Geospatial Assessment of Agricultural Drought Vulnerability Using Integrated Three-Dimensional Model In The Upper Dwarakeshwar River Basin In West Bengal, India

It is essential to measure the degree of agricultural drought vulnerability in an underdeveloped rain-fed agro-based economy at the local, regional, and national level. Agricultural drought has become a major concern in respect to the global food crisis for investigation and development of a sustainable agricultural system that sustains the food security of a country. In this research, delineation of the agricultural drought vulnerability (ADV) status has been carried out by multidimensional mixed-method index approach using remote sensing and geographic information system. An integrated three-dimensional model has been adopted for this study. The three indices of this model are - Exposure Index (EI), Sensitivity Index (SI) and Adaptive Capacity Index (ACI). Exposure Index has been calculated using NDDI, LULC, ADI, DF, ADD and PI. Sensitivity Index has been calculated using satellite-based remote sensing factor VHI, NDWI, EVI, NDVI, VCI, NDWI, LST, and TCI. The ACI has been formed by combining the Environmental Adaptive Capacity (EAC), Social Adaptive Capacity (SAC) and Economical Adaptive Capacity (EcAC) Index. Each index has been computed by assigning the weights based on their relative importance by using the Analytic Hierarchy process (AHP) approach. Final results were classied into ve vulnerability zones, e.g., very low, low, moderate, high, and very high covering an area 362.32 km 2 , 186.68 km 2 , 568.69 km 2 , 547.05 km 2 and 266.89 km 2 respectively. Finally, results have been validated with long term Aman paddy yield data (2004 to 2014) through the Yield Anomaly Index (YAI).


Introduction
Drought is a least known complex climatic phenomenon that affects the Earth's environment (Boken et al. 2005; Badamassi et al. 2019; Pandey et al. 2012), that impacts on the economy of people who are directly and indirectly dependents on agriculture and applied activities (Wilhite 2000;Nagarajan 2009). This crisis is gradually increasing due to climate change and global warming (Bates et al. 2008). Recently, population growth and increasing demand of water for domestic use led's toward water shortage for agriculture all over the world (Dembele and Zwart 2016).
West Bengal (WB) is a predominantly agricultural based state of India. Almost 70% population of WB directly or indirectly depends on agriculture for their livelihood, whereas two-thirds of agricultural land is still dependent on monsoonal rainfall for cultivation (Ghosh 2019). The western parts of WB, mainly Bankura and Purulia districts are prone to this recurring drought phenomenon (Bhunia et al. 2020). So, the agriculture is the worst drought-hit sectors in this region (Palchaudhuri and Biswas 2019). The crop production is disrupted due to insu cient root zone soil moisture availability during the growing period of crops (sowing to maturity). As a result, agricultural drought is one of the serious threats to the agro-based rural economy of WB (Dutta et al. 2015).
Agricultural drought vulnerability is the signi cant aspects of drought management and monitoring for both short and long-term strategies (Murthy et al. 2015;Karet al. 2018). Drought vulnerability differs from one region to another region due to physioclimatic conditions (Sehgal and Dhakar 2016).
Accurate assessment of agricultural drought vulnerability can reduce the probable threat in agricultural sector. So, vulnerability assessment is essential step to increase economic development and agricultural stability in developing countries like India. There Although it is noticeable that there is no well accepted universal method is exist to forecast agricultural drought vulnerability. But, recently remote sensing (RS) and Geographical information system (GIS) based work is playing a signi cant role to monitor agricultural drought forecasting (Chockalingam et al. 2015;Hundera et al. 2016;Sanchez et al. 2018;Trnka et al. 2020). This paper focuses on multi-dimensional holistic index based approach using RS, GIS and AHP techniques constructing integrated Agricultural Drought Vulnerability Index (ADVI) based on three dimensional models in the upper Dwarakeshwar river basin in West Bengal (India) and also to evaluate the geo-spatial distribution of agricultural drought.

Study area:
The upper Dwarakeshwar river basin is situated in the eastern lower part of Chota-Nagpur plateau. The basin comprises a bordering area amidst the two districts (e.g., Bankura and Purulia) which are located in the western province of West Bengal, a state in the eastern part of India. The study area lies between 23°08´58.80 N to 23° 31´55.88 N and 86° 30´52.43 E to 87° 09 13.34 E, covering an area of 1934 sq km. The basin is situated in the middle part of the Puruliya district comprising the Kashipur, Puncha, Hura, Santuri, Para and Raghunathpur-I blocks while in Bankura district it covers the Chhatna, Bankura-I, Bankura-II, Indpur, Onda, Gangajalghati and Saltora blocks ( Figure.1). The region receives an average annual rainfall about 1528.37 mm. It is also noted that about 80% of the rainfall is recorded due to the in uence of south-west monsoon (June to October). Here, Kharif is the main cropping season and the major crops are Aman paddy that is normally cultivated during this time. On other side, Wheat, mustard, till and potato are cultivated in winter season as robi crops.

Data used
To perform work, different types of data (e.g., remote sensing, metrological, morphological, soil, drainage, groundwater, and socioeconomic status) have been used in calculation of the aforesaid three composite indexes. The detail description of data source and their speci cations for Agricultural Drought Vulnerability Index (ADVI) are given below in Table 1.  Table 2. Them, multi-dimensional index-based integrated ADVI has also been measured using this formula [ADVI= (EI + SI) -ACI]. The details methodology of this research is shown using ow chart (Pic. 2). Table 2 Selected parameters for agricultural drought vulnerability assessment in the upper Dwarakeshwar river basin. Each individual composite index was developed by following formula: Where, nCI is the three composite index of vulnerability, Wa is the each factor assigned of weight, and Ra is the Relative rating weights of the pair-wise comparison values under a classi ed factors.
AHP technique is build up by two leading segments. The primary segment is the prime scheme of Normalized pair-wise comparison matrix and was calculated by the weights for each factor. The secondary segment is calculated by the relative rating weights of all the factors into sub-classes by using pair-wise comparison matrix of each factor. To create a matrix of pair-wise comparison, each criterion is assigned against the other criterion by allocating a relative rank on Satty's scale (Saaty 1980), between 1 (minimum signi cant) to 9 (maximum signi cant) ( Table 3). The relative scales of all these factors are given based on different criteria, relative in uences, preferences and importance etc. In the Pair-based comparison matrix, each parameter in the row follows the opposite value and its signi cance with the other parameters. Weights for every factor ware obtained from pairwise comparison matrix by normalizing the values and this was determined by dividing each cell with corresponding sum of the column and then averaging the rows of each criterion. The general pair-wise comparison matrix P1 is constructed as follows, At last, the consistency of the Pair-based comparison matrix is assessed by Consistency Ratio (CR). CR is calculated by the following equation: Consistency Ratio (CR) = CI/RI Where, CI =Consistency Index and

RI =Random Index
If, CR is less than or equal to 0.1, the comparison matrix is considered as consistent, else it will be corrected.
The Random Index (RI) value obtained from the Satty's standard RI table, which is shown in Table 4. The Consistency Index (CI) is applied and it is calculated using the following equation-

Consistency index (CI) = ((γmax -n)∕(n -1))
Where, λmax = the principle Eigen value of matrix. n = Number of parameters used in the analysis. Here, all the index parameters of normalized weights value shown in Table 5 and sub-classes weights value of all the parameters shown in Table 6 & 7.  Figure   3) to create the exposure index by using GIS based AHP technique. The Parameters of evaluation exposure indexes have been shown in Table 8. Sensitivity in vulnerability assessment is a measure of how much the local climate will change in vulnerability to during drought.
Sensitivity is assessment of the susceptibility of moisture stress or water threads for agricultural drought. Vegetation health and freshness, soil moisture, soil temperature, evaporation, and transpiration are all critical factors for assessing agricultural drought sensitivities. Here the Sensitivity index is created using the satellite based remote sensing factor VHI, NDWI, EVI, NDVI, VCI, NDWI, LST, and TCI ( Figure 4) and all thematic layers has been prepared and resampling in 30 m spatial resolution through GIS environment. Parameters of evaluation sensitivity indexes have been shown table 9. In addition, a combination of eight different indicators has been developed using data from various sensors like Landsat and Modis in 2016 to better understand the agricultural drought sensitivity. Adaptive capacity elaborates the e ciency of acclimatize power. So, Adaptive capacity provides the ability to recon gure with minimal loss of resilience, environmental, Ecomonic, and human socio-economic system functions. An adaptive capability includes social and technological skills and strategies that allow multiple individuals or groups to adjust the environmental and socio-economic changes. In the context of the food system, adaptive capacity is usually developed or deployed to maintain livelihoods, food production or food access.
In eld of drought vulnerability, Adaptive capacity is the inherent strength of the agricultural area to cope with the reduction of the crop productivity and probable loss in the agricultural drought. Here, The ACI is a composite index of three indices, namely Economic Adaptive capacity (EcAC), Environmental Adaptive Capacity (EAC), and Social Adaptive Capacity (SAC).
The SAC and EcAC data was normalized by using the following equations-If p has positively related to vulnerability then pn = (pα -pmin)/(pmax -pmin) And if p has negatively related to vulnerability then used pn = (pmax -pα )/(pmax -pmin) Where, pn is normalized parameters, pα is each individual parameter, and pmax and pmin respectively represent maximum and minimum value of each parameter.

Environmental Adaptive Capacity (EAC)
The environmental elements control the amount of potential damage from a potential hazard or disaster and also build the EAC index. To measures EAC index, average groundwater depth, rainfall, drainage density, drainage buffer, soil drainage, NDVI, aquifer media, soil texture, soil depth, relative relief, elevation, slope factors have been used and it has also been attached by AHP technology. All collected data has been thematically mapped in GIS platform at 30 m spatial resolution ( Figure.   The adaptive capacity of a society is created by bringing together the social elements that empower the society from a single disaster. Social adaptive power controls the severity and duration of any kind of catastrophe. Social infrastructures such as education, health, labor force, unity, technology and productivity have the power to control the consequences of any kind of disaster. Here, to diagnose social adaptive capacity, six parameters have been used, such as, agricultural labor density, farmer density, rural literacy rate, old-age dependency population ratio, rural health facility, and population density ( Figure.6). Parameters of evaluation SAC has been shown table 11.
The SAC index has been constructed using the GIS overlay method with AHP based assigned weightage on the thematic layers of all the permits based on their normalized value. Thematic layers are farmers, agricultural labor, rural literacy, population density, old age dependency population and health. The EcAC Index is formed by the elements that Economically control the ability to adapt any kind of natural phenomena. EcAC index depend on various natural factor which is important index to determine region-based agricultural droughts. These are the road density, drinking water facility, irrigation area, agricultural area, total crop production, ratio of seed stores and livestock for determining the EcAC of agricultural drought in the agro-based upper Dwarakeshwar River Basin which is showing in gure 7. Table. 12. Pisciculture is an alternative source of income, so higher Piscicultural density areas indicate higher probability of drought adaptive capacity.

Positive
Road density Length of road / total area A higher Road density indicates higher probability of drought adaptive capacity.

Ration of seed stores
No of seed stores/ total area A higher seed stores density area indicates higher probability of drought adaptive capacity.

Ratio of fertilizer depots
No of fertilizer depots/total area A higher ratio of fertilizer depots indicates higher probability of drought adaptive capacity.

Results
The GIS-based 3 indicators have been used to assess agricultural drought vulnerability. The vulnerability of agricultural drought depends on the regional distribution of these three indicators such as, EI, SI and ACI.  (Fig. 9). The Sensitive Index of the Deep Green Region, which is sparsely scattered in the northern and western parts of the study area, is very low, and northern and western region falls within the low sensitivity index.
Moderate and high sensitive index has been seen in the middle portion of the study area. Areas with a very high sensitive index are distributed scatterly throughout the study areas which are shown by red color.
Adaptive Capacity Index (ACI) (Fig. 10) these three indices namely EAC, SAC, and EcAC (Fig. 11)  Higher adaptive capacity indicates lower vulnerability. Adaptive capacity can be controlled through environmental and sustainable socio-economic development. Sensitivity can be changed by changing the type of crop. But the exposure is associated with patterns of rainfall. So, it is not possible to control it. However, real time forecasting can be of some bene t.

Validation
There is no universally accepted, accurate, and direct mechanism has not developed for determining the validity of agricultural  (Fig. 13). It can be seen that almost every block has a negative YAI in the dry year of 2010 and a positive YAI in the normal wet year of 2012. The result shows that the entire region is a drought vulnerability region, which validates the prepared ADVZ.

Discussion
The above agricultural drought vulnerability analysis and mapping of the study were revealed for agricultural drought management purposes. This AHP and GIS-based unique methodology was used to reveal the EI, SI, and ACI that determine the basin-scale drought vulnerability. Here, the principal exposure factor is rainfall that can determine the water scarcity due to precipitation de cit. Calculation of drought duration, intensity, and frequency of SPI over the 10 years' time period was done. On the other hand, satellite-based NDDI disclosed the drought condition of the study area. LULC is showing that the region is highly dependent on an agro-based economy. As a result, when drought phenomenon happens, the probability of potential losses of the agricultural sector in this basin is relatively very serious. The sensitivity of this area is mainly dependent on vegetation, soil moisture, and temperature. These are assessed by using remote sensing-based index meanly VHI, NDVI, EVI, NDVI, VCI, and TCI.
TCI materially monitors the surface temperature conditions. NDWI essentially monitors the soil moisture and vegetation index basically monitoring the greenness and health of crop conditions which can assess water stress on the crop over the whole seasons. The total adaptive capacity index demonstrates a region's capability to defend drought vulnerability.
The exposure index and the sensitivity index of region together increases the vulnerability to drought, while the adaptive capacity index builds the capacity of drought tolerance which reduces the drought vulnerability. The agricultural drought vulnerability map showed that Indpur, some part is puncha and para, northern part of Saltora, western part of Onda, middle portion of Chhatana, north western and middle parts of Kashipur block have high vulnerability for agricultural drought due to very low adaptive capacity. Western Hura, south western Kashipur, south eastern Saltora and Gangajalghati, eastern Onda, north eastern Chhatna block has moderate vulnerability for agricultural drought due to low sensitivity, low adaptive capacity, high to moderate exposure index. Bankura-I and Santuri block has under the low vulnerability zone because moderate to high adaptive capacity and moderate to low sensitivity index. Bankura -II block has very low vulnerability due to very high adaptive capacity and low sensitivity index. So assessment of agricultural drought vulnerability adaptive capacity is a principal controlling factor that regulating to decrease the agricultural drought vulnerability. The irrigation area density, rural education and ground water depth, also drainage and soil condition, agricultural labour density, crop production density, agricultural are density are the important regional parameter for agricultural drought vulnerability assessment. According to regional nature of upper Dwarakeshwar river basin that region is under highly agricultural vulnerable when drought occurs. So, 1) To minimization of the ratio of agricultural dependent population, 2) spreading of economic diversi cation, 3) promotion of drought resistant crop farming and 4) increase investment for replace the traditional irrigation system to modern (sprikel, drip, pipe, in ltration) irrigation system are the main way to reduce the ADV. That is why; diagnostic assessment of three indices-based agricultural drought vulnerability map of agrobased economic region is required for a drought reduction plan.

Conclusion
Drought is a major hidden catastrophe of agricultural production in any region of the world. As a result, due to climate change, agricultural drought risk management is very necessary for maintaining food security. The current study has accepted a three dimensional holistic perception for assessment and spatial distribution of agricultural drought vulnerability map. Agricultural drought vulnerability considers address the multidimensional nature of multiple parameters such as exposure factors like LULC,NDDI and daily rainfall based 3-month SPI(drought duration, intensity, frequency), sensitive factor like NDVI, NDWI, VCI, TCI, VHI, EVI and adaptive capacity factors like Economic, social and environmental adaptive capacity. Different input indicators were selected by studying the various vulnerabilities related to climate and these inputs are managed in a systematic way to diagnose the spatial distribution of agricultural drought vulnerability. Weights were selected based on the ability of the parameters involved in the subjectivity to affect that particular index. Finally, an agricultural drought vulnerability map has been created using those three indices. Ultimatly vulnerability maps will assist in drought management, identifying agricultural areas affected by extreme drought. This will greatly bene t the planners and government o cials in formulating government policy for local scale drought management strategy. It also demonstrates the effectiveness of remote sensing and GIS-based three-dimensional methodology for identifying drought-related stresses in the agricultural economy. Thus, by modifying or directly using this methodology, it is possible to assess the agricultural drought in any part of the world and to formulate management policies based on it.       Agricultural drought vulnerability zones.