Exploratory Data Analysis of Climatic Trends in the Koyna Biodiversity Hotspot

The climate is a vital part of the Earth's system, and its significance for nurturing forest ecosystems cannot be undervalued. We are observing shifts in the patterns of the climatic cycle, leading to increased intensity that consistently surpasses the average threshold values of different climatic variables with every passing year, it is essential to study how the climate is influencing the forest ecosystem and identify the trends and major shifts observed during a specific period of time. “Climate-change impacts have led to visible shifts on the terrestrial systems in many parts of Asia in the phenologies, growth rates and distributions of plant species” (IPCC COP 27, Chapter 10-Asia) [1]. Extirpation risks in percentages of taxa (plants, amphibians, reptiles, birds, and mammals) for 2℃ and 4.5℃ global warming in Asia's 'priority places' without adaptation by the 2080s show 18.8% extirpation at 2℃ and 41.67% extirpation at 4.5℃ in the Indian Western Ghats Biodiversity Hotspots (IPCC COP 27, Chapter 10-Asia) [1].

The study performed by Vose et al. [2] stated that climate directly affects forests by impacting their makeup, behaviour, and well-being. Alterations in weather like rain and temperature can directly influence forests, but the outcomes vary based on location and the type of forest. Certain outcomes can be positive, while others might be negative. For instance, warmer conditions could boost tree growth in specific forests but reduce it in different ones. In many studies, the temperature and precipitation relationships are referred in any climate studies while other parameters like relative humidity and surface pressure (atmospheric pressure) are often kept aside though all these four have a crucial interplay in driving the climate of a particular region.

The climatic trend analysis study is carried out on the data of Koyna Wildlife Sanctuary (WLS) which is in Satara District of Maharashtra State. Nandargi and Mulye [3] studied the interplay between Intensity of Indian Summer Monsoon Rainfall at the Koyna during the period from 1961 to 2005. While Jha et al. [4] studied the geothermal gradient of boreholes to estimate the past variations in the climate. Bajirao et al. [5] used remote sensing and geographical information system technology to conduct an area-altitude analysis of erosional topography investigation of Koyna River basin. While the Koyna region has seen various research studies in the fields of seismology, Dam structure, and geology, the focus on Climatic Studies remains relatively limited. Aside from the seasonal rainfall analysis conducted by Nandargi and Mulye [3], there is a notable gap in research concerning the region's climatic factors and their impact on the forest ecosystem. Therefore, it is imperative to gain a comprehensive understanding of the trends in climatic variables and their influence on the region's forest ecosystem. The study by Patil et al. [6] state that on August 5th, 2019, the regions of Sangli and Karad (Satara District) experienced heavy rainfall and a sudden surge in water discharge from the Koyna Dam, surpassing 50,000 cusecs. Similarly, the Chandoli Dam released approximately 20,000 cusecs of water, causing disruption in communication between numerous villages in Sangli and Kolhapur District. These unprecedented events led to the first-ever devastating floods in the history of western Maharashtra. The floodwaters submerged vast areas in the vicinity of the river and its tributaries, severely impacting the three districts and causing significant devastation. Table 1 summarizes the recent studies done on the Koyna WLS with respect to climate study.

Table 1. Research studies done in Koyna WLS with respect to climatic parameters research

Table 2. Research gap with respect to parameters considered in previous studies

What is Indian Ocean Dipole (IOD)?

The Indian Ocean Dipole (IOD) is a climate phenomenon characterized by changes in sea surface temperatures in the Indian Ocean. It is similar to El Nino and La Nina events, but it is specific to the Indian Ocean region. The Indian Ocean Dipole (IOD) has two phases: a positive phase with warmer seas in the west and cooler waters in the east, and a negative phase with the opposite pattern. These sea surface temperature anomalies can have significant effects on weather patterns, influencing things like rainfall and temperature in the surrounding areas [8-10].

Furthermore, the Indian Summer Monsoon (ISM) exhibited remarkable behaviour, being abnormally dry in June but becoming impressively wet by September. The positive IOD event, which was characterized by sea surface temperature anomalies in the Indian Ocean, had a substantial impact on weather patterns in the region.

The research gap with respect to study performed till date in Koyna WLS can be seen in Table 2 and parameters that are considered most and least in the studies till date.

According to “Expert Study Committee Report: FLOODS 2019 (KRISHNA BASIN) by Water Resource Department, Government of Maharashtra, India” [8], the region experienced persistent and intense rainfall activity, coinciding with an active monsoon spell that commenced on July 27, 2019, affecting Maharashtra. This resulted in significant floods in Konkan and North Madhya Maharashtra. Following that, on August 3, 2019, another vigorous monsoon spell began, drove by the emergence of a low-pressure system over the Northeast Bay of Bengal. In the next days, the system deepened into a deep depression and travelled westward, causing catastrophic flooding in South Madhya Maharashtra. The deep depression system moved across central India, characterized by an active low-pressure system with wind gusts. This movement had a significant impact on Madhya Maharashtra's west coast and ghat areas, resulting in increased rainfall. These areas saw heavy to very heavy rainfall and even extremely heavy rainfall events over the period of more than a week, culminating in serious flooding. The prolonged and heavy rain during this timeframe exacerbated flood conditions and provided considerable problems to the impacted areas.

Ratna et al. [9] performed a study to better comprehend the underlying causes that contributed to the abrupt increase in heavy downpours in 2019. The study's findings confirmed the presence of a positive Indian Ocean Dipole (IOD) event in the year 2019, which was one of the strongest in the recorded history.

The Koyna Wildlife Sanctuary (WLS), recognized for its rich forest vegetation comprising Southern Tropical Evergreen Forest and Southern Moist Mixed Deciduous Forest [11], stands as a vital ecological and hydrological asset. In 2012, UNESCO designated it as a World Heritage Site, acknowledging its global significance for biodiversity conservation. Additionally, the Government of India had earlier, in 1985, declared it a protected area, underscoring its environmental importance. The sanctuary holds the status of an IUCN Category 4 Site [12], emphasizing habitat and species management for conservation. Central to the sanctuary's ecological significance is the Koyna Dam on the Koyna River, a major tributary of the Krishna River. This dam not only serves hydrological purposes but also houses a hydroelectric power plant with a 1920 megawatts capacity, marking the intersection of environmental conservation and energy generation. Despite its recognized importance in various domains, including hydrology and seismology, there remains a gap in comprehensive climatic studies within this region. This oversight highlights the urgent need for detailed climatic trend analysis to understand and mitigate the impacts of climate change on this critical biodiversity hotspot, making the present study both timely and essential.

The reasons that justify this research are: 1) The study region lacks substantial investigations concerning climatic variables with a focus on temporal variability, a primary aspect addressed in this study. 2) The chosen study period of 41 years is suitable for climate change assessment and to understand the trend of variables. Throughout this research, the study intends to answer the following questions: 1) How did the variables that are selected in the research area shows a trend from 1981 to 2022? 2) When was the significant shift in climatic intensification seen during a 41-year period, and how constant has it been since then? 3) Is there an impact of Indian Ocean Dipole (IOD) Effect on any identified climatic intensification?

In this research, the Rainfall (mm/day), Temperature (Max, Mean, Min), Relative Humidity (%) and Surface Pressure (hPa) variables are considered, and data is taken between the year 1981 to 2022 to understand the climate change trend in climate over the region. In this study, daily data for a single year is collected for each variable, and a correlation analysis using Pearson's coefficient (r) is conducted, visualized through heatmaps. This analysis aimed to explore the relationships and interplay between these factors, shedding light on their potential influences on weather patterns and conditions. To understand the overall trend of the correlation (r) values, a line plot was generated, encompassing data from 41 years for each variable pair. Interestingly, the year 2019 stood out as a pivotal point, exhibiting a significant shift in climate patterns, notably with the highest recorded values for Temperature (max, mean, min), and Rainfall. After this shift, both variables have displayed a consistent growth pattern. These findings underscore the importance of investigating the impact of climatic changes on the studied variables over the past four decades.

Table 5. Man-Kendall test result for ERA5-Land at annual timescale dataset from 1981 to 2022

The Mann-Kendall test results as seen in Table 5 for the rainfall, temperature (max, mean, min) data show a statistically significant trend at the 5% level. The p-value of these variables are almost 0.002 is less than the chosen significance level of 0.05, providing strong evidence against the null hypothesis and supporting the presence of a meaningful trend in the data. The surface pressure and relative humidity data Mann-Kendall test results show that the trend is not statistically significant at the 5% level. The p-value of 0.385 and 0.14 respectively is higher than the 0.05 significance level that was selected, indicating that there is not enough evidence to rule out the null hypothesis. As a result, the study is unable to draw the conclusion that the surface pressure and relative humidity data across the research period show any apparent trend yet remains constant with few fluctuations observed.

3.2 Correlation analysis using Pearson coefficient (r)

The below Heatmap, Line plot representation of Pearson coefficient (r) for Annual 41 years of data calculated on daily timescale dataset are generated that can be seen in Figure 3 and Figure 4 respectively.

This study has conducted a correlation analysis among the parameters considered, namely Relative Humidity, Temperature (Max, Mean, Min), Surface Pressure, and Rainfall. The results, visualized using a heat map as shown in Figure 3, revealed intriguing relationships between these climate variables. The climate factors under the 41 years (1981 to 2022) study showed several strong relationships. Notably, the relationship between rainfall and humidity had a substantial positive correlation (r = 0.53), indicating that more rainfall likely to be accompanied by higher humidity levels. On the other hand, Surface Pressure (r = -0.67) and Temperature (Max) (r = -0.53) revealed considerable negative relationships with Relative Humidity (r = -0.53). This shows that maximum temperature and surface pressure tend to drop as relative humidity rises. Furthermore, a strong negative connection (r = -0.51) between rainfall and surface pressure was discovered, showing that larger rainfall totals correspond with lower surface pressure readings.

There is a minor tendency for relative humidity to rise when mean temperature rises, according to a weakly positive correlation between relative humidity and temperature (mean) that was found (r = 0.092). On the other hand, rainfall, and temperature (mean) showed a minor negative association (r = -0.2), indicating that larger rainfall levels are connected to marginally lower mean temperatures. Despite the existence of these connections, they are somewhat weak, suggesting that other factors may have a stronger impact on the associations between relative humidity and mean temperature as well as between rainfall and mean temperature. The reason to perform correlation analysis and generate the heat map was to collect the Pearson’s coefficient value (r) of different relationships between parameters and generate the trendline of 41 years how this relationship varies and its highest value and lowest value to confirm that particular year event and climatic variables trend.

3.png

Figure 3. Heatmap representation of Pearson coefficient (r) for daily timescale data of climatic variables for Koyna WLS from 1981 to 2022

4.png

Figure 4. Line Plot Trend for Pearson coefficient (r) generated from Heatmap for the years from 1981 to 2022

The line plot in Figure 4 illustrates a notable decline in the correlation between Rainfall and Relative Humidity in the year 2015. Remarkably, this decline coincided with a period of reduced rainfall and drought-like conditions not only in the study region but also across the state of Maharashtra. In 2016, the correlation between Rainfall and Relative Humidity experienced a sudden increase and gradually continued to rise until 2019, reaching its highest strength. This upswing in the correlation aligned with the occurrence of a devastating flood in Western Maharashtra in the same year, 2019. The flood resulted from excessive rainfall, causing major districts like Satara, Sangli, and Kolhapur to be inundated as rivers overflowed. These findings suggest a potential link between the correlation patterns of Rainfall and Relative Humidity and the observed extreme weather events, highlighting the importance of understanding these relationships for climate monitoring and disaster preparedness.

3.3 Scatterplot for the years from 1981 to 2022 on daily timescale dataset

To gain insights into the relationship and trendline, Google Collab is used to generate scatter plots from a 41-year ERA5-Land dataset. The data, which has been assembled into a single file, has 15340 data points for each variable, allowing for a detailed evaluation of the interplay between the variables listed. These scatter plots help to validate the variables data values and its authenticity by understanding the theoretical predefined laws and relationship by practically generating scatterplots and check on the nature of relationship between two variables as they align with existing theory and teaching of atmospheric variables dependency.

For example, when the rainfall is expected, one of the favourable conditions is the high relative humidity, resulting a strong downpour which is experience in the daily interactions with climate. From Figure 5, scatterplot1 (Humidity vs Rainfall) denotes a directly proportional trend as relative humidity approaches 100%, the amount of rainfall also increases. However, it is crucial to note that relative humidity alone is not sufficient for rainfall to occur. Other factors, such as the formation of lowered pressure levels, play a significant role in precipitation events. In combination, these variables contribute to the conditions necessary for rainfall to occur. Scatterplot3 illustrates a pattern of higher rainfall associated with a decrease in pressure levels, while depicts that lower surface pressure corresponds to higher relative humidity percentages as seen in scatterplot 2. Scatterplots 2 and 3 display declining trend lines, indicating an inverse relationship between the variables. Specifically, as surface pressure levels decrease, rainfall tends to increase, and as surface pressure decreases, relative humidity tends to increase.

The dynamic relationship is observed in scatterplot4 between mean temperature and rainfall, as well as mean temperature and surface pressure in scatterplot 5. Higher rainfall is associated with a natural cooling effect on the atmosphere, leading to a decline in mean temperature. Conversely, lower surface pressure levels correspond to an increase in mean temperature. While scatterplot6 shows a constant trend between relative humidity and mean temperature. These figures demonstrate how changes in rainfall and surface pressure can influence mean temperature, contributing to the intricate interplay of these variables in the climate system.

5.png

Figure 5. Scatterplot for different atmospheric parameters from ERA5-Land dataset on daily timescale 1981 to 2022

3.4 ECWMF ERA5-land data plot

Figure 6 shows significant rise in rainfall pattern where year 2019 have shown highest rainfall on the other hand year 2015 shows a significant concerning drop in rainfall which was indeed a year of drought too. From Figure 7, the month of July receives highest rainfall and even seen lasting till October and slightly dropping in month of November. And hereby, there is extended rainfall, and October heat getting skipped since 2019.

6.png

Figure 6. Precipitation - ECMWF ERA5-Land daily 1981 to 2022

7.png

Figure 7. Precipitation monthly average- ECMWF ERA5-Land monthly 1981 to 2022

Figure 8a, Figure 8b and Figure 8c show a constant rise in all the temperature variables (maximum, minimum and mean respectively) indicating a significant turning point in the climate dynamics of the ECMWF ERA5-Land daily timescale dataset (1981-2022), particularly since 2019. Notably, the year 2019 had the highest annual rainfall of 3281.23mm, while the year 2015 had the lowest at 1655.11mm. Based on the monthly average of the daily timescale data, the trendline analysis shows an overall linear increase in rainfall, with July having the greatest average rainfall of 814.68mm. While Table 6 highlights the significant observations from the mentioned figures.

8a.png

(a)

8b.png

(b)

figure_8c.jpeg

(c)

Figure 8. (a) Temperature- maximum (ECMWF ERA5- Land) 1981 to 2022; (b) Temperature- mean (ECMWF ERA5-Land) 1981 to 2022; (c) Temperature- Minimum (ECMWF ERA5-Land) 1981 to 2022

Furthermore, 2019 stands out as a year with a significant increase in maximum temperatures above the 35°C threshold. This rapid spike is concerning for the forest ecology, which often sees temperatures ranging from 24℃ to 25°C. Furthermore, the minimum temperature, which ranged from 15℃ to 18℃, increased significantly in 2019. Similarly, the maximum temperature climbed above 33℃. While all three temperature points show a linear increase, 2019 holds out with a sharp and rapid surge. These findings highlight the need of monitoring and comprehending climate change, particularly the significant changes reported in 2019, which may have ramifications for the forest ecology and local climate.

3.5 NASA GPM_3IMERGM v06 data plot

To cross validate the rainfall trend with ERA5-Land results, NASA GPM_3IMERGM v06 data from 2000 to 2022 is plotted.

9a.png

(a)

9b.png

(b)

Figure 9. (a) GPM annual data (2000-2022); (b) GPM seasonal data (2000-2022)

Source: NASA Giovanni

Figure 9a using GPM data reveals a significant shift in the pattern of highest rainfall occurrences, particularly in the year 2019. Notably, a similar trend is observed, where 2015 recorded the least rain, followed by a sudden rise in 2016. Subsequently, in 2018, rainfall decreased again, though slightly higher than 2015, before spiking to the highest in 2019. The subsequent decline in 2020 suggests a cyclical pattern of high and low rainfall years, each with an equal interval of new values surpassing the previous ones. This emerging trend signals a potential change in the rainfall patterns, as indicated by the GPM merged satellite-gauge precipitation estimate final run. However, due to the availability of data only up to 2020, further trend analysis for 2021 and 2022 remains uncharted.

In Figure 9b, similar trend to Figure 9a is observed, reaffirming the cyclical pattern of rainfall variations. Particularly noteworthy is the consistency of June, July, and August (JJA) as the months with the highest rainfall, which aligns with the cyclical pattern seen in Figure 9a trend. This suggests that the cyclical fluctuations in rainfall are more pronounced during the JJA months, with these months consistently experiencing the highest levels of rainfall during the observed period. The overlapping trends in both figures further support the notion of a developing rainfall pattern and its association with specific months.

The findings of this study have centred on the examination of long-term climatic trends of dataset of the one of biodiversity hotspot, Koyna Wildlife Sanctuary (WLS), encompassing variables such as Rainfall, Temperature, Relative Humidity, and Surface Pressure. The methodology employed offers a sophisticated framework for undertaking analogous climatic trend analyses in diverse biodiversity hotspots that currently lack such investigations. Importantly, this approach is not restricted solely to the specific sites but can be universally applied to all biodiversity hotspots. In instances where ground observation data is scarce, the study emphasizes the use of experiential validation through a retrospective examination of past events within the biodiversity hotspot. This methodology holds promise for enhancing our understanding of climatic trends in various ecological and biodiversity hotspots worldwide.

During the analysed period from 1981 to 2022, the data of an area under study exhibited a general rise in the average values of Rainfall and Temperature (Max, Min, Mean). In contrast, Surface Pressure remained relatively consistent, exhibiting no notable fluctuations. Relative Humidity, on the other hand, displayed a stronger dependency on Rainfall, Surface Pressure, and Maximum Temperature, collectively affecting its fluctuations.

Interestingly, the year 2015 experienced a weaken annual trend of Pearson coefficient (r) correlations. This indicated a decline in the interplay between the climatic variables during that period, which resembles a pattern observed back in 1987. In contrast, the year 2019 displayed a stronger trend in the interplay between the variables with highest annual rainfall of 3281.23mm throughout the 41 years of time span. This figure even slightly surpassed the annual rainfall of 3272.82mm recorded in 2006.

In 2019, significant changes in weather patterns occurred in the Koyna WLS, potentially impacting biodiversity, hydrology, and ecological balance as what South-western Maharashtra State experienced in 2019 Floods. These findings are consistent with a strong positive phase of the Indian Ocean Dipole (IOD), one of the strongest in historical records, validating the study's reliability. Additionally, the prolonged Indian summer monsoon since 2019, leading to the absence of the October heat transition, which now may pose a risk for ecosystems, agriculture, and soil conditions due to disrupted climate patterns.

This study's findings on climatic trends in the Koyna Wildlife Sanctuary are subject to limitations, notably the lack of long-term ground observation data, which could enhance the precision of satellite-derived measurements. The use of retrospective event analysis, while insightful, cannot replace the subtle understanding that continuous on-site monitoring would provide. Additionally, the resolution of satellite data may not fully capture the microclimatic diversity of the sanctuary's ecosystem. Future research would benefit from integrating higher-resolution ground data to validate and refine these findings, addressing the potential non-stationarity in climate behavior and its implications for the sanctuary's biodiversity.

Despite this constraint impacting the depth of this trend analysis, the research findings represent a crucial advancement in comprehending the changing climatic patterns within the biodiversity hotspot throughout the experiential validation of climatic events occurred in past. This additional data source will provide valuable ground-based insights. The present study remains of great importance, serving not only Koyna WLS but also as a valuable model for investigating climatic trends of various protected areas through a similar approach.