Assessment of Relationship between Climate Change, Drought, and Land Use and Land Cover Changes in a Semi-Mountainous Area of the Vietnamese Mekong Delta

Abstract

Agriculture in the Global South is innately susceptible to climatic variability and change. In many arid and semi-mountainous regions of the developing world, drought is regularly cited as a significant threat to agricultural systems. The objective of this study is to assess the impacts of climate change on drought and land use and land cover (LULC) change in a semi-mountainous region of the Vietnamese Mekong Delta. We assessed previous drought trends (1980–2020) and future drought in the context of climate change, in accordance with three selected scenarios from the Coupled Model Intercomparison Project Phase 6 global climate models which have recently been released by the Intergovernmental Panel on Climate Change (IPCC) (2021–2060) using the Standardized Precipitation Index (SPI). The change of land use for the period 2010–2020 was then assessed and the associated climatic variability explored. The results show that for the period 1980–2019, SPI 3 responds quickly to changes in precipitation, whereas SPI 9 showed a clear trend of precipitation over time. The first longest duration occurrence of drought for SPI 3, SPI 6, and SPI 9 patterns were respectively 15–16, 21, and 25 months at Chau Doc station, and respectively 11, 14–15, and 16–17 months at Tri Ton station. Future precipitation and both maximum/minimum temperatures are projected to increase in both the wet and dry seasons. In addition, for all-time series scales and climate change scenarios, the levels of drought were slight, followed by moderate. In the future, the humidity at Chau Doc station is expected to decrease, while the occurrence of drought events is expected to increase at Tri Ton station, particularly in SPI 6 patterns (110 drought events in 1980–2020, and up to 198 drought events in the future). Moreover, between 2010–2020, the agricultural land area was seen to decrease, replaced by non-agricultural land uses that were found to increase by 22.4%. Among the agricultural land area, forestry, rice crops, and upland rice were found to reduce by 7.5, 16.0, and 21.2%, respectively, while cash crops and perennial crops increased by 26.4% and 170.6%, respectively. Amongst other factors, it is concluded that the variability of climate has led to drought and thus impacted on the conversion of LULC in the study area. Due to low economic efficiency, changing climate conditions, and a lack of irrigated water, the area of rice crops, forestry, aquaculture, and upland rice decreased, replaced by land for orchards for fruit production and other cash crops.

1. Introduction

Due to its natural sensitivity to weather fluctuations, agriculture is one of the sectors most at risk from the effects of climate change and variability [1,2]. Therefore, there is growing concern that both food security and agriculture-dependent livelihoods may be at risk [3]. In particular, drought occurrences pose a serious risk to the livelihood security of rural families [4,5]. Droughts occur virtually in all climatic zones and negatively impact natural habitats, ecosystems, society, and the economy. They are considered as the most common and harmful natural hazard [6,7,8,9]. Droughts severely influence the socioeconomic state of Thailand, Cambodia, Laos, and Vietnam on a regular basis and affect about 85% to 90% of the livelihoods of people residing in rural areas in the Lower Mekong Basin (LMB) region [10]. The LMB has experienced several extreme drought events in recent decades, including in the years 1992, 1999, 2003, 2015–2016, and 2019–2020 [10,11]. These led to substantial economic losses, primarily due to crop damages. The most extreme drought in 2016 caused USD 1.7 billion in damages in Thailand and caused water shortages in 18 out of 25 provinces, also affecting 2.5 million people in Cambodia. In Vietnam, the drought in 2016 is estimated to have cost USD 669 million, while the estimated recovery costs amounted to approximately USD 1.5 billion [10]. According to the results of Wang et al. [12], who used multiple indicators based on eleven Coupled Model Intercomparison Project Phase 6 (CMIP6) global climate models (GCMs) to assess global drought characteristics, future droughts will increase in most parts of the world. However, few studies have focused on the seasonal characteristics of drought in the LMB, in particular using the new CMIP6 outputs [7].

Precipitation shows strong spatial–temporal heterogeneity under global climate change, and assessing the variability of future precipitation is essential to understanding the impact of climate change on a range of natural and socioeconomic systems [13,14]. The latest outputs of CMIP6 models, based on a new set of possible alternative future scenarios, is more suitable for use thanks to improved spatial specificity and less uncertainty of precipitation compared with the CMIP5 projections [15]. The Sixth Assessment Report (AR6) of the Intergovernmental Panel on Climate Change (IPCC) highlights that the global surface temperature was 1.09 °C warmer during 2011–2020 than 1850–1900, with a higher increase across the land (1.59 °C) than over the oceans (0.88 °C). Precipitation is a key component of the hydrological cycle, combining atmospheric and land surface processes, and there is likely to be an increase in globally averaged precipitation across the land since 1950, with a higher rate after the 1980s [16,17,18]. However, at a regional scale, precipitation has been considerably found to increase (decrease) at high (low) latitudes in recent decades [19]. Changes in precipitation have impacted both water resources supply and management and progress towards the 2030 Sustainable Development Goals adopted by the United Nations (UN) in 2015 [20,21]. Therefore, understanding and exploring the relationship between climate change, future drought risk, and land use and land cover is important for the overall sustainability of the LMB because of its high human population density, susceptible infrastructure, and poor planning of land use and management practices [22,23].

Several authors have assessed climate change and its impacts in Southeast Asia (SEA) recently. Fei Ge et al. [24] used outputs from 15 climate models from the CMIP6 to evaluate projected changes in precipitation extremes for SEA by the end of the 21st century and suggested that CMIP6 multi-model ensemble medians showed better performance in characterizing precipitation extremes than individual models. Moreover, projected changes in precipitation extremes increased significantly over the Indochina Peninsula and the Maritime Continent. Supharatid et al. [25] projected changes in temperature and precipitation over mainland SEA using CMIP6 models to highlight the threats to Cambodia, Laos, Myanmar, Vietnam, and Thailand. They found that the effectiveness of anticipated climate change mitigation and adaptation strategies under SSP2-4.5 results in slowing down the warming trends and decreasing precipitation trends after 2050. All these findings imply that countries of SEA need to prepare appropriate adaptation measures in response to climate change. Recently research focused on the impact of future climate change on river discharge in Ho Chi Minh City (HCMC) [26]. They used a calibrated SWAT to simulate the discharge under seven GCMs derived from CMIP6 under three SSPs, with results suggesting that the climate of HCMC will be warmer and wetter by the end of the 21st century.

It has been clearly demonstrated that a knowledge gap exists, with a lack of studies considering the downscaled climate change impacts on drought and land use and land cover change at smaller scales, i.e., the provincial level required for adaptation planning. Short- and medium-term drought indicators like SPI 3 or SPI 6 were frequently used in studies on drought and its effects on agriculture in Vietnam and other nations [1,2,27,28,29]. In addition, a variety of drought indices are employed to evaluate the severity of a region’s drought, but this requires more information [30,31,32,33]. The influence of SPI on LULC in the context of climate change is also found in many studies [33,34].

Short-term land use changes are often associated with human activities due to the development of society and change in land use according to market economy fluctuations in the absence of a long-term strategy [35,36]. In Vietnam, the two dominate land use functions are agricultural land and non-agricultural land [37].Usually there is a conversion of agricultural land for socioeconomic development purposes. Moreover, a change can occur within a land-use category (mainly within the agricultural land category). Such changes usually occur as a result of a land use type changing into a type of higher economic efficiency [38,39,40]. Often driving factors for land use change are low crop productivity due to erratic weather changes, irrigation water availability, soil suitability, and hydrometeorological and climate-related changes due to prolonged drought, saltwater intrusion, and sea level rise [41,42,43].

Therefore, the objective of this study is to assess the impacts of climate change on drought and land use and land cover changes (LULCC) in the upper section of the Vietnamese Mekong Delta (VMD). We first assess current and future drought in the context of climate change, according to three selected scenarios, using Standardized Precipitation Index (SPI). In addition, we explore change of land use for the period from 2010 to 2020 in An Giang province, taking the associated climatic variability into consideration.

2. Methodology

2.1. Study Area

An Giang is an upstream province of the Mekong River in Vietnam, with both plains and a semi-mountainous area concentrated in the districts of Tri Ton and Tinh Bien (Figure 1). An Giang has a natural area of 353,683 ha, equal to 1.03% of Vietnam’s total land area, and is ranked fourth among the Mekong Delta provinces in terms of total agricultural area. In total, An Giang has 296,719 ha of land used for agriculture, according to the official land use data of 2019 (including 242,337 ha of land for rice cultivation, 11,648 ha of land for other annual crops, 25,343 ha of land for perennial crops, 11,643 ha of forestry land, 5530 ha of aquaculture land, and 219 ha of other agricultural land) [44,45]. The province has a population of approximately 1,907,401, with an ethnic population of 91,408 Khmer concentrated in the two mountainous districts of Tinh Bien and Tri Ton [46]. In 2020, the total planted area in the province was 707,100 ha (of which 637,200 ha are rice, 50,600 ha are crops, and 19,300 ha are fruit trees), with 1582 ha being planted forest [44]. The majority of An Giang’s domestic and agricultural water needs are met by reservoirs in the semi-mountainous region [47]. As a result, changes in precipitation here have a significant effect on the local economy and livelihoods.

2.2. Drought Analysis

The Standardized Precipitation Index (SPI), developed by McKee et al. [27,48] is widely used worldwide [28,49,50,51]. In this study we used the SPI to calculate the impact of drought on agriculture and establish its frequency using various time steps, as well as for the two cropping seasons (summer–autumn and winter–spring). In order to assess the short- and medium-term drought levels, SPI 3, SPI 6, and SPI 9 tests were performed. Furthermore, the seasonal SPI was also calculated.

Assuming a precipitation series at a given timescale X = x₁, x₂, x₃, …, x_n, n = 1, 2, 3, …, n represents the length of the series, and the distribution function f (x, θ) can best describe the distribution of the precipitation series X.

$$X~f\left(x,\theta \right)$$

(1)

where θ is the parameter sets of the function f(x) that can be estimated based on the sample series X.

Then, the cumulative distribution function can be expressed as,

$$F\left(x\right)={{\displaystyle \int }}_0^{x}f\left(x,\theta \right)dx$$

(2)

Equation (2) can be used to determine the cumulative probability of each observed precipitation event, xi. The SPI then undergoes an equiprobability transformation from the cumulative probability to the unit variance, zero mean, standard normal random variable Z.

$$\mathrm{SPI}={\mathsf{\Phi }}^{-1}\left(F\left(X\right)\right)$$

(3)

where Φ⁻¹ (•) is the inverse of a standard normal cumulative distribution function Φ (•) with zero mean and unit variance. Considering the fact that some kinds of distribution functions are not defined for x = 0, such as the Gamma function, an adapted statistic H (x) can be expressed as Equation (4).

$$H\left(x\right)=q+\left(1-q\right)F\left(x\right)$$

(4)

where q is the probability of no precipitation, which can be computed by simply dividing the number of months (N) by the number of months (n) that have no precipitation (Vicente-Serrano, 2006) [52]. Equation (5) can then be used to describe the SPI computation process.

$$\mathrm{SPI}={\mathsf{\Phi }}^{-1}\left(H\left(X\right)\right)$$

(5)

In this study, we used the SPI-based drought classification for Vietnam, in which SPI = −0.25 signifies the onset of drought conditions [53,54]. Drought begins when the SPI first falls below the negative value of 0.25 or less and ends when the SPI returns to a positive value of 0.25 or higher (

Table 1. The classification of drought and wet according to the SPI index.

Drought Levels	SPI Values	Drought Levels	SPI Values
Normal	0–0.24	Normal	(−0.24)–0
Very slightly wet	0.25–0.49	Very slight drought	(−49)–(−0.25)
Slightly wet	0.5–0.99	Slight drought	(−0.99)–(−0.5)
Moderately wet	1–1.44	Moderate drought	(−1.44)–(−1)
Severely wet	1.5–1.99	Severe drought	(−1.99)–(−1.5)
Extremely wet	>2	Extreme drought	<(−2)

).

2.3. Climate Change Scenarios

This study utilized the recent output from the CanESM5-CanOE (2.81⁰ × 2.81⁰ grid) developed by the Canadian Centre for Climate Modeling and Analysis, Victoria, Canada. Data associated with this study is available from https://esgf-node.llnl.gov/search/cmip6/ (last accessed 10 September 2022). New outputs from CMIP6 have recently been released by the IPCC. While the CMIP5 projected the future climate based on greenhouse gas emissions represented by four representative concentration pathways (RCPs), namely RCP2.6, RCP4.5, RCP6.0, and RCP8.5 [55], the CMIP6 designed five scenarios called shared socioeconomic pathways (SSP), namely SSP1-2.6, SSP2-4.5, SSP4-6.0, and SSP5-8.5, to include socioeconomic factors such as the growth of population, economics, urbanization, and other factors [56]. The improvement of CMIP6 scenarios includes wider equilibrium climate sensitivity (ECS) with an increasing temperature range of 1.5–4.5 °C. The models in CMIP6 are expected to improve their capability and reduce uncertainty over the previous CMIP5 and CMIP3 [13,57].

In this study, the three scenarios, SSP1-RCP2.6 (SSP1-2.6), SSP2-RCP4.5 (SSP2-4.5), and SSP5-RCP8.5 (SSP5-8.5) were chosen for analysis [58,59]. Historical simulations (1850 to 2014) and projections (2015 to 2100) from CMIP6 GCMs were employed in this study, with the climatic variables consisting of daily maximum near-surface air temperature (tasmax, in K), daily minimum near-surface air temperature (tasmin, in K), and precipitation rate (pr, in mm/day). The study period was divided into three periods for equal comparison including historical (2000–2020), future (2021–2040), and (2041–2060). The mentioned climatic variables were downloaded by extracting the grid value that was the closest to the observation station grid point [7].

Since there is large uncertainty as to the extent to which climate change responses might alter emission scenarios, it should be noted that the selected scenarios (SSP1, SSP2, and SSP5) for this study are only two plausible descriptions of how future emissions might develop at low-average and high socioeconomic pathways. Precipitation output from this cannot be used due to biases between the simulated variables for the current (control) climate and observed values. The delta change approach was used to correct these biases by transferring the signals of climate change derived from a climate model simulation to the observed database, and this method is considered a better bias correction of precipitation on a monthly and seasonal basis [60,61]. In this study, the means were calculated on a monthly basis for each 20-year period of climate output. The 12 delta change factors for precipitation were used to perturb the observed database and calculated as follows:

$$\Delta P\left(j\right)=\frac{{\overline{P}}_{scen}\left(j\right)}{{\overline{P}}_{contr}\left(j\right)}$$

(6)

$$P_{\Delta }\left(i,j\right)=\Delta P\left(j\right)\times P_{obs}\left(i,j\right) \left(i=1~31;j=1~12\right)$$

(7)

where $\Delta P\left(j\right)$ is the delta factor; ${\overline{P}}_{contr}\left(j\right)$ and ${\overline{P}}_{scen}\left(j\right)$ are the precipitation in month j averaged for the 20-year control (2000–2020) and scenario (2021–2060) periods centered in 2040 simulated by the GCM; P_Δ(i, j) is the projected precipitation (bias-corrected precipitation) under the SSP1, SSP2, and SSP5 scenarios; P_obs(I, j) is the observed precipitation representing current climate; suffixes i and j stand for the i^th day and the j^th month, respectively.

For temperature, absolute change is used for the delta change factors, as follows:

$$\Delta T\left(j\right)={\overline{T}}_{scen}\left(j\right)-{\overline{T}}_{contr}\left(j\right)$$

(8)

$$T_{\Delta }\left(i,j\right)=T_{obs}\left(i,j\right)+\Delta T\left(j\right) \left(i=1~31;j=1~12\right)$$

(9)

where ΔT(j) is the delta change factor; ${\overline{T}}_{contr}\left(j\right)$ and ${\overline{T}}_{scen}\left(j\right)$ are the temperature in month j averaged for the 20-year control (2000–2020) and scenario (2021–2060) periods centered in 2040 simulated by the GCM; T_Δ(I, j) is projected temperature (bias-corrected temperature) under the SSP1, SSP2, and SSP5 scenarios.

The projected precipitation (bias-corrected precipitation) under the SSP1, SSP2, and SSP5 scenarios was then used to calculate the future SPI as presented in Equation (5).

2.4. Land Use and Land Cover Change

Data collected to assess the change in land use in the period from 2010 to 2020 in An Giang province included data, reports, and maps of land inventories from the Department of Natural Resources and Environment of An Giang province (DONRE). Then, we explored the change in land use through the synthesis, analysis, and construction of tables and charts to compare and evaluate the change in area of land use types by the descriptive statistics method. We also assessed the transition between land use types in the same period. At the same time, we explored the drivers of land use change. In addition, the map method has been used to normalize the data and overlap to determine the areas that have undergone change in the period from 2010 to 2020 using ArcGIS 10.6 software. The overlap diagram determining the land use change area for the period 2010 to 2020 with the GIS tool is depicted in Figure 2.

3. Results and Discussion

3.1. Current Drought Assessment

Current drought based on the time series at 3, 6, and 9 monthly SPI for the Chau Doc and Tri Ton stations are shown in Figure 3 and Figure 4, respectively. From Figure 3, it can be seen that the 3-month continuous drought (SPI 3) responds faster to changes in precipitation than the 6-month and 9-month continuous droughts, respectively. However, it is also indicated that the 6-month and 9-month continuous droughts show more clearly the seasonal variability of precipitation or events of drought or wetness over time. Comparing drought in Chau Doc and Tri Ton stations, it is clearly indicated that the drought SPI 3, SPI 6, and SPI 9 at Tri Ton station tend to gradually increase over time, with slightly less variation seen than those at Chau Doc station.

In addition, the frequency and duration of dry periods were determined using the SPI 3-, SPI 6-, and SPI 9-month drought time series from 1980 to 2019. Figure 3 and Figure 4 also show that the variation in drought and wet periods over time is greater in Chau Doc than at Tri Ton Station. As SPI decreases over time, Tri Ton station shows more variation than Chau Doc station. This means that this area was drier between 1980 and 2019. The standard deviations for SPI 3-, SPI 6-, and SPI 9-month, respectively, were 0.74, 0.60, and 0.50 at Tri Ton Station, and 0.70, 0.46, and 0.47 at Chau Doc Station. Three consecutive months were found to appear more frequently at both stations than six and nine consecutive months. However, the number of occurrences differed little between the 6-month and 9-month consecutive terms.

Details of SPI values for drought and wet durations for the 1980–2019 period at Chau Doc and Tri Ton stations are listed in

Table 2. SPI values for periods of drought and wet durations in the timeframe of 1980–2019 at Chau Doc meteorological station.

	Chau Doc
Characteristics	3-month	6-month	9-month
Number of SPI $\le$ −0.25	129	91	72
The first longest duration/occurrence times of drought (month)	15–16/2	21/1	25/1
Time period, first longest duration	2012–2014, 2014–2015	1991–1993	2013–2016,
The second longest duration of drought (month)	11–12/2	19/2	21–23/2
Time period, second longest duration	1990–1991, 2001–2003	2012–2014, 2014–2016,	1989–1991, 1991–1993
The third longest duration of drought (month)	8–9/3	15/1	14–16/2
Time period, third longest duration	1989–1990, 1992–1993, 2009–2010	2002–2003	2002–2003, 2012–2014
Number of SPI $\ge$ 0.25	135	133	123
The longest duration of wet	10	15/1	33
Time period, first longest duration (month)	2016–2017,	1988–1989	1998–2001,
The second longest duration of wet (month)	9	13/1	12–13/2
Time period, second longest duration	2008–2009	2007–2009	2007–2009, 2010–2011
The third longest duration of wet (month)	8	12/2	10/2
Time period, third longest duration	1999–2000	1999–2000, 2010–2011	1981–1982, 2016–2017

and

Table 3. SPI values for drought and wet duration in the period of 1980–2019 at Tri Ton meteorological station.

	Tri Ton
Characteristics	3-month	6-month	9-month
Number of SPI $\le$ −0.25	116	73	34
The first longest duration/occurrence times of drought (month)	11/1	14–15/2	16–17/2
Time occurrence first longest duration	1985–1986,	2003–2005, 2012–2013,	2003–2005, 2012–2014,
The second longest duration of drought (month)	8–9/2	12–13/2	14–15/2
Time occurrence second longest duration	1991–1992, 2003–2004,	1984–1986, 2009–2010	2014–2016,
The third longest duration of drought (month)	7/3	8–9/3	11–12/3
Time occurrence third longest duration	1989–1990, 2012–2013, 2015–2016,	2001–2002, 2011–2012, 2014–2016	1984–1986, 2009–2010, 2013–2015
Number of SPI $\ge$ 0.25	122	109	96
The longest duration of wet	10/1	15/1	19/1
Time occurrence first longest duration (month)	1987–1988	1980–1981	1998–2000
The second longest duration of wet (month)	9/1	10/1	18/1
Time occurrence second long duration	1998–1999	1998–1999	1980–1982,
The third longest duration of wet (month)	7/2	9/3	12/1
Time occurrence third longest duration	1980–1981, 2000–2001	1981–1983, 1987–1988, 2000–2001	1986–1988

, respectively. From Table 2 and Table 3, it can be seen that the longest durations of dry and wet periods were observed from 2013 to 2016 (25 months from October 2013 to June 2016) and 1998 to 2001 (33 months from March 1998 to July 2001), respectively, at Chau Doc station. The second and third droughts, on the other hand, lasted approximately as long as the first; specifically, the two occurrences of the second drought occurred for periods lasting between 21 and 23 months (February 1989 to July 1991 and September 1991 to January 1994). The first longest duration of drought for SPI 3, SPI 6, and SPI 9 patterns at Chau Doc station were 15–16, 21, and 25 months, respectively. At Tri Ton station, the first longest duration of drought for SPI 3, SPI 6, and SPI 9 patterns were 11, 14–15, and 16–17 months, respectively. At Chau Doc Station, the first longest duration of wet periods for SPI 3, SPI 6, and SPI 9 patterns were 10, 15, and 33 months, respectively, whilst at Tri Ton Station, the first longest duration of wet periods for SPI 3, SPI 6, and SPI 9 patterns were 10, 15, and 19 months, respectively. These results clearly indicate that the duration of drought and wet periods at Chau Doc is longer than those at Tri Ton.

3.2. Future Drought Assessment

3.2.1. Climate Change

Delta changes in the monthly precipitation extracted directly from the grid cells of the CanESM5-CanOE emission scenarios SSP1-2.6, SSP2-4.5, and SSP5-8.5 closest to the stations in the study area were used to generate the monthly precipitation series, which was then used to calculate the corresponding SPIs. Details on the monthly delta change factors of precipitation (%), Tmax (°C), and Tmin (°C) of the scenario period (2021–2060) compared with the control period (2000–2020) under scenarios SSP1-2.6, SSP2-4.5, and SSP5-8.5 in the study area (at Chau Doc and Tri Ton stations), respectively, are shown in Figure 5a,b. The results of downscaled precipitation revealed that future precipitation in the study area would increase in both the wet and dry seasons under the three scenarios, with the highest increases seen in the wet season (period from mid-May to mid-November) in the month June, followed by September and July. Although it was found that there was an increase in precipitation in the dry season, with the highest percentage found in the month February and January, it should be noted that the “real” amount of increased precipitation volume is very modest.

In addition, the maximum temperature was projected to increase both during the wet and dry seasons under all scenarios and at all stations. The dry season (mid-November to mid-May), showed the largest increase in maximum temperature. This increase in maximum temperature, consequently, would lead to an increase in potential evapo-transpiration (ETo). Thus, this would impact the agricultural water resources and in turn lead to the conversion/change of crops or land use in the study area.

3.2.2. Future Drought Assessment (SPI 3, SPI 6, SPI 9 Patterns)

The different time series (SPI 3, SPI 6, SPI 9) for monthly drought patterns at Chau Doc and Tri Ton meteorological stations using various climate change scenarios are shown in Figure 6 and Figure A1 with scenarios SSP1 2.6 (a), SSP2 4.5 (b), and SSP5 8.5 (c), respectively.

According to the precipitation data from the climate change scenarios for Chau Doc and Tri Ton stations, the level of precipitation tends to decrease, with only two extremely wet events occurring at Chau Doc station between 2020 and 2060. In comparison, Tri Ton station displayed significantly more events.

The results highlight that extreme wet or dry conditions are unlikely to occur in the VMD under the assessed climate change scenarios. There are multiple extremely wet seasons solely in the SSP1 2.6 scenario. However, in climate change scenarios, there is no discernible difference between SPI 3, SPI 6, and SPI 9. Additionally, the trend of SPI variation over time is ambiguous.

Table 4. Number of occurrences of drought and wet period in the different patterns of SPI using climate change data at Chau Doc station.

Categories	SSP 1_2.6			SSP 2_4.5			SSP5_8.5
	SPI 3	SPI 6	SPI 9	SPI 3	SPI 6	SPI 9	SPI 3	SPI 6	SPI 9
Extreme drought	0	0	0	0	0	0	0	0	0
Severe drought	2	0	0	2	0	0	2	0	0
Moderate drought	41	18	0	41	18	0	41	18	0
Slight drought	99	91	30	99	91	30	99	91	30
Very slight drought	0	0	0	0	0	0	0	0	0
Very slightly wet	29	68	91	29	68	91	29	68	91
Slightly wet	55	119	63	55	119	63	55	119	63
Moderately wet	64	24	6	64	24	6	64	24	6
Severe wet	25	2	1	25	2	1	25	2	1
Extreme wet	7	0	0	7	0	0	7	0	0

and Figure A1 show the number of occurrences of drought and wet periods in the different patterns of SPI using climate change data at Chau Doc and Tri Ton stations, respectively. It can be seen that for Chau Doc meteorological station in the future, the duration of the drought shows little change, while the wet occurrence time has decreased for SPI 3, SPI 6, and SPI 9 under scenarios SSP1-2.6, SSP2-4.5, and SSP5-8.5. Meanwhile, at the Tri Ton station, the humidity level has changed little, while the occurrence of drought increases, particularly in SPI 6 patterns. SPI 6 shows 110 drought events from 1980 to 2020, but this is forecasted to increase to 198 drought events in the future.

3.3. Land Use Land Cover Change from 2010 to 2020

Figure 7 presents the area of land use types in 2020 for An Giang province. It shows that agricultural land was the dominant land use, including specialized rice crops (triple rice and double rice), upland rice, cash crops, fruits, and other agricultural land, in which the area of specialized rice cultivation occupied 23.9 × 10⁴ hectares (ha) and was distributed mainly in the districts along the Tien and Hau rivers such as An Phu, Cho Moi, Phu Tan, Chau Phu, Chau Thanh, and Thoai Son districts, as can be seen in Figure 8. In addition, the area of fruits accounted for 5.0% (1.76 × 10⁴ ha) of the natural land area of the province; this type of farming has grown strongly in recent years because of its high economic efficiency compared to other agricultural land uses. Next, the area of cash crop land occupied 1.18×10⁴ ha; this was one of the land use types that farmers preferentially select to replace inefficient rice crops. Other types of land use showed a smaller area and were distributed widely (Figure 7).

The area of land use types of An Giang province in the period from 2010 to 2020 showed a gradual increase in the area of non-agricultural land (by 22.4%), as can be seen in Figure 9. The reason was the conversion of agricultural land to non-agricultural land that aimed to meet the needs of socioeconomic development of the province, especially to meet the needs for the development of infrastructure systems and residential areas due to population growth. Moreover, forest and rice crop land showed a decreasing trend. In contrast, the area of land for cash crops and fruits increased (by 24.4 and 170.6%, respectively). The above transformation trend was due to the fact that the production efficiency of rice and forest land was not high, due to the influence of acid soil, unusual weather due to climate change, and fluctuating market prices. As such, farmers have tended to switch to other crops which have higher economic efficiency such as fruits and cash crops.

Table 5. Land use change matrix for the period of 2010–2020 in An Giang province.

Year 2010 Year 2020	Non- Agricultural	Unused Land	Water Surfaces	Forestry	Rice Crops	Upland Rice	Cash Crops	Perennial Crops	Others Agricultural	Area 2010 (ha)
Non-Agricultural	26,118.8	8.6	1531.7	137.8	1464.8	94.5	432.4	1240.0	7.8	31,036.4
Unused Land	112.3	689.6	20.2	453.5	546.4	16.9	12.2	98.9	-	1949.9
Water Surfaces	3504.7	180.2	21,184.9	97.8	5570.2	61.1	736.6	849.5	15.1	32,200.1
Forestry	408.4	83.5	95.6	10,436.6	1726.1	117.7	322.8	684.6	-	13,875.2
Rice crops	5253.0	0.8	5941.0	160.2	22,5931.6	1495.6	5376.9	8621.3	51.9	252,832.3
Upland rice	397.9	10.4	66.2	77.1	1807.8	2469.0	524.6	296.7	7.4	5657.2
Cash crops	720.6	28.2	759.6	36.2	1510.8	54.6	3932.9	2278.0	-	9320.9
Perennial crops	1463.4	0.9	189.8	249.2	471.7	148.8	442.2	3530.0	10.7	6506.7
Others agricultural	9.7	1.4	2.1	0.5	47.4	-	-	5.2	84.7	151.0
Year 2020 (ha)	37,988.8	1003.5	29,791.1	11,649.0	239,076.5	4458.0	11,780.6	17,604.3	177.7	353,529.6

shows the area of rotation of different types of land use in the period of 2010–2020. As can be seen in Table 5, the transformation of land use types in An Giang province in the period from 2010–2020 shows widespread change.

Figure 10 presents the map of land use changes for the period of 2010–2020 of An Giang province. From Figure 10, it is indicated that the area of other land types, such as surface water (aquaculture), has been converted to other types of land use, such as rice crops, cash crops, and fruits (as can be also seen in Table 5). This may be due to the reduced economic efficiency of aquaculture in recent years, following the sharp drop in the selling price of products. The area cultivated for rice crops also shows strong a conversion to upland rice and cash crops due to drought, which leads to a shortage of water for irrigation in the dry season. In addition, the area of land use for fruits converted from other types of land use, mainly rice and cash crops that have not brought high economic benefits. In general, the trend of changing land use types in An Giang province in the period of 2010–2020 was consistent with the province’s socioeconomic development trend. However, the transformation was fragmented, not centralized, and difficult to manage and monitor for the local authorities.

The change of land use in the period from 2010 to 2020 was assessed, and when taking the associated climatic variability into consideration, it is shown that both maximum and minimum temperature were found to increase in both the wet and dry seasons under the three mentioned scenarios (especially in the dry season (mid-November to mid-May)); the increases in maximum temperature are found to be highest. The increase in the current drought frequency, as discussed previously, was found to be associated with LULCC. It is clearly indicated for the period of 2010–2020 that the area of agricultural land in An Giang province decreased and was replaced by non-agricultural land. This is due not only to its low economic efficiency, but also to changing climate conditions (especially max/min temperature) leading to a lack of irrigated water. Consequently, the area of rice crops, forestry, aquaculture, and upland rice was found to decrease within the agricultural land group (Table 5 and Figure 10).

4. Conclusions

For the period of 1980–2020, SPI 3 responded quickly to changes in precipitation, whereas SPI 9 showed a clear trend of precipitation over time. Moreover, for all-time series scales and climate change scenarios, the levels of drought in the study area were typically slight, followed by moderate. In the future, the wet period events at Chau Doc station are expected to decrease, while the occurrence of drought events is expected to increase at Tri Ton station in the semi-mountainous area.

Future precipitation and both maximum and minimum temperature are projected to increase both during the wet and dry seasons under scenarios SSP1-2.6, SSP2-4.5, and SSP5-8.5 at both Chau Doc and Tri Ton, of which, in the dry season (mid-November to mid-May next year), the increases in maximum and/or minimum temperature are found to be highest. This consequently would lead to an increase in potential evapo-transpiration (ETo) and impact on the water resources available for agriculture, and thus may ultimately lead to the conversion or change in crops or land use within the study area.

It is noted that in the period of 2010–2020, the area of agricultural land in An Giang province decreased and was replaced by non-agricultural land uses. Due to low economic efficiency, changing climatic conditions, and a lack water for irrigation, the area of rice crops, forestry, aquaculture, and upland rice tended to decrease within the agricultural land group, replaced by land for growing fruits and cash crops.

As only one RCM (CanESM5-CanOE) was considered in this study, future studies should investigate the outputs from other GCMs available from CMIP6 for the associated uncertainty analysis. In addition, the ensemble output from reliable GCMs based on their skill scores should be taken into consideration in the future.