1 Introduction

The unequivocal warming of the earth system as a response to the gradual increase in the ambient concentration of the greenhouse gases (GHG) through anthropogenic activities is well acknowledged now (Solomon et al. 2007). Such warming of the earth system is leading to the genesis of some of the abrupt weather and climate systems across the globe. The erratic patterns of rain, temperature events, and increased frequency of the hydro-climatic extreme events in the recent time lead us to such inferences. Following the warming and the increase in the global mean surface temperature, it was suggested that the increase in the moisture content of the atmosphere (Trenberth et al. 2007) might lead to enhanced precipitation. A considerable increase in the moisture transport possibly resulting into an increase in the extreme precipitation events over India has been suggested by Bhaskaran et al. (1995) and May (2002).

A wide range of limitations and uncertainties attributed to different models and scenarios is associated with climate projections as evident from a varied range of temperature and precipitation projections (Chaturvedi et al. 2012). However, it is claimed that the climate projections are more reliable at the global scale as compared to the regional scales (Taylor et al. 2012). The regional climate projections are important to understand the region-specific changes for more comprehensive impact and adaptation studies. A projection based on CMIP5 models has suggested a warming of the ensemble mean from 2 °C (RCP2.6) to 4.8 °C (RCP8.5) over India (from 1880 to 2080s) delineating different climate possibilities under a range of scenarios and their associated uncertainties (Chaturvedi et al. 2012). Also, an increase in all India precipitation by a magnitude of 6, 10, 9, and 14% under the RCP2.6, RCP4.5, RCP6.0, and RCP8.5 scenarios, respectively, has been projected by 2080 with respect to 1961–1990 baseline period with a larger variability in the spatial distribution.

It is found out that global climate models (GCMs) are capable of addressing the large-scale climatic features to simulate the atmospheric general circulation because of their relatively coarser resolution at the continental scale in a satisfactory manner but are not capable of capturing the regional-scale meteorological processes especially those associated with complex topography (Giorgi and Mearns 1999; Rummukainen 2010; Kumar et al. 2013). In this respect, dynamical downscaling of GCMs has shown competency in simulating the regional climate over the regions having complex topography (Samuelsson et al. 2011). There has been considerable progress in the use of regional climate models for impact studies over different regions across the globe due to their potential in simulating the regional climate with greater confidence. However, regional climate models (RCMs) are not always helpful in understanding the climate over the regions having complex topography, as in some of the cases, they have shown an intensification or overestimation of the precipitation over those regions (Hirakuchi and Giorgi 1995). Especially in respect to Indian summer monsoon rainfall, it has been seen that over the regions of high precipitation (due to orography), regional climate models tend to overestimate the precipitation (Choudhary et al. 2017). A general pattern of overestimated precipitation over the same region has been reported in different RCM-based studies (Bhaskaran et al. 1996; Jacob and Podzun 1997; Ji and Vernekar 1997; Ratnam and Kumar 2005; Dash et al. 2006; Nguyen and McGregor 2009; Dobler and Ahrens 2010; Mathison et al. 2013; Dimri 2012; Moors et al. 2012; Kumar et al. 2013). The reason for such systematic biases in the regional climate simulation has been attributed to the cumulus parameterization in most of such studies. Moreover, it has also been shown that over the mountainous regions having complex terrains, it is difficult to represent the velocities accurately in the numerical models. This shortcoming leads to unrealistic forcing of the grid-scale precipitation schemes leading to errors in the simulation (Samuelsson et al. 2011).

Many studies have been carried out to assess the trends as well as the changes in the climate system over India. Unanimously, it has been reported that the monsoon rainfall is trendless in nature and often behaves in a random fashion while analyzed over a long time (Mooley and Parthasarathy 1984; Guhathakurta and Rajeevan 2008). Though no significant trend in the precipitation was found, significant changes in the long-term rainfall over different spatial and temporal scales are shown (Dash et al. 2007; Kumar et al. 2010; Guhathakurta et al. 2011). Dash et al. (2007) also reported that the summer monsoon rainfall over Northeast India (hereafter NEI) and east of Madhya Pradesh along with adjoining parts of Gujarat and Kerala has shown falling trend for the period 1871–2002. A decrease in the trend of annual rainfall and number of rainy days in the 15 out of 22 basins over India was reported by Jain and Kumar (2012).

Using a set of regional climate models for the present climate, it has been shown that most of the COordinated Regional climate Downscaling EXperiment (CORDEX) RCMs are able to capture the spatial patterns of the precipitation over most parts of India including Northeast India (Choudhary et al. 2017). However, models have inherent biases associated with them in the simulation of precipitation. While using the similar set of experiments, Singh et al. (2017) demonstrated that no significant improvement or value was added by the regional climate models to their GCM counterparts in simulating different features of Indian summer monsoon precipitation.

Moreover, for future climate, a study based on the use of regional climate model (PRECIS) for the projection of climate under increased greenhouse gas and sulfate aerosol scenarios has suggested a considerable increase in the temperature as well as precipitation over India by the end of the twenty-first century (Kumar et al. 2006). The study also reports that, however, warming has a monotonous pattern spread across the Indian region, but precipitation changes in the future have a substantially different spatial pattern over the country. While talking about the regional scales, decreasing trends (− 6 to 8% of the normal per 100 years) in the monsoon rainfall have been reported over the northeast peninsula, NEI, and northwestern part of the country (Kumar et al. 1992). A significant warming trend of 0.57 °C per 100 years for the period of 1881–1997 with higher warming in post-monsoon and winter season was reported by Pant and Kumar (1997). A rising trend in the mean annual temperature over the west coast, north central, interior peninsula, and NEI region along with the whole India was observed during the period 1901–1982 (Hingane et al. 1985). A falling trend in the diurnal temperature range at sites from NEI was found for annual, seasonal (pre-monsoon and monsoon), and monthly (September) time scales (Jhajharia and Singh 2011). Goswami et al. (2006) in their study claimed that NEI is a region of the high mean and high variability of the rainfall events attributed to the strong influence of the local orography on the rainfall. Jhajharia et al. (2007) by using station data over Agartala have shown that a decrease in the total pan evaporation accompanied by an increase in the relative humidity has occurred with the simultaneous rising trend of mean air temperature with the slope 0.32 °C/year. Same authors have also mentioned a declining trend of annual evaporation for the period 1969–2002. At the same time, a falling trend of total annual rainfall for the period 1953–2000 with a rate of 2.4 mm/year has been suggested in the same study. In another study, Jhajharia et al. (2012) over Assam reported a decreasing trend of precipitation during the month of June and July. A significant decreasing trend of precipitation with a rate of 3.2 mm/day was observed over the central Northeast India during the period 1889–2008 (Subash et al. 2011). Dash et al. (2012), using a regional climate model projection under A1B scenario, have suggested that the annual mean temperature over northeast may rise by 5 °C by the end of the twenty-first century. The same study also projects an increase in precipitation in general across the region along with a decrease in the precipitation of the parts of Sikkim, Assam, Meghalaya, and Mizoram.

Although some studies have been undertaken to study and assess the present climate over NEI, very few efforts have been made to incorporate the information available from climate models. Also, recent efforts using the regional climate models from the CORDEX South Asia (SA) framework did not emphasize the regional aspects of climate over the northeastern region. The RCM simulation used in the current study has not been evaluated in the previous studies. Further, in order to plan for the appropriate adaptation and mitigation measures under the changing climate scenario, generating reliable and accurate climate projections is necessary following the available climate scenarios. In the current study, an effort has been made to investigate and summarize the trends of the recent climate followed by the future projections of climate under the CORDEX South Asia framework.

The paper is structured as follows. Section 2 comprises a brief description of the study area, CORDEX South Asia experiment, the observational dataset, and methodology used for presenting the projection of the future climate. Section 3 consists of the results on the trends and changes in the climate under different scenarios. Section 4 presents the summary of major findings and conclusions of the study.

2 Data and methodology

2.1 Study area

The easternmost part of India consisting of the seven sister states (Arunachal Pradesh, Assam, Meghalaya, Mizoram, Manipur, Nagaland, and Tripura) which are along with Sikkim is known as Northeast India. The region consists of different physiographic categories across the states such as Eastern Himalaya, Northeastern Hills, and the plain areas including Brahmaputra and Barak valley plains. The climate of the region has been characterized as sub-tropical humid type as per the Köppen’s classification (Oliver and Wilson 1987) and is governed by distinct relief features as well as monsoonal circulation during both southwest and northeast monsoon seasons. The region is abode to a diverse variety of flora and fauna and also lies the world’s wettest place Mawsynram, which receives an approximate of 1187.3 cm of rain every year (Dash et al. 2012). The distribution of topography over the study region has been shown in Fig. 1.

Fig. 1
figure 1

Surface elevation (m) over the Northeast India region

Full size image

The area under the Indian landmass with longitude 88°E–98°E and latitude 22°N–30°N has been considered for the analysis of climate variables. The region shares its international boundary with China (North), Myanmar (East), Bangladesh (Southwest), and Bhutan (Northwest). The climate of the region is markedly distinct from other regions in India due to influences from the features like orography, alternating pressure cells of NEI, Bay of Bengal giving rise to maritime tropical air masses, southern Indian Ocean, periodic western disturbances, and the local valley and mountains due to specific terrain features (Barthakur 2004). The region is endowed with one of the largest river basins in the world comprising a network of major rivers and also has a number of lakes, streams, falls, and watersheds important to sustain an agricultural-based economy and survival base over the region. The NEI region has a fragile mountain ecosystem with a diverse variety of vegetation types ranging from sub-tropical, sub-montane, montane, and sub-alpine to alpine systems (Das et al. 2009). The region is also acknowledged as the biodiversity hotspot due to its rich species diversity and high endemism. Thus, it becomes even important to focus on the region for any possible changes in future climate keeping in mind the high sensitivity and fragile nature of this unique ecosystem.

2.2 CORDEX experiment and model data

The importance of the downscaling technique in providing the regional climate information has been well appreciated in the climate community in the recent times. The World Climate Research Programme (WCRP) supports the COordinated Regional climate Downscaling EXperiment (CORDEX) with the major aim to provide the reliable regional climate change scenario across the globe. The programme was launched with the vision to develop an international coordinated framework for the generation of improved present as well as the future projection of regional climate information (Fernández et al. 2010).

Under the CORDEX framework, the regional climate data from the simulation over South Asia domain at the spatial resolution of 50 km (0.44°) for the regional model (REMO) (Jacob 2001) driven by MPI-ESM-LR global model has been obtained for the study. The global model used to drive the REMO belongs to r1i1p1 realization ensemble family. The configurations of the model for the simulations have been discussed in detail in the study by Jacob et al. (2012) followed by Teichmann et al. (2013).

In order to investigate the possible future changes and the associated uncertainties, the data from model integrations available for different representative concentration pathways (RCPs) (Moss et al. 2010) have been utilized. The RCPs do not facilitate to predict the future; rather, it helps understand the uncertainties in order to reach decisions robust enough to a wide range of possible futures. The meteorological variables viz. 2 m air temperature (near-surface air temperature) and precipitation have been analyzed for the present ~ historical (1970–2005) climate. Further, the long-term simulations of climate for the period 2006–2100 under three different emission scenarios, namely RCP2.6, RCP4.5, and RCP8.5, as prescribed under Intergovernmental Panel on Climate Change, Assessment Report 5 (IPCC AR5) have been analyzed. The RCPs are characterized by the changes in the radiative forcing of the atmosphere to the pre-industrial values by the amount in (W/m2) prefixed to each RCP. The data for the REMO experiment under the CORDEX South Asia framework has been acquired from the Centre for Climate Change Research (CCCR), Indian Institute of Tropical Meteorology, Pune, India.

2.3 Observational dataset

The mean near-surface air temperature dataset for the present climate (1970–2005) at a spatial resolution of 0.5° from Climatic Research Unit (CRU) (Harris et al. 2014) has been used for the comparison of climatology, calculation of bias, and trends of temperature over the study region. For precipitation data, the Asian Precipitation–Highly Resolved Observational Data Integration Towards Evaluation of Water Resources (APHRODITE) dataset (Yatagai et al. 2012) at the horizontal resolution of 50 km has been used. The NEI region has only 11 meteorological stations across the 8 states; thus, in order to present a homogeneous climate information, use of high-resolution gridded dataset is a must.

2.4 Methodology

The climatology of the mean near-surface air temperature (2 m air temperature) and precipitation has been calculated from the model as well as the observation dataset for the present climate period (1970–2005) to facilitate the comparison of the spatial pattern of the mean climate between the observation and the model. A season-wise appraisal of the variables has been made as winter (December, January, and February (DJF)), pre-monsoon (March, April, May (MAM)), monsoon (June, July, August, September (JJAS)), and post-monsoon (October and November (ON)). The bias of the model along with the linear trend for the temperature and precipitation has also been calculated for the present climate (1970–2005) for the sake of comparison of the model with the available observations. In order to assess the changes in the immediate as well as the far future, the period 2006–2100 has been sliced into two parts, namely near future (2020–2049) and far future (2070–2099). The changes in the temperature climatology for near-surface air temperature for the near future and far future with reference to the present climate have been calculated for the abovementioned seasons in all the three RCP scenarios. Moreover, the percentage changes in precipitation for future emission scenarios have also been computed for different seasons and scenarios. To investigate the rate of changes (if any) in temperature and precipitation, the long-term linear trend for the period 1970–2099, including the historical period and RCP scenario, has been computed.

3 Results and discussion

3.1 Temperature

As mentioned earlier, the NEI region comprises a wide variety of climatic regimes within across the region. The mean climatology of the near-surface air temperature from the observation as well as the model along with the model bias has been presented in Fig. 2. The model (Fig. 2e–h) is able to capture the spatial distribution of the temperature climatology in a consistent manner when compared with the corresponding observations (Fig. 2a–d) in almost all the seasons. Such characteristics can be attributed to the tuned parameter for soil thermal characteristics from moist to dry over the simulation domain as described in Kumar et al. (2014). The observations show that the peak warming occurs in the monsoon season followed by the pre-monsoon with higher temperatures in the central parts of Assam and Meghalaya; this feature is well captured by the model. However, a considerable bias in simulating the mean temperature climatology has been observed in the model output. A cold bias of up to 2–3 °C has been seen especially in the post-monsoon and winter seasons. The magnitude of the bias is comparatively higher in the higher-elevation regions such as the upper reaches of Arunachal Pradesh and the provincial regions of Sikkim. Again, a warm bias associated with the warm seasons in the plain areas has been found. The cold bias in the model climatology has also been reported by Dash et al. (2012) using the regional climate model RegCM3, where a cold bias of the magnitudes − 2 to − 10 °C has been suggested in the annual mean temperature. The model shows a less bias in simulating the temperature during the monsoon season especially in the southern and eastern parts of the region.

Fig. 2
figure 2

Near-surface air temperature climatology (°C) during 1970–2005 over the Northeast India region from Climatic Research Unit (CRU) dataset (a–d) and for REMO driven by MPI-ESM-LR global model (e–h) and their corresponding bias (°C) (i–l) under CORDEX South Asia experiment for DJF (December, January, February), MAM (March, April, May), JJAS (June, July, August, September), and ON (October and November) seasons

Full size image

The spatial distribution of the linear trends of the near-surface air temperature has been shown in Fig. 3. A continuous warming trend can be noted in all the seasons from the observation data. Warming is more pronounced in the post-monsoon followed by the winter season with its magnitude as high as 0.04 °C/year similar to those reported by Dash et al. (2012). Although the model is able to capture the general warming trend, but in most of the cases, it has the tendency to underestimate the magnitude of warming for the present climate period as shown in Fig. 3e–h.

Fig. 3
figure 3

Trends of near-surface air temperature (°C/year) during 1970–2005 over the Northeast India region from Climatic Research Unit (CRU) dataset (a–d) and for REMO driven by MPI-ESM-LR global model (e–h) under CORDEX South Asia experiment for DJF (December, January, February), MAM (March, April, May), JJAS (June, July, August, September), and ON (October and November) seasons

Full size image

Further, the model shows highest warming rate during the pre-monsoon season contrary to the observation which shows peak warming in winter season. Again, higher magnitude of warming trend has been found in the high-elevation northern regions in both the model and observations indicating towards the possibility of elevation dependence of warming. The time series of spatial correlation (figure not presented) of the seasonal mean temperature between model and observation for the period 1970–2005 shows a fairly good correlation suggesting a fair performance of the model in terms of capturing the spatial patterns of the seasonal mean across the NEI region.

The annual cycle of the temperature over the NEI region (figure not presented) shows the underestimation of the temperature in comparison to the available observation. Also, the corresponding bias is less in the pre-monsoon and monsoon months while it is higher in the winter and post-monsoon seasons. The peak of the temperature occurs in the month of August in observations but it occurs in July for the model-simulated output.

The projected changes in the temperature climatology for different seasons under RCP2.6 emission scenario for near future and far future have been presented in Fig. 4. In the near future, a change in the temperature as high as 2–3 °C may be expected during the winter season followed by the summer season.

Fig. 4
figure 4

Changes in near-surface air temperature (°C) calculated as near future (2020–2049) minus present (1970–2005) (a–d) and far future (2070–2099) minus present (1970–2005) (e–h) for RCP2.6 over the Northeast India region from REMO forced with MPI-ESM-LR global model for DJF (December, January, February), MAM (March, April, May), JJAS (June, July, August, September), and ON (October and November) seasons

Full size image

A similar change can also be noticed in the far future period projection. The central parts of the NEI region are more influenced by the temperature changes during winter over the higher-elevation regions in the north. The monsoon and post-monsoon seasons’ response towards the warming is moderate as the temperature changes in both periods (near and far future) are not as high as the other two seasons. Interestingly, the magnitude of change is less in far future, which may be attributed to the fact that under RCP2.6, the radiative forcing of the atmosphere is less and stabilizes itself towards the end of the century to limit the change in the global mean temperature by a maximum of 2 °C (Vuuren et al. 2011; Moss et al. 2010). Figure 5 shows the changes in seasonal mean temperature climatology for RCP4.5. The model simulates a change in the climatology up to 1–1.5 °C in general in the near future period with more warming in the winter season and least warming in post-monsoon.

Fig. 5
figure 5

Same as Fig. 4 but for RCP4.5

Full size image

The same feature can be seen in the far future as well but the magnitude of change is even higher and ranges up to 3–4 °C at several places including the interior parts of the NEI region. Since RCP4.5 is a stabilizing scenario, thus limiting the atmospheric radiative forcing to 4.5 W/m2 without ever exceeding the value at the end of the century, i.e., 2100 (Thomson et al., 2011), the warming in the far future may be the manifestation of this phenomenon. A considerable difference in the magnitude of change between far future and near future can be seen under RCP8.5 (Fig. 6).

Fig. 6
figure 6

Same as Fig. 4 but for RCP8.5

Full size image

This is due to the fact that, under RCP8.5, due to the higher emission of GHG, radiative forcing is higher and allowed to overshoot without any control and policy interventions (Riahi et al. 2011). Under such emission scenario, a rise in temperature by 3–3.5 °C in the near future and an increase by 6–8 °C in the far future have been projected in the model simulations. The pre-monsoon and winter months are affected by the warmer temperature changes especially in the southern parts of the region constituted by the provincial states of Manipur and Mizoram.

The long-term trends of the temperature confirm the warming of the region in the months of DJF and MAM under different RCPs. The rate of change for the period 1970–2099 under RCP2.6 (Fig. 7a–d) has been found to range between 0.015 and 0.03 °C/year during the winter, with a higher rate of warming over parts of Meghalaya, Tripura, and Sikkim. The warming has the similar spatial pattern for the pre-monsoon and monsoon seasons under the least intensive emission scenario. In general, a continuous warming homogeneously distributed across the NEI region under all the emission scenarios can be observed in Fig. 7. However, the rates of temperature change over the period 1970–2099 (which includes present climate as well as the future projection) may vary but all the scenarios have projected a unanimous warming of the atmosphere. Under RCP4.5, the temperature rise may occur at the rate of 0.04–0.045 °C/year over several parts of the region, while Sikkim and upper reaches of Assam, Meghalaya, Manipur, Mizoram, and Tripura may experience higher warming rate, especially during the winter season. As discussed earlier, due to the absence of control and policy intervention, RCP8.5 is the strongest emission scenario allowing the overshoot of greenhouse gases by the end of the century; a consistent warming trend dominates across the region. The rate of temperature rise in such a case has been found to range up to 0.06–0.07 °C/year at places. Again, the spatial patterns of the temperature rise under RCP4.5 and RCP8.5 are much more similar, though their magnitudes are different.

Fig. 7
figure 7

Trends of near-surface air temperature (°C/year) over the period 1970–2099 over the Northeast India region as simulated from the REMO under RCP2.6 (a–d), RCP4.5 (e–h), and RCP8.5 (i–l) under CORDEX-SA experiment for DJF, MAM, JJAS, and ON seasons

Full size image

The area-averaged time series of near-surface air temperature for the period 1970–2099 including the present climate as well as RCP scenarios over the study area have been presented in Fig. 8 for all the seasons. A significant bias in the model-simulated climatology can be observed for the period 1970–2005, especially for the DJF, JJAS, and ON seasons. For all the seasons, the time series for RCP2.6 suggest that initially due to the emission of greenhouse gases the temperature rises till the middle of the twenty-first century but after 2060, the response of the climate system stabilizes itself by the end of the century. A large interannual variability in the temperature climatology in different RCPs is evident in all the seasons. However, the responses of climate system under RCP2.6 and RCP4.5 scenarios are similar to each other as they behave similarly till the middle of the century, and then RCP4.5 shoots up a bit to a comparatively warmer climate. It is seen that, while assuming RCP8.5 scenarios, with a slow rate of warming in the initial half of the twenty-first century, the temperature rises up quickly and continues to rise in the latter half without any moderating effect, unlike RCP2.6. It has also been seen that less interannual variability in simulating temperature has been shown by the REMO for the period 1970–2005 for DJF and JJAS seasons in comparison with the other two seasons (i.e., MAM and ON). It is also evident that RCP2.6 temperature climatology lies mostly under the mean ± 1 standard deviation of the present climate particularly during pre-monsoon and post-monsoon seasons suggesting a moderate response of the climate system towards the radiative forcing of 2.6 W/m2.

Fig. 8
figure 8

Time series of near-surface air temperature (°C) averaged over the Northeast India region for the period 1970–2099 for a DJF (December, January, February) season, b MAM (March, April, May) season, c JJAS (June, July, August, September) season, and for d ON (October and November) season from REMO forced with MPI-ESM-LR global model . The series in gray, green, dark blue, and red corresponds to present climate and RCP2.6, RCP4.5, and RCP8.5 warming scenarios, respectively. The dotted black line corresponds to the mean ± 1 standard deviation of the present climate period (1970–2005). The series in cyan represents the observation

Full size image

3.2 Precipitation

The climate of the NEI region is predominantly influenced by the monsoonal precipitation accompanied by some pre-monsoonal showers and periodic events of convective precipitations. The daily mean precipitation climatology from the observation and REMO experiment has been presented in Fig. 9. The climatology of the observation as well as the model shows maximum precipitation during monsoon followed by pre-monsoon showers being the second major source of precipitation.

Fig. 9
figure 9

Daily mean precipitation (mm/day) during 1970–2005 over the Northeast India region from APHRODITE dataset (a–d) and for REMO driven by MPI-ESM-LR global model (e–h) and their corresponding bias (i–l) under CORDEX South Asia experiment for DJF (December, January, February), MAM (March, April, May), JJAS (June, July, August, September), and ON (October and November) seasons

Full size image

The model is able to capture the spatial features of precipitation across the NEI region reasonably well. The precipitation associated with the upper reaches of Arunachal Pradesh and Sikkim and stretches of Khasi hills in Meghalaya is well captured in the model climatology. However, a strong bias in the simulated output for the present climate can be observed in Fig. 9i–l for different seasons. A substantial dry bias up to 5–7 mm/day over Assam, Meghalaya, and parts of Arunachal Pradesh has been observed during the monsoon season. At the same time, a wet bias of 6–7 mm/day has also been found in the northernmost reaches of Arunachal Pradesh in all the seasons which may be associated with the complex interactions over the higher elevations. In other seasons, the magnitude of precipitation bias is relatively less than the monsoon. The underestimation of precipitation during monsoon season has also been reported by Jacob et al. (2012) and Dash et al. (2006) and attributed to the insufficient characterization of monsoon features and the large-scale flow direction.

The mean annual cycle (figure not presented) of the daily mean precipitation suggests a bimodal distribution of the precipitation. The observation shows a peak of precipitation in the month of July, while for the model, the peak precipitation is simulated in the month of June; this suggests an early onset of the monsoon over the region. This might also be the reason for a considerable positive bias of precipitation in model-simulated climate in the pre-monsoon months.

The trends of daily precipitation (Fig. 10) over the NEI region for the period 1970–2005 suggest an increasing rate of precipitation (~ 0.03 mm/day/year) across the region during pre-monsoon season. Again, a slightly decreasing trend over the central Assam, Arunachal Pradesh, and the states of Manipur and Nagaland accompanied by increasing trends (up to 0.03 mm/day/year) over Meghalaya has been observed. Although the model is able to capture the spatial pattern of precipitation, it does not capture the trends of precipitation well.

Fig. 10
figure 10

Trends of daily mean precipitation (mm/day/year) during 1970–2005 over the Northeast India region from APHRODITE dataset (a–d) and for REMO driven by MPI-ESM-LR global model (e–h) under CORDEX South Asia experiment for DJF (December, January, February), MAM (March, April, May), JJAS (June, July, August, September), and ON (October and November) seasons

Full size image

To investigate the changes in precipitation in the future climate, percent change in the daily mean precipitation has been calculated for different emission scenarios. The changes in the precipitation in the near future under RCP2.6 have been presented in Fig. 11a–d. The projection under RCP2.6 suggests that a strong deficit up to 40% in the winter accompanied by an increase in the post-monsoon precipitation might be experienced by the NEI region in the near future. A mixed pattern of change including a slight increase and slight decrease across the region during pre-monsoon and monsoon months has been projected in the near future. When talking about the far future, the pre-monsoon precipitation along with the post-monsoon precipitation has been found to increase. The magnitude of the decrease of wintertime precipitation was found to be less as compared to the far future responding to the stabilized response of the climate towards the end of the century.

Fig. 11
figure 11

Changes in precipitation calculated as percentage change of near future (2020–2049) minus present (1970–2005) (a–d) and far future (2070–2099) minus present (1970–2005) (e–h) for RCP2.6 over the Northeast India region from REMO forced with MPI-ESM-LR global model for DJF (December, January, February), MAM (March, April, May), JJAS (June, July, August, September), and ON (October and November) seasons

Full size image

Due to stronger warming under RCP4.5, the wintertime precipitation has been found to decrease strongly in both near and far future (Fig. 12). A significant increase in post-monsoon over the parts of Assam and Meghalaya in the near future along with a decrease up to 20–25% over southern and easternmost parts of the region has also been found. Again, similar to RCP2.6, pre-monsoon (monsoon) season precipitation shows a moderate response to warming with a small increase (decrease) in the near (far) future.

Fig. 12
figure 12

Same as Fig. 11 but for RCP4.5

Full size image

The appraisal of changes in the precipitation under RCP8.5 has been presented in Fig. 13. RCP8.5, being the strongest emission scenario, suggests a strong decrease in DJF precipitation under near and far future (up to 60%). In the near future, a slight increase (5–10%) in the precipitation may be experienced in pre-monsoon and 25–30% increase in the far future. It is noteworthy that, under the far future, a general decrease amounting up to 10–20% has been found to occur in the pre-monsoon, monsoon, and post-monsoon seasons.

Fig. 13
figure 13

Same as Fig. 11 but for RCP8.5

Full size image

The long-term trends of daily mean precipitation in the winter season for the period 1970–2099 over the NEI region suggest a general decreasing trend with a rate of up to − 0.006 to − 0.01 mm/day/year at several places under different RCPs (Fig. 14). An increasing trend of precipitation with magnitude 0.01 mm/day/year has been found to occur in the pre-monsoon season under RCP2.6 over the eastern and northernmost parts of the region. Conversely, RCP4.5 projects mostly decreasing trend except in pre-monsoon over upper reaches of Assam and Arunachal Pradesh. RCP8.5 projects a mixed pattern of trends in pre-monsoon, monsoon, and post-monsoon seasons.

Fig. 14
figure 14

Trends of daily mean precipitation (mm/day/year) over the period 1970–2099 over the Northeast India region as simulated from REMO under RCP2.6 (a–d), RCP4.5 (e-h), and RCP8.5 (i–l) under CORDEX-SA experiment for DJF, MAM, JJAS, and ON seasons

Full size image

The area-averaged long-term time series of the seasonal mean precipitation has been presented in Fig. 15. A large interannual variability associated with the model simulation is evident in the area-averaged seasonal mean of precipitation in all the seasons except the JJAS.

Fig. 15
figure 15

Time series of total seasonal precipitation (mm) averaged over the Northeast India region for the period 1970–2099 for a DJF (December, January, February) season, b MAM (March, April, May) season, c JJAS (June, July, August, September) season, and for d ON (October and November) season from REMO forced with MPI-ESM-LR global model. The series in gray, green, dark blue, and red corresponds to present climate and RCP2.6, RCP4.5, and RCP8.5 warming scenarios, respectively. The dotted black line corresponds to the mean ± 1 standard deviation of the present climate period (1970–2005). The series in cyan represents the observation

Full size image

Again, a large year-to-year variation in the seasonal mean precipitation during DJF and ON seasons has been found to occur by the end of the century under the least warming scenario RCP2.6. Under the RCP4.5 and RCP8.5, the seasonal mean precipitation has been found to decrease in all the seasons by the end of the twenty-first century. An overestimation of the total seasonal mean precipitation, in all the seasons except monsoon, has been simulated by the model output.

4 Summary and conclusions

The NEI region is characterized by its distinct geographical setting known for enormous natural resources and a wide variety of endemic flora and fauna. The complex topographical setting accompanied by the extreme variation in the elevation and aspect gives rise to peculiar climatic regimes from east to west across the region. In the current study, an effort has been made to assess the recent changes in the precipitation and temperature regime over the NEI region followed by the changes in the projected future climate using regional climate model (REMO) under the CORDEX South Asia experiment framework.

The mean climatology of the near-surface air temperature and precipitation suggests that the REMO is able to capture the spatial patterns of these two variables reasonably well similar to observations. The linear trend of temperature for present climate (1970–2005) indicates a warming trend for all the seasons except for pre-monsoon. The changes in the temperature climatology under different RCPs suggest warming of the winter season in the near as well as far future in a unanimous fashion. The warming of the winters is also confirmed from the long-term (1970–2099) trends of temperature across all the three emission scenarios. This may lead to higher temperatures during the winter season with further implications for the regional-scale climate and associated agriculture, lifestyle, and health of the people residing in the region. Further, in this context, it becomes necessary as well as interesting to assess the change in the minimum and maximum temperature regimes along with the possible changes in the diurnal temperature range over the study area. A significant bias and a considerable interannual variability in the temperature have also been observed in the model; thus, before using these projections for any planning and policy executions, such shortcomings should be addressed.

During the present climate (1970–2005), an increasing trend of the daily mean precipitation has been observed for the pre-monsoon season followed by a decrease in the monsoonal precipitation in many parts over NEI region. This increase may be attributed to the changes in the pre-monsoonal convective activity, driven by the evaporation and evapotranspiration changes. However, this needs to be confirmed through a process-based study over the study area. For the future climate, a consistent decrease in the wintertime precipitation in all the scenarios has been found to occur. The magnitude of the decrease may range up to 40–60% in different scenarios. A slight increase (decrease) in the near (far) future in the precipitation has been observed in RCP4.5 and RCP8.5 during pre-monsoon (monsoon) seasons across the study area. This behavior needs further investigation, using a theoretical mechanism to explain such changes in the future. In general, a decreasing trend of the precipitation has been found to occur in the long-term climate especially in the stronger RCPs, i.e., RCP4.5 and RCP8.5. However, a strong bias followed by a large year-to-year variability has been noted in the model-simulated precipitation regime; thus, more appropriate measures should be taken in order to use the climate projections in a better way.

Under the purview of the current study, an effort has been made to summarize the recent climatic changes over the NEI region and also to foresee the future changes in the climate regime under different emission scenarios using a regional climate model. The signals of past and future climate changes may have a significant impact on the population in many ways such as effects on agriculture, water resources, health, and flora and fauna, and extreme weather events across the region. However, with several shortcomings, a number of uncertainties may be associated with the single-model projections; thus, for more accurate projections of future climate, an ensemble projection approach may be adopted, though a single regional model experiment provides a baseline on the changing climate across the study region. Further, more dedicated studies over the NEI region are necessarily required to confirm the obtained climate change signals in order to lead to better mitigation and adaptation measures in response to the changes in the climate system.