Introduction

The hydrological cycle is a vital component in the functioning of an ecosystem as it links various biogeochemical and energy cycles (Wilson and Baldocchi 2000). The forests are a significant part of the hydrological cycle (Roberts 2009). In tropical and sub-tropical areas, the forests receive most of the water through precipitation. A part of water, received in the form of precipitation, is intercepted by the canopy and evaporated from the surface of vegetation before reaching the soil. The rate of water infiltration, run-off, and percolation are also affected by the density and depth of the root channels (Dunne et al. 1991). During CO2 assimilation by the process of photosynthesis, leaves take up CO2 by diffusion through the stomatal pores; simultaneously they also lose water by the process of transpiration.

The transfer of water from the Earth's surface to the atmosphere, through the process of evaporation from various sources including plant surface, soil, and transpiration from plants is known as evapotranspiration (ET) (Hamon 1960). This water transfer or exchange comprises a change of state of water from liquid to vapor, which absorbs energy and cools the land surface (Wang and Dickinson 2012). The latent energy involved in ET is termed as the latent heat of vaporization. At the ecosystem level, the partitioning of energy, available through the net radiation, into ET (latent heat flux, LE) and sensible heat flux (H) affects various aspects of weather and climate (Wever et al. 2002).

Evapotranspiration (ET) provides a nexus between the terrestrial water, carbon, and energy exchange processes (Monteith 1965; Sheffield et al. 2010; Wang and Dickinson 2012). It has been considered a predominant variable needed to understand the efficiency of water utilization by vegetation (Allen et al. 1998; Anderson et al. 2011; Nandy et al., 2021). It is also an indicator of extreme events like flash droughts (Anderson et al. 2013; Otkin et al. 2016). Moreover, long-term changes in ET at a regional scale can ascertain the persistent drought and desertification caused by climate change (Sheffield et al. 2012; Greve et al. 2014; Mao et al. 2015). Hence, robust and long-term ET measurements are vital to enhance the understanding of the global water cycle and hydrological function of forests (Tong et al. 2017).

To estimate ET, several methods have been established including simple empirical methods, soil–water budget methods, Bowen ratio, or eddy covariance (EC)-based measurements and modelling techniques, which have been implemented at varying scales. The analytical methods involve discrete measurements of transpiration, rainfall interception, and soil and understory ET using various tools like sap flow sensors (Ford et al. 2007), rain gauges above and below canopy (Herbst et al. 2008), and measurement chambers (Ford et al. 2007) whereas the soil water budget method depends on the resolution of the soil water budget equation using rainfall, soil water content (SWC) and drainage measurements (Schwärzelet al. 2009). In forests, the species composition, distribution, and sapwood area of different species affect the upscaling of sap flow-based studies from tree to stand level (Watham et al. 2017a). The variability in rainfall interception depends on the forest density, structure of the trees as well as on weather conditions (Soubie et al. 2016). The soil and understory ET in mature forests are usually low, but it depends on energy reaching the soil, thus, on leaf area index (LAI), hence, it can vary significantly for some ecosystems (Vincke et al. 2005). The information obtained from these methods is limited in their scope to explain the biophysical controls of ET on hourly to daily time scales (Baldocchi and Ryu 2011).

Lately, the ET and energy fluxes are being studied in several ecosystems using the EC technique. The EC-based method provides in situ, stand scale measurements of forest ET and energy partitioning with negligible interference (Baldocchi and Ryu 2011). These systems usually consist of instruments that allow high-frequency measurements of CO2 and H2O fluxes. These systems provide long-term valuable observations of fluxes (Wang and Dickinson 2012). There has been a steady increase in the application of the EC method because of its ability to directly measure the gas and energy exchange between ecosystems and the atmosphere over the forest areas (Baldocchi and Ryu 2011; Aubinet et al. 2012). Several EC flux towers have also been established over different forest ecosystems across India. The diurnal and seasonal variability of CO2 fluxes has been reported for these forest ecosystems (Jha et al. 2013; Rodda et al. 2016; Deb Burman et al. 2017; Watham et al. 2017b, 2020). However, the dynamics of ET have not been studied in detail.

In general, for the process of photosynthesis to occur, the rate of CO2 transfer from the atmosphere to the carboxylation sites is directly linked to the water loss occurring through the process of leaf transpiration (Marques et al. 2020). These exchanges occurring in an ecosystem are controlled by the interaction of various environmental factors, including incoming solar radiation (Rg), air temperature (Tair), vapor pressure deficit (VPD), and SWC, in combination with the vegetation biological processes like leaf emergence and development (Zha et al. 2013). It has been reported that these biophysical variables play a significant part in regulating the diurnal to inter-annual variability of ET (Baldocchi et al. 2004; Baldocchi and Xu 2007). Hence, an in-depth understanding of these controls is essential to assess how the changes in climate may affect the variability in ET. The present study aims to report the EC-based measurements of ET and energy fluxes, in the moist deciduous plant functional type (PFT) of northwest Himalayan (NWH) foothills of India for a period from 2016 to 2018. The main objective of the study is to document how ET and energy fluxes vary in the moist deciduous PFT. Additionally, it explores how the biophysical factors control ET in this ecosystem as it represents one of the major PFTs in the NWH foothills of India.

Material and methods

Study site

The study was conducted at Barkot Flux Site (BFS, 30° 06′ 44.40″ N, 78° 12′ 43.06″ E), which is located in the NWH foothills of Uttarakhand, India (Fig. 1). The elevation at the study site is 415 m above sea level which is relatively flat to undulating terrain. It has a monsoon-influenced humid subtropical climate. The site lies in the moist deciduous PFT (Srinet et al. 2020), which was also categorized as Tropical Moist Deciduous forest by Champion and Seth (1968). The overstory is dominated by sal (Shorea robusta) and its associates present in the area include Terminalia tomentosa, Lagerstroemia parviflora, T. bellirica, and Syzygium cumini (Nandy et al. 2017). The understory mainly comprises Mallotus philippensis, Ehretia laevis, and Cassia fistula. The maximum canopy height observed at the study site is approximately 32 m. It is a climax forest with a maximum LAI of 4.2 (Watham et al. 2020). The forest around the flux tower site experiences leaf fall from the end of February to the beginning of May (Srinet et al. 2019). The mean annual temperature observed from 2016 to 2018 was 21.87 °C and the annual precipitation was 1115.40 mm.

Fig. 1
figure 1

Location of Barkot flux site

Full size image

Data used

The study was conducted over 3 years, from 2016 to 2018. The energy and water fluxes between the forest and the atmosphere were measured using the EC method. The EC system installed at the site consists of an integrated CO2/H2O open-path infrared gas analyzer and 3D sonic anemometer (IRGASON, Campbell Scientific). It takes continuous 10 Hz measurements of CO2 and H2O fluctuations at 46 m height above the ground surface. Rg was measured at 46 m height using a 4-component net radiation sensor (CNR4-L, Kipp and Zonen, The Netherlands). Tair and relative humidity (RH) were measured at 6 levels (2 m, 4 m, 8 m, 16 m, 32 m, 48 m) using relative humidity and temperature probes (HygroclipsS3, Rotoronic, Switzerland). Soil temperature and SWC were measured at 5 levels (5 cm, 10 cm, 20 cm, 60 cm, 100 cm below the ground surface) using soil moisture and temperature sensor (CS650, Campbell Scientific). All these data were recorded using CR3000 data logger (Campbell Scientific, Logan, UT, USA). The meteorological data were recorded at half-hourly intervals.

To analyze the control of biophysical variables on ET, the observed biophysical variables including Rg, Tair, VPD, precipitation, and SWC were taken into account, whereas, the LAI values were taken from the Moderate Resolution Imaging Spectroradiometer (MODIS) combined Leaf Area Index (LAI) and Fraction of Photosynthetically Active Radiation (FPAR) 8-day composite dataset (MOD15A2H).

Data processing and analysis

The flux data collected by the EC system was processed using EddyPro 6.2.0 software (Li-COR Biosciences, USA) and carbon, water, and energy fluxes were calculated. The processing was carried out using the standard procedure which includes splitting the data into 30 min files; despiking; block averaging; 2D coordinate rotation; spectra correction; Webb, Pearman, and Leuning (WPL) and other corrections; and quality control (Burba 2013). For the night-time filter, frictional velocity (u*) filtering was applied to the obtained half-hourly time-series for 2016–2018. The 30-min mean fluxes were screened for precipitation periods, instrument failure, and out-of-range records. The data was checked for gaps and gap filling was carried out using the ReddyProc package (Wutzler et al. 2018) in R, using marginal distribution sampling (MDS) (Reichstein et al. 2005; Foltýnová et al. 2020).

Half-hourly ET over the study site was calculated as (Ma et al. 2017):

$$ET= \frac{LE}{L{\rho }_{w}}$$
(1)

where LE is the latent heat flux; \(L\) is the latent heat of vaporization of water (2.45 kJ g−1); \({\rho }_{w}\) is the density of water. To study the seasonal pattern of ET, based on the climatic variations and plant phenology, the year was divided into four seasons (Watham et al. 2020): (a) winter (DOY: 01 to 90), (b) dry summer (DOY: 91 to 181), (c) monsoon (DOY: 182 to 273), and (d) post-monsoon (DOY: 274 to 365) seasons.

A relative weight analysis (RWA) using a multiple regression model was carried out to understand the effect of variability of various biophysical variables including Rg, Tair, VPD, precipitation, SWC, and LAI on ET in R environment. As the biophysical variables were highly correlated, therefore, to understand their proportionate contribution to the regression model (R2), considering both its unique contribution and its contribution when combined with other variables (Johnson 2000), the RWA was implemented. The LAI data was available at a temporal resolution of 8-days, hence, the RWA was carried out at 8-days timescale.

Results

Variations in energy fluxes and evapotranspiration

Variability of energy fluxes

Figure 2 shows the diurnal variation in the half-hourly means of H and LE for all the seasons from 2016 to 2018. The diurnal variation in H was the largest during the dry summer season (-6.72 to 207.83 W m−2 with the peak at 1300 h). The values of H ranged from -8.31 to 118.34 W m−2 in winter with a peak observed at 1230 h, -4.34 to 61.03 W m−2 in monsoon with a peak at 1300 h, and -3.97 to 60.73 W m−2 in post-monsoon season with a peak at 1130 h. The mean diurnal variation in H for 2016 to 2018 was found to be from -1.31 to 109.35 W m−2. The night-time values of H were found to be negative in all the seasons. The variability in LE was the highest in the post-monsoon season; it ranged from 2.52 to 223.54 W m−2 with a peak at 1230 h. LE ranged from 7.10 to 214.41 W m−2 in monsoon with a peak at 1300 h, 1.79 to 175.64 W m−2 in winter with a peak at 1230 h, and 2.72 to 144.31 W m−2 in dry summer with a peak at 1130 h. The mean diurnal variation in LE for the seasons ranged from 4.47 to 186.89 W m−2.

Fig. 2
figure 2

Seasonal averaged diurnal variations of sensible (H) and latent heat flux (LE)

Full size image

Figure 3 represents the variability of mean daily H and LE from 2016 to 2018. H and LE values showed significant seasonal and day-to-day variation. The highest value of H was observed during dry summer whereas the highest value of LE was observed during the monsoon season. With the onset of the leaf-fall period, during the end of February to March, daily LE showed a decreasing trend and H showed an increase in values. H was found to be higher than LE in dry summer. H sharply decreased by the end of dry summer whereas LE started increasing. LE continued to increase and H tended to decrease at the beginning of the monsoon season. The value of H maintained a stable trend during monsoon and post monsoon seasons, whereas, LE decreased in post-monsoon season. However, the value of LE was always higher than H in monsoon and post-monsoon seasons.

Fig. 3
figure 3

Daily sensible (H) and latent (LE) energy variations at Barkot flux site from 2016–2018

Full size image

Variability of evapotranspiration

The annual ET was found to be 693.67 ± 46.70 mm. The diurnal variation of ET is shown in Fig. 4. The highest diurnal variation was observed during the post-monsoon season (0.003 to 0.32 mm h−1) followed by monsoon (0.01 to 0.31 mm h−1), winter (0.002 to 0.25 mm h−1), and dry summer (0.003 to 0.21 mm h−1). The annual variation of ET at the study site is shown in Fig. 5. ET exhibited a multi-peak trend in all the 3 years. The daily ET varied from 0.19 to 4.27 mm day−1 in 2016, 0.25 to 4.07 mm day−1 in 2017, and 0.35 to 4.04 mm day−1 in 2018. The total ET was found to be 745.80 mm year−1 in 2016, 679.52 mm year−1 in 2017, and 655.67 mm year−1 in 2018. The peak ET was observed during September, which was during the late monsoon period. Mean daily ET started increasing from January till March–April and reduced during the dry summer months with few fluctuations. The peak ET was observed in the monsoon season with a few dips in the values and it started reducing slightly again in post-monsoon season.

Fig. 4
figure 4

Seasonal mean diurnal variation of evapotranspiration (ET, mm h−1)

Full size image
Fig. 5
figure 5

Variation of daily evapotranspiration (ET) at Barkot flux site from 2016–2018

Full size image

Biophysical control on evapotranspiration

To understand the influence of the local environment on ET, a statistical analysis was carried out. Seasonal variations in Rg, Tair, VPD, precipitation, and SWC vis-à-vis ET are presented in Fig. 6. The maximum Rg was observed during the end of the dry summer season while the minimum Rg was observed during the monsoon season. The Rg values showed significant day-to-day fluctuations apart from the post-monsoon season. These fluctuations were attributed to the presence of cloud cover. The trends observed in the daily ET values were very similar to the trend in Rg. From 2016 to 2018, the inter-annual mean Tair was 21.85 °C with a winter average of 16.10 °C and a summer average of 27.80 °C. The maximum value of Tair was observed in May in the dry summer season whereas the minimum Tair was observed during January in the winter season. The annual mean VPD was found to be 20.87 hPa. The seasonal means were found to be 17.38, 33.79, 16.62, and 15.68 hPa in winter, dry summer, monsoon, and post-monsoon seasons, respectively. The annual precipitation was 1115.40 mm. In 2017, the precipitation was 8.75% higher than the three-year average. The SWC varied slightly from season to season, apart from the monsoon season when the highest value of SWC was observed.

Fig. 6
figure 6

Variation of daily meteorological variables a incoming solar radiation (Rg, W m−2), b air temperature (Tair, °C), c vapour pressure deficit (VPD, hPa), d precipitation (mm), e soil water content (SWC, v/v) and evapotranspiration (ET, mm day−1) from 2016 to 2018

Full size image

The RWA was carried out to understand the contribution of each biophysical variable in the variability of ET at the site scale. A combination of Rg, Tair, VPD, Precipitation, SWC, and LAI were able to explain 73% of the variability of ET. Amongst the biophysical variables used for the present analysis, Rg (rescaled relative weight (RRW) = 26.65) showed the highest control on the variability of ET, followed LAI (RRW = 22.26), VPD (RRW = 20.57), Tair (RRW = 17.85), SWC (RRW = 8.11), and precipitation (RRW = 4.57) (Fig. 7). It was found that Rg, LAI, Tair, and SWC had a positive relationship with the ET values, whereas VPD and precipitation had a negative correlation with ET. The lower relative weights of precipitation and SWC with respect to the variability in ET suggested that the ET in this ecosystem was not limited by water availability.

Fig. 7
figure 7

Rescaled relative weight of various biophysical variables on evapotranspiration (the signs represent the type of relationship)

Full size image

Discussion

The EC method facilitated the study of ET from half-hourly to yearly time scales by providing continuous time-series data for the study site. This method provides a direct measurement of energy fluxes (LE and H). The only problem is the gaps, which occur due to bad weather conditions and instrumental issues. Suitable gap-filling approaches were applied to obtain continuous time-series data of LE and H. To evaluate the performance of EC measurement at the present flux tower site, Watham et al. (2020) examined the energy balance closure for 2016–18 and reported a 73% closure.

The diurnal and seasonal patterns of H were closely related to the pattern observed in Rg as the radiation is the principal energy source for daytime surface warming and evaporation (Grachev et al. 2020). The variability of LE is dependent on radiation as well as the availability of moisture. The lower value of LE in the dry summer season can be due to high temperature and low moisture conditions. In the monsoon season, more moisture is available in the ecosystem with enough sunshine, which leads to higher LE and ET. The stomatal openings are responsible for carbon fixation via photosynthesis as well as for water loss through transpiration. In the post-monsoon season, with moisture availability, warm temperature, and abundant sunlight, the optimum conditions for photosynthesis are available; hence, the LE and ET values were higher.

There is a dearth of ET studies on the forests of India; therefore, the results of the present study were compared to the forests with similar conditions. The mean annual ET of forests from across the globe has been reported to be 503 ± 388 mm year−1 (Baldocchi 2020). For the forests of China, in subtropical monsoon climate, the ET values of Populus sp. was found to be 957.8 mm year−1, whereas, in subtropical monsoon humid climate, the ET values of an evergreen broadleaf forest was 630.0 mm year−1 (Xiao et al. 2013). The ET in Populus sp. was higher than that observed at BFS as the water demand of Populus sp. is higher than that of sal. The conditions at Chinese evergreen broadleaf forest in subtropical monsoon humid climate are close to the sal forest in monsoon-influenced subtropical climate at BFS. The difference in ET observed at both the sites can be attributed to the dynamics of LAI in sal forest as it shows slight deciduous behavior.

The variability in climate potentially influences the rates of ET from the forest canopies as these changes affect the surface conductance and transpiration (Humphreys et al. 2003). In dry conditions, the forests attained a physiological control of water loss by closing the stomata resulting in reduced canopy conductance whereas in water-saturated conditions the canopy conductance does not limit the water flux (Jarvis and McNaughton 1986). The biophysical variables play a vital role in controlling the ET dynamics through their possible effect may be site-specific (Yu et al. 2020).

In the present study, the RWA analysis reflected that Rg, LAI, Tair, and SWC had a positive influence on the ET values. However, VPD and precipitation had a negative influence on ET in this moist deciduous PFT. Rg has a strong positive influence on the ET values, as it is the only input energy, which after partitioning can affect the heat and water vapour transport to and from the leaf surfaces (Oke 1987). The negative influence of VPD on ET variability can be attributed to the self-adaptive behavior shown by the plants to reduce the utilization of resources by closing the stomata to conserve water. The seasonality of ET can be affected by LAI as it provides the surface for transpiration to take place and precipitation interception by the canopy (Jin et al. 2017), which may result in an increase in ET. Singh et al. (2014) also reported that the relationship between ET and LAI showed tight seasonal coupling. As reported in the study, the temperature was also found to have a significant role in controlling the variability of ET (Xiao et al. 2013; Yu et al. 2020). An optimum temperature is required for photosynthesis to take place, which in turn controls the transpiration process from leaf stomata. Both the processes are intrinsically related in terrestrial ecosystems (Jarvis 1976). The temperature also influences the evaporation process in an ecosystem. Precipitation can encourage ET by increasing SWC (Legesse et al. 2003) and promoting vegetation growth (Jin et al. 2017). On the other hand, precipitation can sometimes limit ET by reducing radiation due to cloudy conditions (Yu et al. 2020). In the present study, the impact of precipitation on ET was found to be negative, which indicates that the ET in moist deciduous PFT is not water-limited. Whereas it is positively correlated with Rg and Tair, which reveals that it is energy-limited. A combination of these variables was able to give an insight into the biophysical controls on the variability of ET of the moist deciduous PFT present at BFS. With an increase in temperature and atmospheric CO2 in view of climate change, it is very difficult to predict the behavior of ecosystem ET. With the increase in temperature, the evaporation may increase, however, the greater CO2 concentration and increase in VPD may lead to stomatal closure, which will result in reduced transpiration (Baldocchi 2020). Therefore, such long-term ET measurements can be crucial for understanding the role of forests in the changing climate.

Conclusion

The forests are an intrinsic part of the terrestrial hydrological cycle. EC method has been considered as one of the most reliable techniques to investigate the carbon, water, and energy fluxes and to understand the biophysical controls in their variability by providing continuous, long-term observations for these complex processes. ET provides a nexus between these complex processes. Hence, it is essential to get accurate estimates of ET to decode the trade-off between these processes and to perceive the hydrological cost of the carbon sequestration process. This study presented comprehensive estimates of ET using the EC flux tower measurements for a period from 2016 to 2018 and also provided an insight into the biophysical controls on ET in moist deciduous PFT of NWH foothills of India. The average annual ET was found to be 693.67 ± 46.70 mm year−1. The highest seasonal ET was found in the post-monsoon season. Rg, LAI, and VPD primarily controlled the variation in ET. The study provides the first direct, long-term estimates of ET for an Indian forest.