Volume 45, Issue 9 p. 4337-4344
Research Letter
Free Access

Moisture Supply From the Western Ghats Forests to Water Deficit East Coast of India

Supantha Paul

Supantha Paul

Interdisciplinary Program in Climate Studies, Indian Institute of Technology Bombay, Mumbai, India

Search for more papers by this author
Subimal Ghosh

Corresponding Author

Subimal Ghosh

Interdisciplinary Program in Climate Studies, Indian Institute of Technology Bombay, Mumbai, India

Department of Civil Engineering, Indian Institute of Technology Bombay, Mumbai, India

Correspondence to: S. Ghosh,

[email protected]

Search for more papers by this author
K. Rajendran

K. Rajendran

CSIR–Fourth Paradigm Institute (CSIR–4PI), Bangalore, India

Search for more papers by this author
Raghu Murtugudde

Raghu Murtugudde

Department of Civil Engineering, Indian Institute of Technology Bombay, Mumbai, India

Earth System Science Interdisciplinary Centre/DOAS, University of Maryland, College Park, MD, USA

Search for more papers by this author
First published: 30 April 2018
Citations: 31

Abstract

The mountainous western coast of India, known as the Western Ghats, is considered to be a biodiversity hot spot, but it is under a constant threat due to human activities. The region is characterized by high orographic monsoon precipitation resulting in dense vegetation cover. Feedback of such a dense vegetation on the southwest monsoon rainfall is not yet explored. Here we perform regional climate simulations with the Weather Research and Forecasting model and find that evapotranspiration from the vegetation of Western Ghats contributes 25–40% of the southwest monsoon rainfall over the water-deficit state of Tamil Nadu. This contribution reaches 50% during deficit monsoon years or dry spells within a season. Our findings suggest that recent deforestation in this area will affect not only the biodiversity of the region but also the water availability over Peninsular India, which is already impacted by water scarcity.

Key Points

  • Forests of West Coast in India supply 25–40% (average) of moisture to the southwest monsoon rainfall over the east coast
  • The contributions from the forests of Western Ghats to the monsoon rainfall over the East Coast reach 50% during dry periods
  • The urgent need to stop the deforestation of WG not only to retain biodiversity but also to maintain the water cycle over Peninsular India is emphasized

Plain Language Summary

Forests over the Western Ghats (WG) region at the west coast of India are suffering from severe deforestation. We find that the vegetation over the WG region contributes moisture to the precipitation of the water-deficit state of Tamil Nadu and it reaches as high as 40% in many of the regions. Tamil Nadu is at present is under severe water crisis due to interstate water sharing and related controversies. We emphasize the urgent need of enforcing strict laws to stop the deforestation of WG not only to retain bio-diversity but also to maintain the water cycle over these semiarid parts of Peninsular India.

1 Introduction

The Western Ghats (WG) runs parallel to the entire western coast of the Indian peninsula and extends across the states of Maharashtra, Karnataka, Goa, Kerala, and Tamil Nadu. The mountain range generates very high orographic rainfall since it is almost perpendicular to the moisture-laden Southwest monsoon winds. This bountiful rain supports a thick vegetation and forest cover over this region. The WG are also considered one of the top eight hottest hot spots for biodiversity in the world (Myers et al., 2000) with over 400 biological species and seven distinct vegetation types (Utkarsh et al., 1998). Diversity of plant species is closely related to the number of rainy days along the mountain range (Gadgil, 1996). The length of the dry season, human interference, and the ecological resilience of the landscape (Pascal et al., 2004) influence the distribution of primary vegetation types in the WG. This region also has one of the richest evergreen forests with a number of distinct species. However, during recent periods, there has been a gradual but significant loss of forest cover (Reddy et al., 2016). In the district of Uttara Kannada in Karnataka, forest cover has declined by 50% (Gadgil, 1996), while the evergreen forest in Kerala is currently facing extinction (Ramesh et al., 1997) with the rate of deforestation accelerating during recent decades (Chattopadhyay, 1985; Ramesh et al., 1997; Reddy et al., 2016).

Such a rich vegetation cover can be expected to interact with the atmospheric moisture loading through evapotranspiration (ET) and potentially play an important role in monsoon precipitation downstream. However, such interactions have not been reported yet to the best of our knowledge. Here we attempt to quantify the contributions from the WG vegetation and forest cover to the southwest monsoon rainfall (rainfall during June to September). Recent studies (Pathak et al., 2017) have argued that the Ganga Basin is one of the major land moisture sources for monsoon rainfall. The Ganga Basin gets much attention due to its huge size and intense agricultural activities. In comparison, the area of WG is quite small and hence its impacts on monsoon rainfall over the WG itself and Peninsular India has not attracted any attention. Other studies (Devaraju, Bala, & Modak, 2015; Devaraju, Bala, & Nemani, 2015; Quesada et al., 2017) show significant impacts of deforestation on monsoon rainfall, but none of them really have focused on the biodiversity hot spot of India, that is, the forests of WG. Global studies show that vegetation plays an important role (Bonan, 2008) in generating precipitation over the Amazon forest. During the dry season over the Amazon, the forest cover results in a lower albedo, higher net radiation, and greater ET compared to pastureland (da Rocha et al., 2004; von Randow et al., 2004). These factors are found to help in forming a shallow, cool, and moist boundary layer affecting precipitation. Medvigy et al. (2013) report that complete deforestation of the Amazon will lead to 10%–20% precipitation reduction in the remote coastal northwest United States. Lawrence and Vandecar (2014) have shown that beyond a critical deforestation threshold (clearing forests beyond 30–50%), rainfall declines irrespective of the type of forest cover. Reduction in forest cover over the tropical band covering 18.75°S—15°N was found to decrease precipitation over the Amazon by 138 mm/year with increase in temperatures (Bathiany et al., 2010). Boisier et al. (2012) and de Noblet-Ducoudré et al. (2012) studied the role of different representations of land use land cover (LULC) maps from multiple sources on the uncertainties in model simulations of heat fluxes. And yet, there is a dearth of studies on the role of WG vegetation in the precipitation variability during the Indian summer monsoon. Here we present the first such study. We perform numerical simulations of the Indian monsoon with the Weather Research and Forecasting (WRF) model coupled to the Community Land Model (CLM). Two simulations are performed: one with and one without WG vegetation. The results are compared to understand the role of vegetation.

2 Method

In the present study, we employ the WRF model (Skamarock et al., 2008) to simulate the Indian Summer Monsoon Rainfall (ISMR) and to perform sensitivity experiments to understand the impacts of WG. WRF is considered the next-generation mesoscale numerical weather prediction system. WRF is coupled with the state-of-the-art CLM version 4 (Oleson et al., 2010) to analyze land surface feedback processes and impacts of LULC change on regional climate (Paul et al., 2016; Pielke et al., 2002).

In each grid cell, the land surface is represented by five primary subgrid land cover types (glacier, lake, wetland, urban, and vegetated). The vegetation is represented with plant functional types (PFT), each with a distinct leaf area index, stem area index, and canopy height. Each subgrid land cover type and PFT patch is considered as a separate column for energy and water calculations (see the supporting information for further details).

The coupled WRF-CLM prescribes static geographical information and dynamic atmospheric variables as boundary conditions. Static geographic data set generally includes LULC information, vegetation fraction, orography, and soil properties. This data set is global and is provided with the model (WRF) but can be updated for a region of interest. Atmospheric variables specified as boundary conditions at lateral boundaries include pressure and surface level data at 6-hourly intervals and are obtained from ERA (European Reanalysis) data. The model domain is presented in Figure 1a. The horizontal resolution chosen to be 30 km, which is sufficient to resolve the WG reasonably well (Srinivas et al., 2013). The model domain has 178 grid points in the east-west direction and 193 grid points in the north-south direction with 30 levels in the vertical. The number of grid cells is 1,470 for peninsular India and 321 for WG region. A spectral nudging technique is applied in both zonal and meridional directions for winds (U and V) and temperature (T) with a predefined wave numbers for U, V, and T at each level (Miguez-Macho et al., 2004) above the planetary boundary layer to retain the large-scale features while allowing regional model to develop its own small-scale atmospheric features within the planetary boundary layer. Earlier studies (Devanand et al., 2017; Paul et al., 2016) have suggested that such a nudging improves simulations of monsoon precipitation.

Details are in the caption following the image
(a) The domain for the WRF simulations used for this study. (b) The MODIS land cover classification. (c) Western Ghats with modified land cover (no vegetation) is also shown.

The land cover data that provide the details of vegetation over the WG are obtained from Moderate Resolution Imaging Spectroradiometer (MODIS) ) for the year 2005. The major vegetation types over the WG as obtained from MODIS are evergreen broadleaf forests and woody savannas (Figure 1b). Gridded fractions of different PFTs considered for the regional simulations are shown in Figure S1. Here we perform two simulations: (1) a control simulation with the prescribed vegetation over the WG and (2) a sensitivity experiment with no vegetation over the WG (Figure 1c). Both the simulations are performed for the monsoon season (June to September) during 1990–2015. For each of the seasons, a one-month spin-up is performed, which is considered sufficient for atmospheric adjustments under imposed lateral boundary conditions (Giorgi & Mearns, 1999). For example, a 20-day spin-up time was used for the seasonal simulations of Indian monsoon rainfall performed by Srinivas et al. (2013). Further, the premonsoon rainfall is very low compared to the monsoon seasonal rainfall. Hence, the contribution from per-monsoon rainfall to the monsoon precipitation that comes through recharge of soil moisture and subsequent transpiration from vegetation (with low leaf area due to dry premonsoon condition) is assumed to be small. The differences between the two model experiments are analyzed below to quantify the impact of WG vegetation cover on ISMR. Details of coupled WRF-CLM (Bonan et al., 2002; Dudhia, 1989; Hong et al., 2004, 2006; Jiménez et al., 2011; Kain, 2004; Mlawer et al., 1997; Monin & Obukhov, 1954; Oleson et al., 2010; Subin et al., 2011) with different parameterization schemes are discussed in the supporting information.

3 Results and Discussion

We first evaluate the control run with respect to the observed gridded rainfall data at 0.25° spatial resolution obtained from India Meteorological Department (Pai et al., 2014). The evaluation is performed for entire India (Figure S2, first column), Peninsular India including WG (Figure S2, second column) and the WG alone (Figure S2, third column). The model faithfully reproduces precipitation climatology for all the three regions, but there is a slight underestimation over WG, which may be because of the inadequate model resolution to capture the details of the highly varying orography. We further test the ability of the model to capture the spatial patterns. The spatial probability density functions of precipitation are plotted for observed and simulated precipitation for all the three regions (Figures S2d–S2f). The patterns are well captured although the peaks are underestimated. This also may be attributed to the scales of topography (Hariprasad et al., 2011), which are finer than model resolution and hence are not fully resolved. The spatial correlations Figures S2g–S2i) offer confidence for using the model for diagnosing the impacts of WG vegetation on ISMR.

We present (Figure 2) the differences between the modeled precipitation over Peninsular India from the sensitivity simulation with no vegetation over WG and control simulation (i.e., with prescribed WG vegetation). This presents the likely impacts of deforestation over WG on the monsoon precipitation over peninsular India. We find a significant drop in rainfall (statistically significant at 0.05 level) when vegetation is removed with values ranging from 1 to 2.5 mm/day (Figure 2a). This amounts to an average of around 25% of the total over the southern Peninsular India (state of Tamil Nadu shown in Figure 2d). Tamil Nadu receives an annual mean rainfall of 960 mm (average of years 1951–2000; data are obtained from Indian Meteorological Department), of which 33% occurs during June to September. During the Kharif season (agricultural season starting during June), Tamil Nadu receives very low rainfall and that significantly affects the agricultural activities. Hence, the moisture contribution from the vegetation of WG plays a very important role in meeting the water demand of Tamil Nadu during the southwest monsoon season.

Details are in the caption following the image
(a–f) Differences between the simulated precipitation as obtained without vegetation (simulation without vegetation) and with vegetation over Western Ghats (control simulation). The top row presents the values, and bottom row presents the same as percentages. The columns represent location of Tamil Nadu (first column), seasonal rainfall (second column), rainfall during dry spells (third column), and wet spells (fourth column).

We further present the likely impacts of deforestation over the WG on the rainfall over Peninsular India during the wet and dry spells of summer monsoon months (June-to-September). Our definition for dry spell over Peninsular India is very similar to that used for defining break period over Central India by Rajeevan et al. (2010). A 5-day moving average of the daily rainfall time series (only for summer monsoon) over the peninsular region is prepared and then standardized by subtracting the climatological mean and dividing by the climatological standard deviation. Dry spells are defined as the periods during which standardized rainfall anomaly is less than −1.0 for at least three consecutive days. Similarly, wet spells are defined as the periods during which standardized rainfall anomaly is greater than +1.0 for at least three consecutive days. We find that during the dry spells the impacts of deforestation over the WG on the rainfall over Tamil Nadu is typically higher as compared to that during wet spells (Figures 2b and 2c). This makes the moisture contribution of forests of WG to rainfall over Tamil Nadu critically important and needs to be maintained for agricultural sustainability specifically over the Kharif season. During the dry spells, the decline in rainfall due to deforestation over WG reaches as high as 50% of the total (Figure 2e) and this is evident over the entire state. However, during the wet spells, the contribution is mostly over the southern part of the state with a magnitude of around 25–30% (Figure 2f) of the total. A putative mechanism behind such a variability in contribution may be attributed to the high rainfall during the wet spells over WG which gets evapotranspirated during the dry spells when there is low cloud cover and high radiation (Figure S3). A fraction of the moisture gets transported by the prevailing south westerlies delivering additional moisture and rainfall over Tamil Nadu during dry spells. We find low latent heat flux over the southern WG due to low ET during the four monsoon months after considering deforestation (Figures 3a–3d). However, the northern (upper) WG shows low ET only during the onset of monsoon in the sensitivity simulation without vegetation (Figures 3b–3d). The vegetation of northern WG mostly consists of woody savannas, which have low leaf area index likely resulting in low ET. The dominant vegetation of southern WG is the evergreen broadleaf forest with high leaf area index that produce higher amount of moisture through ET. In the scenario of deforestation, the supply of moisture from the WG decreases along with the decrease in vertically integrated moisture transport over the transition zone of WG forest and Tamil Nadu plain (Figures 3e–3h) leading to a decreased precipitation over Tamil Nadu throughout the monsoon season (Figures 3i–3l). Removing vegetation cover in this region reduces the ET-driven generation of moisture and that leads to a significant loss of precipitation over Tamil Nadu.

Details are in the caption following the image
(a–l) The differences between the simulated meteorological variables as obtained from the sensitivity run (without considering vegetation) and control run (with vegetation) over the Western Ghats during June, July, August, and September. Panels (a–d) are for latent heat flux; panels (e–h) are for vertically integrated moisture flux (VIMT); panels (i–l) are for precipitation.

We find an increase in precipitation over the southern part of WG (Figure S4b) when we eliminate the vegetation over the WG (statistically significant at 0.05 level). This can be explained with the Bowen Ratio. An earlier study by Saad et al. (2010) over the Amazon forest suggests an increase in precipitation with patchy deforestation, and this is due to the increase in local Bowen Ratio over deforested regions. Increased precipitation was observed for deforested simulations near the upwind edges of deforestation patches. The increase in precipitation after deforestation over the southern (lower) WG (Figure S4b), which is associated with an increased Bowen Ratio (Figure S4c), is consistent with that found over the Amazon. However, WG being a rainfall surplus region, these changes in rainfall may not be impactful enough although upstream effects of such a deforestation in terms of the moisture fluxes must be considered in detail with a fully coupled model. To summarize, deforestation in WG leads to a decrease in rainfall over the Tamil Nadu and Northern WG with a slight increase in Rainfall in Southern WG.

We further selected three years with extremely deficit summer monsoon rainfall over Tamil Nadu, viz., June to September of 1993, 1999, and 2002. These years are selected based on the threshold of rainfall amount (mean −1 standard deviation). We find that the deforestation over WG reduces 40–50% of the precipitation over Tamil Nadu during all the three years (Figures 4c–4e). The major source of surface water in Tamil Nadu is the river Cauvery, and this river basin extends over the states Karnataka and Tamil Nadu (Figures 4a and 4b). We find that the vegetation over the WG region supplies the moisture to the southern part of Peninsular India.

Details are in the caption following the image
Locations of (a) Tamil Nadu and (b) Cauvery river basin. (c–e) The differences between the simulated precipitations as obtained without and with vegetation over the Western Ghats for the years 1993, 1999, and 2002. These three years are rainfall deficit years, and this classification is performed based on the anomalies of summer monsoon precipitation over the Peninsular India.

We have further computed the difference between precipitation and ET (neglecting other minor losses), which is an indicator of water availability (Figures S5a–S5c) for the state of Tamil Nadu. The results for the differences in water availability between simulations without vegetation and the control simulations are in agreement with that for precipitation. During a dry spell, significant portion of the available water in Tamil Nadu comes from the vegetation over the WG and this is consistent with our findings obtained with the simulations for precipitation.

We further verified the same with back trajectories applied to the simulations. The model used for back trajectory is the Dynamic Recycling model developed by Martinez and Dominguez (2014). Such an approach should ideally be applied to the observed or reanalysis data; however, the reanalysis data used here are of too coarse resolution to capture the thin belt of WG. Hence, we have used simulated data of fine resolution for the same. We find that the forests of WG contribute as high as 3 mm/day of rainfall during August and September over majority of locations of water scares Tamil Nadu and Cauvery basin (Figure S6). During June and July, the contribution is found to be lower at around 1 mm/day. This fact together with the characteristics of precipitation and ET for both Tamil Nadu and WG region (Table S1) implies that whenever oceanic sources of moisture become restricted or limited, WG acts as a capacitor for moisture supply, and vegetation from WG contributes toward precipitation for both the regions during summer monsoon months. For Tamil Nadu, surface temperature increases by 0.25 °C, when the vegetation of WG region is removed (Table S1). This probably indicates that a cooling is associated with increased precipitation resulting from moisture generated by the forest of WG. These findings are consistent with each other and indicate that WG vegetation plays a greater role on the general meteorology of the area.

The WG orography can also be expected to play a major role in the generation of precipitation (Boos & Kuang, 2010; Chakraborty et al., 2006; Gadgil & Joshi, 1983; Gunnell, 1997; Konwar et al., 2014; Manabe & Terpstra, 1974; Patwardhan & Asnani, 2000). We performed simulations to understand these direct orographic effects and find that orography affects the precipitation not only over the peninsula but also over the Bay of Bengal, which will have a significant feedback to the atmospheric processes (Supplementary Text and Figures S7a–S7h). This can only be tested with a coupled atmosphere-ocean model and will be a potential area of future research.

4 Conclusion

The recent increase in the rate of deforestation over the WG, one of the world's hottest biodiversity hot spots, is a major concern of the ecological community, primarily due to the loss of rare flora and fauna. However, the contribution of this dense vegetation over WG to the water cycle has not been explored thus far to the best of our knowledge. In the present study, we conclude based on regional model simulations that an important and significant contribution from WG vegetation lies in the supply of moisture toward generating precipitation over the water-deficit state of Tamil Nadu, Cauveri river basin, and over the WG regions of Peninsular India. The vegetation of WG contributes 25% of the moisture to the spatially averaged southwest monsoon precipitation over Tamil Nadu and exceeds 40% at several locations. The maximum contribution of moisture to the monsoon precipitation reaches up to 50% during the dry spells within a season as well as during deficit summer monsoon seasons. The back trajectory analysis of atmospheric moisture reveals that the vegetation over WG contributes around 3 mm/day of rainfall during the months of August and September for majority of the areas of Tamil Nadu and Cauvery river basin. Such a huge contribution has significant importance for the Kharif cropping activities, especially over Tamil Nadu. The major source of surface water, the Cauvery River, is already failing to meet the needs of Tamil Nadu. The situation will further worsen if the high deforestation rate continues across WG.

Acknowledgments

The authors sincerely acknowledge the financial support by the Ministry of Earth Science (MoES/PAMC/H&C/35/2013-PC-II). The precipitation data over India have been collected from India Meteorological Department (http://www.imd.gov.in/advertisements/20170320_advt_34.pdf). The land cover data that provide the details of vegetation over the WG are obtained from Moderate Resolution Imaging Spectroradiometer (MODIS, https://modis.gsfc.nasa.gov/data/). The reanalysis data is obtained from European Centre for Medium-Range Weather Forecasts (http://apps.ecmwf.int/datasets/data/interim-full-daily/levtype=sfc/). S. P. acknowledges Amey Pathak for the help in deriving shape files of maps. The authors acknowledge the Editor and the two anonymous reviewers for reviewing the manuscript and providing very useful suggestions.