Climate change intensifies the drought vulnerability of river basins: a case of the Magat River Basin

In the recent past, climate scientists have observed an increasing trend in temperature and have forewarned us that as we move forward (Levine & Steele 2021), it has become imperative for us to acknowledge the natural catastrophes precipitated from climatic variations induced by unprecedented global warming which reflects anthropogenic activities in terms of incremental carbon footprint and greenhouse gas emissions (Houghton et al. 2001; Jorgenson et al. 2019). The frequency of emerging climatic extremes disrupts environmental and societal processes caused by the deleterious impacts of such disasters. Calamities induced by natural extremes worldwide have caused billions of dollars worth of destruction, damage, and injury where about 20% of it is attributed to drought occurrences (Salvacion 2021). Climate change is not singled out to be the sole cause of hazards that severely affect socioeconomic, environmental, and other aspects but its concomitant impacts on environmental and natural resources, agricultural systems which threaten food production and food security, variations in weather, meteorological extremes, and even in human health, has been an extensive subject in climate change vulnerability assessment studies (Thornton et al. 2014; Franchini & Mannucci 2015; Ochieng et al. 2016; Raza et al. 2019). As climate change impacts almost all aspects of society (Carleton & Hsiang 2016; Cáceres et al. 2021), it challenges the resiliency of a particular system which exposes its vulnerability in terms of sensitivity and exposure elements that represent the potential harm it may incur. This also impacts its corresponding adaptive capacity (AC) measures that may alter the actual condition of the system so that when possible adaptation measures are considered, vulnerability is regarded as the representation of potential long-term effects of climate change on a particular unit of exposure (Gumel 2022).

Drought is a commonly recurring element related to climate. It is a natural peril affecting millions of households as it demands a difficult and sophisticated approach to scientifically address its inherent characteristics (Wu et al. 2011). Interdisciplinary measures have been applied globally to mitigate drought-related jeopardies around the world in terms of famine due to reduced production, causing food insecurity, economic losses, natural resources, and environmental systems degradation (He et al. 2019; Bandyopadhyay et al. 2020; Poonia et al. 2021; Wendt et al. 2021). There has been a huge collection of research and sustainability efforts to address the need for a more serious undertaking in drought studies. For detailed information, the reader is referred to Wilhite (2013) and Sheffield & Wood (2013). The Intergovernmental Panel on Climate Change Framework on Vulnerability assessment has become one of the most used tools in assessing the impacts of natural hazards. In their working paper, Brooks & Adger (2003) introduced a preliminary conceptual framework for vulnerability and adaptation that adapts to a variety of applications, systems, and hazards concerning climate change. Vulnerability is either perceived as a consignment of damage as an outcome of a specific climatic hazard or a superimposed aspect that thrives within the system to be stimulated when a climatic risk occurs (Moret 2014; Modica & Zoboli 2016). It is an indicative measure of the susceptibility of an area, with respect to the main focus of evaluation may it be more in an economic, environmental, social, or physical aspect (Dabanli 2018). Drought vulnerability assessment is contingent on the target sector to be evaluated and the geographical features of the area. In other words, drought vulnerability is always related to the way a certain system under a climatic hazard responds in a socioeconomical sense; climate change may also trigger or intensify poverty (Leichenko & Silva 2014). While drought can be correlated with the decrease in rainfall over an area, the real vulnerability might lie in how farmers incur an equivalent loss by not using a drought-resistant crop variety, or how much help a water-impounding reservoir could make if it is available. There are existing examples of multi-dimensional approaches and studies mainstreaming drought vulnerability assessment and mitigation efforts that in some sense complement each other (Naumann et al. 2014; Thomas et al. 2022; Zhou et al. 2022). It is now a globally recognized agenda to weaken the social, environmental, and economic impact of drought by promulgating progressive routes in making societies more resilient to drought risk and vulnerability (Naumann et al. 2014).

Agriculture is often regarded as the backbone of the Philippine economy as more than 30% of its land area is being cultivated. However, the Philippines is identified as being at the forefront of experiencing climatic extremes and disasters. Up to 327 million dollars worth of agricultural damages has been estimated from the El Nino incident from 2015 to 2019. This has caused small-scale farmers to be greatly affected in terms of their livelihood and survival needs (Peñaflor & Gata 2020). The frequency and intensity of climate-related disasters always deter a positive course of development (Perez et al. 2022). Different vulnerability studies and methodologies have been performed to create awareness of the impacts of climate-related disasters. Following the IPCC model, to address the consequences of numerous disasters on society, the development of a thorough method for evaluating vulnerability was done in the Philippines as part of disaster risk reduction efforts conducted by Robielos et al. (2020). Also, Perez et al. (2022) investigated the evolution of drought in the Philippines based on an El Niño event using strategically selected drought indices driven by satellite-sourced data. Related watershed-specific climate change vulnerability studies were also being conducted, as well as crop-oriented susceptibility assessment to crop-based climate-risk vulnerability assessment and climate change investigation under different climatic hazards which are more concerned with exhibiting multi-impact risk and vulnerability analysis (Bimmoy & Ongan 2014; Balderama et al. 2015; Tolentino et al. 2016).

The Cagayan Valley Region where the Magat River Basin (MRB) is situated is identified as one of the highly vulnerable regions in the Philippines (Yusuf & Francisco 2009). The MRB has been considered to be one of the 142 critical watersheds in the Philippines and the majority of its area is under protective custody to secure the health of the watershed. The Department of Environment and Natural Resources (DENR) is mandated to steward the basin's 4,143 km² as a forest reserve and the National Irrigation Administration has been given the authority to develop some 150 km² for irrigation and agricultural purposes (Department of Natural Resources 2016). MRB harbors the Magat Multi-Purpose Dam which plays a role in the economic development of the communities inside the watershed. To put it into perspective, the Magat River Integrated Irrigation System (MARIIS) alone is designed to service more than 860 km² of rice fields and farm-based aquaculture ponds. MRB has been so significant in providing opportunities for livelihood to farmers and fisherfolks and hence, making agriculture the primary endeavor, making the basin a prominent contributor in making the Province of Isabela the second top rice-producing province in the Philippines (Cañete & Temanel 2017; Tongson et al. 2017). MRB is no stranger to such climatic hazards that affect all the sectors and industry that benefits from its resources, and drought is considered to be a heavily devastating climatic hazard in the Philippines (Warren 2018). Climate change transmutes the occurrence, duration, intensity, and extent of drought. The slow-pacing onset of droughts can be observed and ranges over a considerable extent of time from a few months to even a couple of years. Drought impact is location specific; it challenges the stability of the basin to provide for its stakeholders and is considered to be a real problem in irrigated agriculture, particularly in rice production (Manalo IV et al. 2020). However, there is no distinct way to address drought vulnerability. It is an emerging concern to reduce the corresponding social and economic cost of drought and assessments of the corresponding economic damage should be made to evaluate the severity of drought impacts (Neri & Magaña 2016). The socioeconomic factor is concerned with the population of the area of interest, the cost of damage, the extent of lands affected, and the availability of adequate water supply (Jia et al. 2016).

Geographic information system (GIS)-based methods are widely accepted and established to be a great tool also in terms of data processing and calculation. GIS handles input data, its storage, management, processing, and analysis up to the production of output to be used in strategic frameworks and planning for risk and hazard mitigation and preparation. It plays a huge part in the development of highly essential maps, necessary for promoting social security and emergency responses of the concerned agencies for the benefit of the stakeholders (Karmakar et al. 2010). This study was conducted to benchmark and assess the impacts of climate change on the vulnerability of the MRB to drought by considering agriculture as the major sector of focus by employing a quantitative analysis of data available from concerned agencies and reports, and by applying a site-specific expert-judgment approach for agriculture derived from Macawile et al. (2018) and Hagenlocher et al. (2019) which have been tailor-made to tackle the genuine concerns of the stakeholders. Currently, there are no published drought studies conducted for MRB, and this study serves as a pioneering approach in quantifying drought impacts concerning the basin's vulnerability concerning agricultural socioeconomic aspects.

The degree to which climate change exacerbates drought impacts is often difficult to assess. The evaluation and development of different indices and vulnerability indicators have been increasingly well-known and are being extensively applied in disaster risk and sustainability studies (Birkmann 2007; Mori & Christodoulou 2012). The appropriateness of drought vulnerability indicators particularly in assessing the socioeconomic aspect of an agricultural system that heavily relies on water resources is not an easy task. In overcoming this challenge, experts in agriculture and related fields have been coordinated to evaluate and select the suitable drought vulnerability indicator for MRB based on Meza et al. (2019). This study uses the principles of Saaty's analytic hierarchy process (AHP) and entropy method (Saaty 1980; Zhu et al. 2020) of multi-criteria decision-making to reliably appraise the corresponding weight of indicators in both subjective and objective dimensions to appropriately assess the basin's sensitivity, exposure, and AC to drought. AHP is considered to be one of the oldest and most-effective decision-making techniques, which has been extensively applied in different disciplines (Ho & Ma 2018). It is relatively easy to use and is very useful in establishing the corresponding importance of criteria over the others, with a concomitant consistency value that indicates if the analysis is statistically acceptable. The ability to employ AHP in various studies significantly facilitates analyzing uncertain scenarios without sacrificing the subjectivity and objectivity of any evaluation criterion (Chakraborty & Joshi 2016). In the same light, the entropy method has been incorporated to emphasize the objective weights of the level of importance of indicators since entropy operates on the premise that criteria with higher weight indicators are more beneficial than information criteria with lower indicators (Kumar et al. 2021). The integrated AHP-Entropy process was conducted to effectively evaluate the vulnerability indicators in both a subjective and objective manner. The combination of these two MCDA methods accounts for the consideration of the existing pool of knowledge based on the expertise and experience of the experts on the subject matter in evaluating the level of importance of the vulnerability indicators but at the same time, objectively getting rid of human influence in the final weighting process (Nyimbili & Erden 2020). For the benefit of catalyzing appropriate policy framework and decision support systems, this study will also include vulnerability projection which is explicitly affected by the future changes in the rainfall regime estimated utilizing the Representative Concentration Pathway (RCP)-based projection in the Philippines that was generated using the Climate Information Risk Analysis Matrix (CLIRAM) tool obtained from the Philippine Atmospheric, Geophysical and Astronomical Services Administration (PAGASA) that has been explicitly reporting the future changes in rainfall, temperature, sea level rise, and changes in tropical cyclone under different scenarios (DOST-PAGASA 2018).

The success of all drought vulnerability studies is dependent on the availability and quality of data. Data gathering might be the most difficult aspect or part of such studies but the adequacy of data determines the study output which will, later on, affect future policies that will be patterned based on the results generated from this research.

The vulnerability in this study was generally assessed per subbasin level. The data were gathered using paper-based survey questionnaires that were conducted on the field in person, readily available information from concerned agencies, and published information from credible sources.

In order to assess the land use and forest cover indicators, the latest up-to-date, 2015 land cover dataset was obtained from the National Mapping and Resources Information Authority (NAMRIA) where LandSat8 digital and visual image interpretation was used for land cover mapping, acquired from Earth Observing System's LandViewer (EOS). The 2015 land cover data have been validated in the field and yield an accuracy of 98% which implies that the land use and land cover have not significantly changed (WReDC 2021). The stream data have been identified through ocular field inspection and dialogue with the locals. The spatial data have been subjected to geoprocessing using GIS.

In order to determine the plant growth stage at the onset of a drought, the general cropping calendar with respect to the type of climate, type 3 in the case of MRB, and the quantity of agricultural land that is irrigation dependent, data were gathered from the Department of Agriculture, and the National Irrigation Administration where analysis was performed to consolidate and arrive at the required set of data requirements.

The majority of datasets needed to quantify the indicators of corresponding components of vulnerability, i.e. sensitivity, exposure, and AC, were gathered through field surveys where the farming households, as the primary stakeholders of the basin, were the clientele of focus. To gather the necessary information, the researchers conducted a survey in chosen localities per subbasin from 7 to 8 February 2022. The survey was carried out to collect key drought risk information, such as economic losses caused by protracted drought, as well as existing interventions and adaptive measures that assist farmers in fulfilling crop and water demands, such as irrigation infrastructure and crop types. Furthermore, the survey represents the socioeconomic metrics employed in this study's analysis to encourage proper accounting of the basin's susceptibility, exposure, and AC to drought from an agro-economic standpoint.

The weight of each component of vulnerability was subjected to both subjective and objective multi-criteria decision analysis (MCDA) in order to carefully account for the respective contributory factor of indicators in the total vulnerability of MRB. For hazard and disaster-related studies (Zeng & Huang 2018; Liu et al. 2019; Sahana et al. 2021; Li et al. 2022) involving multi-criteria analysis, two of the most frequently employed methods for weight estimation are the AHP (Saaty 1980), and entropy weighting method (Zhu et al. 2020).

Palchaudhuri & Biswas (2016) coupled AHP and GIS in assessing the drought risk in India. Similarly, AHP was employed in this study to establish pairwise comparison matrices among the indicators of drought dimensions to synthesize the vehemence of indicators against each other based on Saaty's scale of relative importance (Table 1). Separate analyses were conducted for each of the dimensions of vulnerability (sensitivity, exposure, and AC) to determine the respective weight of indicators based on pairwise comparisons. Here, the idea is to determine the level of sensitivity, exposure, and AC of the MRB subbasins using the indicators of the dimensions of vulnerability. In generating the sensitivity, exposure, and AC maps of MRB, a reliable weighting coefficient for each indicator is needed to show how an indicator is more important or of lesser importance than the other indicators.

In obtaining the final weights for the dimensions of vulnerability, this study employs the Entropy method, an objective, comprehensive, and widely used method of weighting the components of vulnerability by circumventing human influence. This measures the degree of variance in order to assess value. The greater the deviation of the measured value, the higher the level of differentiation of the indicator and the more information that may be extracted (Taheriyoun et al. 2010).

where is the measured value of the ith indicator in the jth sample and denotes the standardized value of the ith value in the jth sample.

Here, ⁠. And the =1.

Vulnerability is determined by the kind, amount, and pace of climatic change and fluctuation to which a system is subjected or exposed, as well as the sensitivity and AC of the system. Exposure and sensitivity are identified to be the implicit impact or corresponding influence of drought and AC serves as an opposing dimension that covers the coping mechanism and defense mechanism of the system.

From local to global dimensions, indicator-based techniques have been pushed as valuable tools for assessing, comparing, and monitoring the complexity of drought risk (Hagenlocher et al. 2019). The basin's vulnerability has been carefully attributed to three categories: sensitivity, exposure, and AC (Cui et al. 2010) which is adapted to the concept of vulnerability of the Intergovernmental Panel on Climate Change. Agriculture has been the main sector of focus since a considerable number of studies in developing frameworks, methods, and the selection of indicators have been conducted considering agricultural aspects (Shim et al. 2021) There is no standard for identifying hazard vulnerability indicators, as it is always affected by the specific characteristics that govern the local system (Zarafshani et al. 2016). Hence, the indicators were formulated based on an expert-judgment approach from the global expert survey results report by Meza et al. (2019) of the European Commission which includes relevant indicators for agricultural systems, which have been modified to select the indicators that are uniquely applicable to MRB. The report summarizes the pool of knowledge from an expert response of experts in drought vulnerability assessment studies working in academia, industry, and concerned agencies. It also gives a comprehensive overview of the sectoral drought risk assessment of agricultural systems which partly determines the appropriate indicators to be considered for MRB.

Sensitivity is consistently defined as the factor by which a certain system is positively or negatively affected by drought. Or in simple terms, sensitivity pertains to the environment since it includes social, economic, and ecological conditions driven by drought hazards. A higher sensitivity value of a system is equivalent to a higher exposure it will be subjected leading to a higher vulnerability.

For MRB, the sensitivity indicators were analyzed according to class and given a rating or rank, where the highest sensitivity rating is five (5) and one (1) is the lowest. Table 2 shows the different sensitivity indicators used in this study.

The factor of exposure is considered in previous and recent drought studies. The intensity, duration, or frequency of stress on a system is measured by exposure. From the term alone, exposure pertains to the state to which a certain system is exposed or remains in a condition to be affected by consequential drought. This study utilizes drought exposure indicators (Table 3) in order to evaluate the exposure category of vulnerability in MRB.

AC pertains to the coping mechanism and response of the system to lower or lessen the potential impacts of drought, which is considered to be the negative dimension in vulnerability studies since it is the countermeasure in mitigating possible drought consequences. It includes the construction of irrigation facilities, the introduction of crop-resistant varieties, cloud seeding, etc., whose ultimate aim is to alleviate drought implications. Modern vulnerability studies consider the concept of AC as the most important factor that separates it from earlier studies concerning climate-related vulnerability. It is said to be mainly defined by social, economic, and biophysical processes. The AC indicators in this study are shown in Table 4.

In order to make projections on the future climatic trend in the basin, the RCP-based projection report from DOST-PAGASA was used. The projections available in the Philippines that were generated using the CLIRAM tool that has been explicitly included in the report are changes in rainfall, temperature, sea level rise, and changes in tropical cyclones (DOST-PAGASA 2018). Also, from DOST-PAGASA (2018), the current climate projections for the Philippines were generated using regional circulation models (RCMs) such as Conformal Cubic Atmospheric, Providing Regional Climates for Impacts Studies (PRECIS) Model, RegCM4, and the HadGEM3-RA RCMs which were derived from different Global Circulation Models (GCMs) at approximately 100 km resolution and were downscaled at a 25-km resolution. These climate projections for the Philippines were based on the IPCC AR5 where climate scientists, modelers, and experts collaboratively catalyzed resolution-based datasets that have been downscaled on a 0.5 × 0.5 grid resolution (including land use data, air pollutants, etc.). The IPCC Expert Panel (Stocker et al. 2014) has chosen four scenarios based on radiative forcing level as influenced by greenhouse gas emissions and other contributory agents. For this study, we considered the RCP 4.5, a scenario that reaches a peak radiative forcing of 4.5 W/m² (2040) which declines and stabilizes in the year 2100 since it is the scenario that is most likely to happen considering the projected temperature trajectory in the Philippines (DOST-PAGASA 2018). We also considered the RCP 8.5 where the radiative forcing exceeds 8.5 W/m² by 2100 and will progressively increase with the same amount of time (van Vuuren et al. 2011; Hibbard et al. 2011).

Rainfall gauging stations are located in different provinces over the MRB. The downscaled (provincial-level) seasonal projections as provided by the CLIRAM tool were used to estimate the projected rainfall in the years 2036–2065 Mid-21st Century and 2070–2099 Late-21st Century.

The paper-based survey was conducted on 40 farming households that utilize water supply from the MARIIS. The response of the farmers has been gathered to further establish the sensitivity, exposure, and AC of the study area. The farmer respondents, of which 26% were female, all registered an additional source of income, mainly backyard livestock farming (e.g. native chickens, pigs, duck, small ruminants, cattle). One-third of the farmers said that they had a secondary income that supports them aside from farming. The farmer respondents use three main commodities, e.g. rice, corn, tobacco, and a considerable number of them venture into high-value vegetable crops. For corn, the average yield loss is tallied at 20% to as much as 60% in highly exposed areas. For rice, an average of 35% yield loss was also estimated based on the response of the farmers and the comparison of crop yield during normal cropping season to drought season. Tobacco farmers, however, incur some significant 30–60% yield and income loss when facing severe droughts even if they use drought-resistant varieties. For vegetable/high-value crop farmers, it was estimated that 20% of their normal income is being reduced during drought season, or in some isolated cases, the damaged crop does not do well under stressed conditions meaning that they abort farming for that particular season, resulting in a 100% loss of income. Only 21% of the farmer respondents have confirmed that they have crop insurance and only 42% are using drought-resistant varieties. All of the farmers have confirmed that they have access to forecasting and AGROMET stations and more than 95% of them undergo crop diversification practices. Most of the farmers (95%) have access to irrigation facilities where a considerable number of such irrigation sources are privately owned, e.g. deep well for tobacco, vegetable, and some rice farmers. Since drought is a natural phenomenon, the farmers are not blaming anyone, but are only vocal in expressing how the government and related agencies may help them. Their proposed interventions include the improvement of irrigation canals and facilities, agricultural subsidies, and crop insurance.

The raster data from NAMRIA have been subjected to GIS processing in order to acquire the needed land cover information in assessing the land use of the MRB. The rating of every respective subbasin is not far from the basin's sensitivity rating of 3.6 which takes place from moderate to high on the sensitivity scale. In terms of the percentage of forest cover, the MRB's rating is 2.85 or between the low to moderate sensitivity scale. Subbasin 2 has a high sensitivity since it only has 21–40% of forest cover, and part of the land area is developed for agricultural purposes as part of the 860 km² of the irrigated service area of the MARIIS. Ibulao and Matuno subbasins, however, have a very low sensitivity to drought which is attributed to more than 80% of its area being covered by forest and in the mountainous zone of Ifugao. Alimit is under some 41–60% of forest lands (Moderate), and Lamut Subbasin has very low drought sensitivity with a terrain consisting of 61–80% of forest. The presence of a perennial type of stream has been identified by field visits and ocular inspection supported by the responses of the locals verifying a year-long stream flow which has a corresponding sensitivity rating of 1 (very low). The general cropping calendar of the provinces consists of commodities including rice, yellow corn, and high-value crops like cabbage and potato. Alimit, Ibulao, and Lamut Subbasins have been identified to be subjected to a high sensitivity rating based on our data gathering and assessment, since the majority of the crops have been affected by drought during their maturity stage, while the Subbasin 2, Matuno, Santa Fe, and Subbasin 1 were rated at 5 (Very High). Eighty percent (80%) of the agricultural areas are dependent on irrigation. The Ifugao and Nueva Vizcaya areas also have newly built and improved irrigation facilities.

In calculating the current sensitivity of the basin, the 9-year SPI (Table 5) implies that 2014 was relatively dry in comparison to the other years. The monthly cumulative rainfall data in 2014 recorded by the rainfall stations were shown in Table 6. The mean annual rainfall harnessed from the rainfall stations over the MRB has a rating of 3.71 or less than 2,000 mm whereas Santa Fe and Subbasin 2 have an annual rainfall of less than 1,500 mm which in the drought sensitivity scale is moderate to high. The continuous dry days for the entire MRB having the 2014 drought as a baseline have lasted for 42–72 days which is classified as high where Subbasin 2 and Santa Fe Subbasin recorded more than 72 days without rainfall (very high); Ibulao and Alimit Subbasin sustained 30–48 days of no rainfall (moderate), and Lamut and Subbasin 1 fall on a high sensitivity scale having experienced 42–72 days of no rainfall.

The survey implies that the basin's exposure to drought has a mean value of 3.4 for the entire MRB or up to 40% of the total area. About 70% of the agricultural production areas have been affected by drought occurrences where Alimit, Ibulao, Ibulao, Lamut, and Matuno Subbasins have 41–60% (moderately exposed to drought) of the respective production areas affected by drought while Subbasin 1 and 2, and Santa Fe subbasins have a high drought exposure of 61–80% in terms of production areas affected. Yield and income losses for the whole MRB were estimated at about 30%. The Santa Fe and Subbasin 1, from consolidating farmer response, lose 31–40% of yield and income compared to production years without drought. Subbasin 2, however, having the Magat Dam as its main source of irrigation support, accounted for 11–20% losses (very low). The remaining subbasins have an estimated 21–30% loss which is considered to be moderately exposed to drought. Lamut, Subbasin 2, and Subbasin 1 were identified to have 40% of families that are highly dependent on agriculture.

The results of the study determined that each subbasin has an available irrigation system or water-impounding reservoir, making the MRB relatively low in terms of potential impact. Also, 21–30% of the farmers from Alimit, Ibulao, and Lamut Subbasins are doing some diversification and planting new crop varieties to explore opportunities for bigger profit. Meanwhile, there are 15% of farmers in Subbasin 1, Matuno, and Santa Fe subbasins that are practicing crop diversification. Subbasin 2 has less than 10% of farmers practicing crop diversification since the cropping pattern of planting a legume crop after two crops of rice is the conventional practice. The available maps of drought-prone areas are not specific but presented to be in general as part of a wider regional basis so a ‘partially available’ classification has been expertly assigned to it. Moreover, about 40% of the farmers confirmed that they are eligible for subsidies and are registered for crop insurance programs where Subbasin 2 has more than 45% of farmers with their rice fields insured. The other subbasins have about 25–40% of farmers with access to crop loans, insurance, and subsidies. Consequently, 40% of the total budget of agriculture is being utilized for drought response programs in order to alleviate the suffering of the farmers in terms of severely dry climate. The majority of the farmers also claim that they have access to weather information from AGROMET stations being broadcast and circulated locally. Also, in the case of those that are being supplied by the National Irrigation System they are observing a proper cropping calendar and since the MRB has a type 2 climate, the farmers follow a traditional cropping calendar for high-value crops and the MRB was partially covered with cloud seeding programs. The majority of the farmers said that they are planting drought-resistant crop varieties, especially rice. In Alimit, Ibulao, Subbasin 2, Matuno, and Subbasin 1, more than 40% of the total farmers are using crop-resistant varieties. About 35% of the farmers from the other subbasins say that they are using drought-resistant varieties, classifying it as having a high AC value.

The contemporary condition of MRB in terms of its level of sensitivity, exposure, and AC to drought, and its overall drought vulnerability serves as the baseline for future projection.

Using the CLIRAM Tool of PAGASA, the observed climate trends in the Philippines is only contingent on rainfall, temperature, sea level rise, and tropical cyclone. For this study, rainfall and temperature were defined as drought exposure indicators (see Supplementary Information). Given the projected percentage change of rainfall and temperature parameters, the projected drought exposure of the MRB under RCP 4.5 and RCP 8.5 scenarios for both Mid- (2036–2065) and Late- (2070–2099) 21st Century. The current drought exposure level of MRB is at 2.76–3.75 and is projected to have a minute increase to a value of 2.77 for RCP 4.5 Mid-Century (Normal), RCP 4.5 Late Century (Lower-bound and Normal), RCP 8.5 Mid-Century (Lower-bound), and RCP 8.5 Late Century (Lower-bound and Normal). But it is projected to have a minimum value decrease at 2.16 for RCP 4.5 Mid-Century (Lower-bound) and a 2.61 projected minimum exposure rating of 2.61 for RCP 4.5 Mid-Century (Upper-bound), RCP 4.5 Late Century (Upper-bound), RCP 8.5 Mid-Century (Normal and Upper-bound), and RCP 8.5 Late Century (Upper-bound). However, the maximum exposure rating of MRB in all future scenarios is projected to subside to as low as 2.4 (a decrease of 36%) for RCP 4.5 Mid-Century Scenario (Lower-bound) and at considerable decrease at 3.27, 3.44, and 3.45 exposure rating under all other scenarios (Table 7).

The vulnerability rating of MRB has observed a dramatic increase up to a maximum of about 30% beyond the baseline or the current vulnerability rating, for both RCP 4.5 and 8.5 climate change scenarios. The current vulnerability range of 1.9–3.29 is seen to have a minimum value of 2.62 under RCP 4.5 (Mid-21st Century Scenario Lower bound) up to a maximum value of 4.33 (Mid-21st Century Scenario Lower bound). For the RCP 8.5 Scenario, the least projected minimum is at 2.67 (Mid-21st Century Upper bound and Late-21st Century Upper bound) to up to a maximum value of 4.3 (Mid-21st Century Lower bound) (Table 8).

This study was conducted to assess the impacts of climate change on the drought vulnerability of MRB in terms of the social and economic features of agriculture-dependent communities. The results of the projection imply that due to the future changes in rainfall and temperature as attributed to climate change, the basin's overall vulnerability is projected to increase to up to 30% beyond the baseline. The potential impact of the drought that was assessed in the study is the combined effect of the basin's sensitivity and exposure to drought. According to Sahana et al. (2021), the AHP-entropy method of multi-criteria decision-making is one of the best-performing methods for drought susceptibility assessment studies. The AHP-entropy method was applied to this study and calculated the baseline value of the sensitivity and exposure of the basin accounting for 57 and 31% of the basin's overall vulnerability to drought, respectively. This may be attributed to the fact that MRB is subjected to Type 3 climate where the rice (lowland) cropping season usually starts in October; corn (dry season) starts in March; and another high-value commodity like cabbage is planted from January to March. All months were under subsequent drought stress having the cumulative monthly rainfall value of 2014 as the basis where relatively below-average rainfall has been recorded from January to June and the last quarter of the year. On the other side, the extent of drought areas is a difficult indicator to satisfy, since there is no specific local data per municipality in each subbasin that summarizes the extent of drought-affected production. In the study conducted by Macawile et al. (2018), they applied a focus group discussion approach (FGD) by coming down to the site and interviewing the farmers. There is an established interrelationship since sensitivity indicators mainly depend on the meteorology of the area, and the resulting exposure is highly dependent upon these indicators, e.g. the yield and income losses are two inseparable indicators since the latter is dependent on the sensitivity indicators. High-intensity droughts, for that matter, induce lesser yield, and lower-income secured (Tongson et al. 2017).

In hindsight, the vulnerability of the MRB is on a high scale as plausibly indicated by the corresponding values of the basin's sensitivity and exposure to drought. However, the AC values of 3.17–3.99 indicate a high effect of adaptive interventions against drought which makes the AC a negative dimension of vulnerability. Farming is the most prominent livelihood of the communities inside the basin, while most of the farmers are also engaged in other activities as an additional source of income. Farmers are considered to be the poorest of the poor (Philippine Statistics Authority 2017), and the ability to adapt is higher for wealthy societies or those who can afford climate protection than for less fortunate ones (Fankhauser & McDermott 2014). Based on the results of the study by Sahana et al. (2021) using the AHP-entropy method, the AC component has a constituted weight of only 12% of the total vulnerability value of MRB which is the least of all the vulnerability components, implying that as of now, existing policies and practices may still be improved. The entropy weighting method was applied for this matter because it is an objective weighting method and can eliminate human bias in assigning the corresponding weight to the categories of vulnerability (Li et al. 2011). AC in this context is the overall counter-measure of the basin to the potential impacts resulting from the combination of drought sensitivity and exposure. The corresponding value of AC indicators reflects the quality and effectiveness of the mitigation and in battling the potential impact of drought, i.e. sensitivity and exposure.

Communities seem to be unaware of the related impacts of climate change and how temperature and rainfall variation alone can potentially alter their state of living and traditional practices, which can be very alarming. Initial adaptive measures should not only be employed momentarily but should be sustainable in the sense that scientific-based practices should be downscaled to the farmers and related constituents. In their study, Manalo IV et al. (2022) found that the inability of farmers to adapt to climate change is also due to some non-climatic stressors. They stated that there are variables that do not appear to have anything to do with climate change mitigation and adaptation yet have a significant influence on how well-coping mechanisms function, and non-climatic variables that directly inhibit the efforts towards climate change adaptation. From the results of the study, the corresponding yield loss of the main commodities – rice and corn – can be as high as 60 and 20% for high-value crops which are mainly seasonal vegetables, while for some the severity of drought can totally annihilate their crops. This only implies that the current drought impacts directly affect the income of the farmers and their families, in addition to the income generated through paid labor during harvest season since the demand for farm workers consequently declines due to visible yield reduction. Crop yield, particularly of aerobic rice, is also expected to decline in the 21st century as a result of climate change (Alejo 2021). Since the MRB supports a wealth of farming families, the current condition in terms of the basin's response to drought hazards will not be enough for the future. Together with the increasing farm input and labor cost, the farmers can be put in a very difficult situation. Without up-to-date and innovative intervention, drought impacts on the socioeconomic welfare of the MRB communities can be adverse and further exacerbated by climate change.

The vulnerability value that was estimated by this study denotes that parts of MRB are highly sensitive and exposed to drought hazards, and that AC interventions that contribute to counteracting drought-related hazards are still pre-mature and will be continuously challenged by the changing climate. Improving drought forecasting and enhancement of models in accurately evaluating the nature and behavior of drought can serve as a great aid for policymakers and authorities in making crucial drought disaster-related decisions (Band et al. 2022). Furthermore, incorporating other drought indices in monitoring and hydrological drought that can provide a broader perspective on drought severity and its impact on a certain system is also crucial in drought assessment studies (Shamshirband et al. 2020). Other factors, such as population growth, that increase industrial, agricultural, and domestic water demand can also increase vulnerability (Sehgal & Sridhar 2019). Also, future development may impede natural processes that can reshape the system's operations and activity.

For future research endeavors, systematic consideration of factors to be assessed in drought and related studies may also try to look for other opportunities such as using different multi-criteria decision-making methods in weighting relative matrices for more transparent discrimination between variables intrinsic to vulnerability.

This paper emphasizes agriculture as the main sector in the MRB and the farmers as its major stakeholders through intelligent selection and consolidation of the indicators of the respective drought vulnerability components on the agricultural systems of the basin. The AHP-entropy weighting method and GIS-based calculations were successfully employed in estimating the MRB's sensitivity, exposure, and AC to drought at 57, 31, and 12%, respectively. Overall current drought vulnerability rating of the basin was estimated to be at a low to moderate level. However, under the RCP 4.5 Mid-21st Century (normal) climate change scenario, it is projected to increase to a moderate to a high level. This projection, however, is only based on the change in the exposure dimension due to the projected change in rainfall and temperature. The other factors that were not included were the expected increase of losses in yield and in income, the future extent of production areas to be affected by drought occurrences, and the future agriculture-dependent households to be affected by droughts, which are all as vital as the currently included indicators. This study serves as a guide to policymakers in conceptualizing management measures and devising plans to mitigate the impacts of drought extremes by carefully assimilating science-based studies in developing intervention programs to invigorate former AC measures and create new schemes to withstand the effects of climatic anomalies and extremes. This study also shows that the MRB's response to climate disturbances and extremes can be enhanced through the proper collaboration of stakeholders. Additionally, a locally-based drought assessment and analysis using ensemble techniques can give better and more reliable results. Such studies can be undertaken using this study as baseline data. Quantifying the major aspects of drought vulnerability and integrating climate change effects can improve mitigation measures to further benefit the stakeholders under climatic-induced threats and stresses. In a broader sense, this particular study of the MRB shows that climate change intensively stimulates the drought vulnerability of river basins from a socioeconomic perspective.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.