig. 3
figure 3

Response of monthly NDVI with monthly temperature in each zone in the Upper Bhima Sub-basin

Full size image

Chen and Weber (2014) in their study concluded that when temperatures were favorable for early growing season i.e. from May to July, the NDVI was well correlated (R2 > 0.5) with precipitation. Correlation between NDVI and precipitation began to decrease in August as temperature increased. Temperatures during the early growing season had a considerable effect on the NDVI which affects the plant growth and density. Precipitation and NDVI may have high correlation in one season than another and the strength of correlations may vary seasonally.

The connection between average monthly NDVI and GLDAS soil moisture is shown in Fig. 4.

Fig. 4
figure 4

Response of monthly NDVI with monthly GLDAS Soil Moisture in each zone in the Upper Bhima Sub-basin

Full size image

Monthly GLDAS soil moisture data has been considered for analyzing the impact of available soil moisture on vegetation growth. NDVI in all the zones is not that much affected by soil moisture. The R2 values between NDVI and soil moisture in WGZ, TR-I, TR-II, WSZ and ARZ are 0.0031, 0.2308, 0.303, 0.1653 and 0.2032, respectively. The R2 values show that the soil moisture in all the zones has very less or no impact on the NDVI. Zhao et al. (2021) utilized the NDVI to derive temperature vegetation draught index (TVDINDVI) to monitor soil moisture. They concluded that some parts of China showed the best correlations between soil moisture and the TVDINDVI during the cold season and some parts of China during the warm season. This meant that the performance of suggested method was different in different circumstances.

The relationship between average monthly NDVI with GLDAS TWS is shown in Fig. 5.

Fig. 5
figure 5

Response of monthly NDVI with monthly GLDAS TWS in each zone in Upper Bhima Sub-basin

Full size image

Monthly GLDAS TWS data has been used for analyzing its connection with vegetation growth. Trend line in all the zones has upward slope which indicates that NDVI has positive correlation with TWS. The R2 values in WGZ, TR-I, TR-II, WSZ, and ARZ are 0.1703, 0.6572, 0.6131, 0.236 and 0.2032. The high R2 values in TR-I and TR-II indicate that TWS and NDVI are moderately correlated in these zones. In WGZ, WSZ, and ARZ, the R2 values shows there is very less or no connection between NDVI and TWS. Andrew et al. (2017) have also shown that variation in terrestrial water storage throughout different time periods has impact on NDVI.

To analyze the provocative results obtained for WGZ from the response of monthly NDVI with precipitation and temperature, Pearson correlation coefficients of NDVI with precipitation, temperature, soil moisture, terrestrial water storage, and groundwater levels of all the agro-climatic zones for pre-monsoon, monsoon, and post-monsoon period are calculated and tabulated in Table 4.

Table 4 Pearson correlation coefficients of NDVI with precipitation, temperature, Soil moisture, Terrestrial Water Storage, and Groundwater levels
Full size table

Since there is very less precipitation during the pre-monsoon period, there is a very weak and negative correlation of NDVI with precipitation. During the monsoon period, NDVI in WGZ shows a higher negative correlation while TR-I shows a weak negative correlation with precipitation. It is also observed that the correlation of NDVI with precipitation is increasing from west to east i.e. from WGZ to ARZ. During the post-monsoon period, all the zones show a positive correlation with precipitation but WGZ shows a weak correlation.

It is also observed that NDVI has a very weak correlation (0.167) with soil moisture in WGZ during monsoon; however a strong correlation is seen in other zones. Correlation coefficients between NDVI and soil moisture in TRI, TR II, WSZ, and ARZ are 0.409, 0.549, 0.655, and 0.74, respectively.

Similar results are observed for the correlation of NDVI with groundwater levels. NDVI has a moderate negative correlation (− 0.53) with groundwater levels in WGZ, while it has a weak to moderate correlation with groundwater levels in all other zones. Though WGZ receives maximum precipitation during monsoon season, NDVI is not showing a strong correlation with precipitation, soil moisture, and groundwater levels. The reason behind these unexpected results in WGZ is due to the lithological formations, rock structures, and mountainous and hilly regions in that zone (Londhe et al. 2022). These factors facilitate lower groundwater storage, higher runoff, and base flow resulting in lower soil moisture. NDVI shows a strong positive correlation with terrestrial water storage in all zones and during all seasons. This indicates that terrestrial water storage is unaffected in WGZ.

To highlight the variations in the correlation values of NDVI with other parameters, Taylor diagrams have been constructed. In the Taylor diagram (Fig. 6), the numbers represent correlations of NDVI with precipitation (Fig. 6a), soil moisture (Fig. 6b), terrestrial water storage (Fig. 6c), and groundwater levels (Fig. 6d) for WGZ and WSZ. In Fig. 6a, point 2 indicates a negative correlation (− 0.38) between NDVI and precipitation during monsoon in WGZ while points 5 and 6 indicate that NDVI growth in WSZ depends on monsoon precipitation in WSZ.

Fig. 6
figure 6

Taylor diagram representing the correlation of NDVI with a precipitation; b soil moisture; c terrestrial water storage and d groundwater levels for WGZ and WSZ

Full size image

The results in Fig. 6b also show that soil moisture in WGZ during the monsoon period (Point 2) has the lowest correlation with NDVI. During pre- monsoon, NDVI has no connection (points 1 and 4) with soil moisture in both WGZ and WSZ. However, NDVI shows a significant correlation with soil moisture in WSZ in the monsoon and post-monsoon periods (Points 5 and 6). The study conducted by Sharma et al. (2021) studied the sensitivity of NDVI to soil moisture, land surface temperature and precipitation. The results show that lowest average NDVI during pre-monsoon season has been recorded due to higher LST and lower soil moisture.

The results in Fig. 6c show that NDVI has a good correlation with terrestrial water storage in WGZ and WSZ in the pre-monsoon, monsoon, and post-monsoon periods. The NDVI in WGZ has almost similar and positive correlation with terrestrial water storage having values 0.578, 0.581, 0.599 during the pre-monsoon, monsoon and post-monsoon period (Point 1, 2 and 3), respectively. NDVI shows a significant positive correlation with terrestrial water storage in WSZ in pre-monsoon (0.795), monsoon (0.724) and post- monsoon (0.697) periods (Points 4, 5 and 6).

There is a negative correlation between NDVI and groundwater levels in WGZ during pre-monsoon and monsoon period as shown in points 1 (− 0.258) and 2 (− 0.53), respectively while it does not have any correlation (0.015) for post-monsoon period (point 3 in Fig. 6d). There is positive correlation between NDVI and groundwater levels in WSZ 1 during pre-monsoon (0.381), monsoon (0.85) and post-monsoon (0.562) period (point 3, 4 and 5, respectively, in Fig. 6d).

Single-factor influence detection on vegetation

The factor detector of GDM was used to determine the driving force of climate factors such as precipitation and terrestrial water storage on the response variable NDVI. In this analysis, the response variable NDVI and climate factors i.e. precipitation and terrestrial water storage are considered for the monsoon and post-monsoon periods in WGZ and WSZ. Factor detector calculates q values in results. A higher q value designates a stronger contribution to the response variable. The q value of the climate factor to response variable NDVI is shown in Table 5.

Table 5 The q value and p value of precipitation and terrestrial water storage for NDVI distribution
Full size table

In WGZ during monsoon and post-monsoon periods, the q values of terrestrial water storage are higher than that of precipitation whereas, in WSZ, q values of precipitation are higher than that of terrestrial water storage which indicates that terrestrial water storage is the dominant factor affecting vegetation in WGZ and precipitation is the dominant factor affecting vegetation in WSZ. Terrestrial water storage and precipitation passed the significance test since all the p values are less than 0.05.

NDVI variation with elevation

The elevation is also an important influencing factor that controls plant growth in the study area. In the study area, precipitation and temperature variation patterns are much more relevant with elevation, which affect directly the availability of soil moisture and NDVI at a regional scale. With rising elevation, the temperature decreases while the soil moisture and precipitation increase.

The DEM of the study area and the variation of NDVI with the elevation in the study area are illustrated in Fig. 7. With the objective to understand the NDVI variation with elevation, the DEM image of the study area is reclassified into 6 classes of elevation and then NDVI values are extracted for those classes in ArcGIS.

Fig. 7
figure 7

Digital Elevation Model and NDVI variation with elevation

Full size image

Results indicate that the 160 m to 400 m elevation interval covering valleys and low-lying areas in WGZ, shows higher NDVI during the pre-monsoon (0.4657) and post-monsoon (0.7685) period and NDVI decreases during the monsoon period (0.3752). The area covered in the elevation range of 400 to 900 m i.e. area in TR-II and WSZ shows lowest NDVI values as this area is a flat Deccan plateau. The area in the elevation range of 900 to 1472 m is mostly mountainous regions and is in TR-I and WGZ. This area shows the highest NDVI values i.e. NDVI shows positive correlation with elevation. Similar results were obtained by Ghebrezgabher et al. (2020) where results showed that NDVI was positively correlated with elevation and precipitation while it was negatively correlated with temperature.

Spatial variation of NDVI and climatic and hydrological parameters for pre-monsoon, monsoon, and post-monsoon periods

The spatial variation of NDVI for pre-monsoon, monsoon, and post-monsoon periods is shown in Fig. 8a. Spatial variation map of NDVI shows lower NDVI values in WGZ during the monsoon period whereas this zone has the highest precipitation values during monsoon (Fig. 8b). On the other hand, the WSZ with less rainfall than WGZ (Fig. 8b), shows better NDVI values which suggest that this ecosystem is fairly dependent on hydrologic changes.

Fig. 8
figure 8

Spatial variation of a NDVI, b precipitation, c temperature, d soil moisture, e Terrestrial water storage, and f Groundwater levels for pre-monsoon, monsoon, and post-monsoon periods

Full size image

The spatial variation of precipitation for pre-monsoon, monsoon, and post- monsoon periods in agro-climatic zones in the study area is shown in Fig. 8b. These precipitation values are calculated using an average of accumulated precipitation during that particular period from 2000 to 2013. As the geographical location of the study area is considered, it receives precipitation mainly because of the Southwestern monsoon. In the study area, monsoon starts in June and lasts till September. So, June, July and August are considered as monsoon period, and September and October are considered as Post-monsoon period. And, before the monsoon, March, April and May months are considered as Pre-monsoon period.

During all seasons, WGZ and TR-I receive the highest amount of precipitation, and WSZ and ARZ receive the lowest amount of precipitation in the study area. In the pre-monsoon period, the southwestern part of the watershed receives the highest rainfall reaching up to 49 mm while the southern part of WSZ receives lowest rainfall. The monsoon period shows dramatic variation in rainfall throughout the study area. In the monsoon period, WGZ receives rainfall of more than 2200 mm, and WSZ and ARZ receive rainfall of less than 300 mm. While in the post-monsoon period, rainfall in WGZ decrease up to 559 mm, and in WSZ and assured rainfall zone, rainfall is 196 mm to 220 mm, respectively. This indicates that the beginning of the monsoon rainfall period in the WSZ is delayed by some days or weeks as compared to WGZ and TR-I. Temperature values are calculated using the average daily temperature (Fig. 8c). During all seasons, WGZ, TR-I and TR-II have the lowest daily temperatures compared to WSZ and ARZ in the study area. During the pre-monsoon period, average daily temperature values range from 27.698 to 32.443 °C. In the monsoon period, daily temperature values range from 22.954 to 27.656 °C, and in the post-monsoon period, the temperature is ranging from 23.391 to 26.443 °C.

GLDAS soil moisture (Fig. 8d) and terrestrial water storage (Fig. 8e) maps are generated by averaging monthly datasets available at the Givovani website from 2003 to 2013. During pre-monsoon, WGZ and TR-I have soil moisture of 9–15 kg/m2 which is the lowest soil moisture as compared to other zones which show 15–26 kg/m2. In the monsoon period, some part of TR-II and WSZ has maximum soil moisture of 36–41 kg/m2. In the post-monsoon period, WSZ and ARZ are showing more soil moisture than WGZ.

In the monsoon and post-monsoon periods, spatial variation of terrestrial water storage is varying according to the precipitation. WGZ has the highest terrestrial water storage of 1450–1900 mm. WSZ has the lowest terrestrial water storage of 1000–1450 mm. While, in the pre-monsoon period, the northern region of the WGZ and TR-I are showing the lowest terrestrial water storage of 740–850 mm. The response of NDVI to climate factors may differ according to variation in type of vegetation, climatic parameters, type of ecosystem and elevation (Ghebrezgabher et al. 2020; Muradyan et al. 2019). Studies show that precipitation had a positive effect on the growth of all vegetation types. Ghebrezgabher et al. (2020) have shown that elevation has a relationship with precipitation, NDVI and temperature. Muradyana et al. (2019) presented the correlation analysis between NDVI and climatic factors showing that temperature had a negative effect on arid and semi-arid vegetation and a positive effect on humid vegetation growth during the entire growing season. However, the results of the present study show that NDVI varies with the precipitation, temperature, soil moisture, terrestrial water storage, groundwater and elevation in study area differently in different agro-climatic zones. It may be proposed that future studies may involve micro-level analysis of NDVI variation with varying geological classes, soil types and changing agricultural practices.

Conclusion

The vegetation dynamics has been studied in five different agro-climatic zones in the Upper Bhima sub-basin from 2003 to 2013. In this research, monthly data of NDVI, precipitation, temperature, GLDAS soil moisture and GLDAS TWS are collected and analyzed. Vegetation dynamics show different variation patterns in each agro-climatic zone of the basin during the pre-monsoon, monsoon, and post-monsoon periods. The scatterplots shows the relationship between monthly NDVI and climatic and hydrological parameters. Results show that NDVI increases with the increase in precipitation and decreases with increase in temperature in all zones except WGZ. The correlation analysis indicates that the vegetation dynamics in TR-I, TR-II, WSZ, and ARZ are positively correlated to the precipitation of the monsoon season since NDVI in these zones increased during the monsoon period. A negative correlation between NDVI and precipitation was observed in the Western Ghat Zone of the study area. The behavior of NDVI in WGZ is exceptional as NDVI decreases with increase in precipitation. In WGZ, NDVI values are decreasing with an increase in the precipitation which shows a negative connection with an R2 value of 0.2837. Similar results were identified in case of connection of GLDAS soil moisture and NDVI. GLDAS soil moisture has no effect on NDVI in WGZ as R2 value is 0.00031 which is very negligible and has very less connection in other zones. In case of GLDAS TWS, all the zones show similar pattern of trend line which indicates that NDVI increases as TWS increases in all the zones.

The Pearson correlation analysis has been carried out for pre-monsoon, monsoon and post-monsoon periods to find out the correlation between NDVI and other parameters considered. During the monsoon period, NDVI in WGZ shows a moderate negative correlation while TR-I shows a weak negative correlation with precipitation. Temperature has a negative correlation or very weak correlation in all the zones in the study area during all the monsoon phases.

Soil moisture has very weak correlation in all the zones during pre-monsoon period and has a moderate positive correlation with NDVI during monsoon and post-monsoon period except in WGZ where the correlation coefficient value is 0.167 during monsoon. In all the zones, NDVI shows strong positive correlation with GLDAS TWS during all monsoon phases. During monsoon and post-monsoon, NDVI shows very less and negative correlation with groundwater levels. The major finding of present study is that WSZ and ARZ largely depend on the monsoon rainfall because the lowest NDVI values occur in the pre-monsoon period but in the monsoon and post-monsoon periods, NDVI values increase up to 0.5 to 0.6 in these zones. Study also shows that WGZ has a typical geological formation which is compact and non-porous basalt, which controls the soil moisture and NDVI values.