Ecosystem Service Modelling to Support Nature-Based Flood Water Management in the Vietnamese Mekong River Delta

Abstract

Deltas are among the most productive and diverse global ecosystems. However, these regions are highly vulnerable to natural disasters and climate change. Nature-based solutions (Nbs) have been increasingly adopted in many deltas to improve their resilience. Among decision support tools, assessment of ecosystem services (ES) through spatially explicit modelling plays an important role in advocating for Nbs. This study explores the use of the Land Utilisation and Capability Indicator (LUCI) model, a high-resolution model originally developed in temperate hill country regions, to map changes in multiple ecosystem services (ES), along with their synergies and trade-offs, between 2010 and 2018 in the Vietnamese Mekong Delta (VMD). In so doing, this study contributes to the current knowledge in at least two aspects: high-resolution ES modelling in the VMD, and the combination of ES biophysical and economic values within the VMD to support Nbs implementation. To date, this is the highest resolution (5 by 5 m) ES modelling study ever conducted in the VMD, with ~1500 million elements generated per ES. In the process of trialling implementations of LUCI within the VMD’s unique environmental conditions and data contexts, we identify and suggest potential model enhancements to make the LUCI model more applicable to the VMD as well as other tropical deltaic regions. LUCI generated informative results in much of the VMD for the selected ES (flood mitigation, agriculture/aquaculture productivity, and climate regulation), but challenges arose around its application to a new agro-hydrological regime. To address these challenges, parameterising LUCI and reconceptualising some of the model’s mechanisms to specifically account for the productivity and flood mitigation capability of water-tolerant crops as well as flooding processes of deltaic regions will improve future ES modelling in tropical deltaic areas. The ES maps showed the spatial heterogeneity of ES across the VMD. Next, to at least somewhat account for the economic drivers which need to be considered alongside biophysical valuations for practical implementations of ES maps for nature-based solutions (Nbs) in the upstream VMD, economic values were assigned to different parcels using a benefit transfer approach. The spatially explicit ES economic value maps can inform the design of financing incentives for Nbs. The results and related work can be used to support the establishment of Nbs that ultimately contribute to the security of local farmers’ livelihoods and the sustainability of the VMD.

1. Introduction

Deltas, home to 4.5% of the global population, are the most densely populated areas in the world, with a general average of 478 people/km², eight times the global average [1]. Deltas are also among the most productive and economically important global ecosystems [2], providing a wide range of ecosystem services (ES). In these populous, flood-prone regions, trade-offs among agricultural production and reducing floods are at the forefront of their management [3,4,5], especially in tropical regions vulnerable to climate change [6]. Recognising the substantial ecological and socio-economic benefits of investment in natural capital, many deltas have incorporated nature-based solutions (Nbs) as part of flood risk management strategies instead of solely relying on hard infrastructure [7], e.g., the Mississippi Delta [8], the Dutch Delta [9], and the Ganga–Brahmaputra–Meghna Delta [10,11].

Ecosystem service assessments have been advocated as an important part of Nbs for natural flood resilience [12,13,14,15,16]. Among ES assessment tools, spatially explicit ES models have been demonstrated as particularly effective decision support tools [17,18,19,20]. Maps of multiple ES provide information on the existence and spatial heterogeneity of ES delivered by Nbs. Spatially explicit ES models, e.g., InVEST, ARIES, and LUCI, possess tools/functions to facilitate ES spatial distribution mapping, scenario analysis, and ES interaction analysis (trade-offs and synergies) [21]. These functions are not widely available or very accessible in discipline-specific models, such as hydrological models, ecological/environmental models, and habitat distribution models. In addition, ES models are often designed to be less data intensive and more accessible to non-disciplinary experts compared to discipline-specific models. However, deltas have unique biophysical characteristics that have been poorly accounted for in most ES models. To address this research gap in ES modelling, adaptions and enhancements of ES models for applications in deltas and floodplains are necessary if these models are to provide any significant value for these important parts of the world in the future.

In deltaic regions, Nbs often aim to preserve flood benefits while providing agriculture and aquaculture productivity. Suitable models for understanding these ES interactions should contain modelling tools for both flood benefits and agricultural/aquacultural productivity. Reviewing the functionality of ES models, the Land Utilisation and Capability Indicator, LUCI, has particular strengths in hydrology and agricultural productivity analysis [22,23]. In LUCI, the concepts of spatial connectivity are well represented in all ES modelling algorithms which take into account the interacting effects of topography (hydrology routing), soil, water, and biodiversity [22,24]. For example, the flood mitigation tool of LUCI performs spatially explicit topographical routing, considering the storage and permeability capacity of elements within the landscape from soil and land use data, and honoring physical thresholds and mass balance constraints [25,26]. LUCI’s strengths in hydrology make it ideal for analyses of wetland ecosystem services and their interactions [27]. To date, LUCI is the only ES model that can operate at both landscape and field/sub-field scales (5 × 5 m grid cell) [21,24], making it suitable for assisting field-based flood management Nbs. Applications to date suggest that a 5 × 5 m DEM provides more than sufficient resolution for making decisions at the field scale [25]. LUCI has been widely used in England and Wales, e.g., Sharps et al. [21], Emmett and the GMEP team [22] and Jackson et al. [25], and in New Zealand, e.g., Trodahl et al. [24], Tomscha et al. [27], Delpy et al. [28], Nguyen et al. [29] and Tomscha et al. [30]. In the tropical Asia-Pacific region, the model was applied to support ecosystem-based adaptation in Vanuatu [31] and mapped multiple ES under land use change scenarios in the Cagayan de Oro catchment in the Philippines [32].

LUCI’s algorithms were originally developed in temperate hill country regions; therefore, challenges arose in taking into account the typical agro-hydrological characteristics of deltas. Located in the fluvial-to-marine transition zone, deltas have complex water flow with tidal influences and strong interactions with groundwater. Deltas also have cultivation activities adapted to seasonal flood conditions, e.g., aquaculture or mixed agri-aquaculture. The relationship between flooding and agriculture/aquaculture in deltas is temporally and spatially dynamic. These temporal and spatial configurations of agro-hydrological characteristics are important considerations for ES modelling in deltas. In addition, high-resolution spatially explicit biophysical modelling for large deltaic regions remains computationally difficult [3]. This research is the first application of LUCI to a deltaic region. Exploring the application of LUCI in deltas was carried out, in part, to explore advantages and limitations of ES modelling in these regions, and to provide recommendations to extend the applicability of LUCI and ES modelling more generally for flat and low-lying regions.

The Vietnam Mekong Delta (VMD), the world’s third largest delta, is a highly productive agricultural and aquacultural region of Vietnam [33,34,35,36]. Over recent decades, the VMD has witnessed extensive development of human-made water control infrastructure, especially high-dike systems in the upper part of the delta (Long Xuyen Quadrangle, Plain of Reeds, and the areas between them, Figure 1) to fully protect rice fields from floods and enable intensive agriculture production (three rice crops per year) [37,38,39]. Details of the dike systems can be found in the Study Area section. High dikes prevent flood water from connecting to the delta’s floodplains. As such, traditional floodplain ecosystem services (ES), i.e., sediment trapping, water storage, and water purification and provision, are reduced [40,41,42].

Recognising the limitations of hard protection infrastructure, nature-based solutions for flood control are considered a key component of the Mekong Delta Plan [43]. Efforts to reconnect the delta’s floodplains with the Mekong River have been carried out in the upstream VMD, where significant flood mitigation features exist, including river-connected rice fields, swamps, and natural wetlands [44,45,46]. The first effective policy on Nbs in the VMD is the 8 crops in 3 years strategy (3-3-2 crop cycle) introduced by An Giang Province in 2007 [45,47]. The 3-3-2 scheme means paddies in high-dike systems are left fallow and fully flooded in the ninth crop cycle over three continuous years. However, the 3-3-2 scheme as well as some flood-based crop (lotus, water lilies, floating rice) demonstration projects has not succeeded to date, largely due to the lack of social and financial mechanisms to incentivise farmers’ participation in Nbs [44,45]. Obtaining financial and social support for Nbs in the region has been challenging, in part, because we lack the robust biophysical modelling needed to underpin the spatial planning for these efforts. Spatially explicit ES biophysical values can also facilitate spatially explicit economic value mapping [48]. Information on ES economic value at farm scales can help farmers clearly see the benefit of flood control Nbs on their farms. Spatially explicit ES economic values are also useful for the design of financial mechanisms/instruments to encourage farmers to participate in Nbs [49,50]. To date, current efforts in ES research in Vietnam on high-resolution ES modelling and how the modelling results can support Nbs implementation in the VMD have been limited.

Given the multi-pronged challenge of improving ES modelling and implementing practical solutions, the first objective of this study is to explore the use of the LUCI model to map the multiple ecosystem services (ES) as well as their synergies and trade-offs in the Vietnamese Mekong Delta (VMD). We focused on mapping flood mitigation, agriculture/aquaculture productivity, and climate regulation (carbon sequestration), which are three ES provided by flood management Nbs, at two timeframes, 2010 and 2018. The year 2011 marked the completion of large-scale flood control infrastructure in the delta as well as changes in development strategies for the delta via the “Guidelines for economic development in the VMD, 2011–2020” (Conclusion 28). Comparing ES maps between 2010 and 2018 can help us understand how ES change under socio-economic changes. From the application of LUCI in the VMD, we explored how well the model works for tropical flat areas and suggest improvements to make the model more easily implementable for wider use in the VMD and/or areas with agriculture/aquaculture, aquaculture being an increasingly key global economic activity. The second objective is to, at least somewhat, account for the economic drivers which need to be considered alongside biophysical valuations for practical implementations of ES maps for Nbs in the upstream VMD by assigning economic values to different parcels using a benefit transfer approach. We hope the ES maps will facilitate the establishment of Nbs that ultimately contribute to the sustainable development of the VMD. Two particularly significant contributions of this study are (1) the development of high-resolution ES modelling suitable for deltaic regions, and (2) the combination of ES biophysical and economic values within the Mekong Delta to support local decision making on Nbs implementation.

2. Study Area

The Vietnamese Mekong Delta (VMD) is deposited from sediment transported down the Mekong River where the river meets the Vietnamese East Sea [51]. It covers 39,000 km² of delta flats, with an average elevation of 0.8 m and an elevation range of 0.5–5 m above sea level [43,51]. Before running through Vietnam, the Mekong River is divided into three branches near Phnom Penh, Cambodia: the Tonle River connected to Tonle Sap Lake (also called Great Lake, with a storage capacity of 80 billion m³, having a great influence on the hydrological regime of the whole Mekong River basin), the Mekong River, and the Bassac River. The latter two branches then run into the VMD. The delta is affected by two tidal sources, regular semidiurnal tides (3.5 m) from the Vietnamese East Sea and irregular diurnal tides (0.8–1 m) from the Gulf of Thailand [52]. The climate is tropical monsoonal with two seasons: the dry season from December to the end of April, and the rainy season from May to November [37,52]. More than 90% of the rainfall generally falls in the rainy season [53]; for example, average annual rainfall during 1989–2017 in the VMD has been estimated as 1869 mm, with only 107 mm, on average, falling during the dry season [54].

As with most other delta systems, the VMD possesses rich ecosystems which provide valuable goods and services. Wetlands of the Mekong Delta are among the richest ecosystems of the Mekong River basin, with 1.9 million ha of freshwater river wetlands (tidal floodplains) and 1.05 million ha of saline estuarine wetlands (coastal marshes, peatland marsh, estuaries, etc.) [55]. The VMD also has other special types of wetlands including rice fields and aquacultural cultivation areas which deliver multiple benefits for local people, such as flood mitigation, sediment retention, carbon storage, and food provision [47,56,57,58].

Due to its rich water and sediment resources, the VMD is an important agriculture and aquaculture region of Vietnam. It helps sustain the livelihoods and food security of 17 million inhabitants. Nationally, it contributes about 50% of Vietnam’s rice production, 65% of aquaculture production, and 70% of fruits annually [59]. The importance of the VMD extends beyond Vietnam’s boundary [60]. The VMD paddy rice area accounts for 2.5% of the global paddy rice area [61,62], and total food production land in the VMD is about 0.27% of the globe’s food production land [63,64]. Kondolf et al. [65] estimated that the VMD produces around 2.4% of the global paddy rice harvest and 0.5% of the total global calorie supply.

Floods in the rainy season and salinity intrusion in the dry season are the two main challenges in the VMD [66]. During the rainy season (May to November), high rainfall in the Mekong River basin causes flooding in the mainstream of the Mekong River and the Mekong Delta. Inundation of one third of the delta can last up to 3 months [37]. The population in the VMD has extensive experience in living with floods [67,68], and people have generally adapted their lives to their presence [33,66]. In the dry season under the influence of tides, salinisation and the lack of fresh water in coastal areas are the problems to be addressed [66]. Located in the low-lying coastal zone, the tidal effects cause salinity intrusion to penetrate far inland, making salinisation intrusion worse during the dry season [53,69]. Recently, the effects of flood and salinity intrusion have been getting worse due to upstream water usage by neighbouring countries, hydropower dam development in the upstream Mekong basin, and a higher frequency of extreme weather events due to climate change [70]. These impacts are expected to continue to increase in severity. Recent dam status updates indicate that 64 of the 187 existing and proposed hydropower dams are operating [71]. Out of these 64 dams, 18 dams are in China and the other 46 are in the Lower Mekong basin tributaries [71]. Rising sea levels are also likely to infiltrate groundwater aquifers and increase salinity gradients in large parts of the Mekong Delta, particularly during the dry season [60,72]. Within the delta, extensive groundwater extraction and sea level rise have escalated land subsidence issues [73].

The VMD has a relatively dense and complex stream network, in part, due to its flat topography and the extent of human modification, with a huge system of navigation and irrigation channels and a sophisticated dike system. The dike system contains low dikes (“August” dikes) and high dikes. Low dikes were constructed to provide protection against flood peaks arriving around mid-July to mid-August, ensuring the farmers can double their rice crop (i.e., grow two rice crops per year). Flood water then inundates rice fields after harvesting when higher flood peaks are experienced. To maintain the country’s position as one of the world’s top rice exporters, the Vietnamese government instituted a policy (Resolution No. 63/NQ-CP) in 2009 to expand intensified rice farming and to purchase rice from farmers. Farmers were incentivised to increase production by shifting from two rice crops per year to three rice crops per year [45,47,74]. This led to the large expansion of high-dike systems in the upper parts of the VMD’s floodplains to facilitate the third rice crop during the flood season [45,47,75]. The total length of high dikes in the VMD is ~1300 km, and the length of low dikes is ~13,300 km [41]. They are equipped with sluice gates and often additional pumping systems, which form the main linkage of floodplains with channels and the main river.

3. Methodology

In this research, we used the LUCI model to map the spatial distribution of three ES delivered by flood management Nbs including flood mitigation, agriculture/aquaculture productivity, and climate regulation (carbon sequestration) services as well as their synergies and trade-offs across the VMD. Two approaches were used to parameterise soil and land cover information for LUCI. The first approach was coarse but enabled a rapid initial implementation, matching the VMD soils and land covers to their closest counterparts from other countries in datasets already supported by LUCI. The second approach was a user-defined parameterisation specific to the region and its available datasets, in which values for land and soil parameters were assigned based on local information and knowledge. As it becomes clear later in the paper, when applying LUCI in a region where datasets have not been explicitly parameterised for use in the model, a first cut application can be established by matching regional datasets to already supported datasets. This is not recommended for the final application of LUCI but can be a useful way of exploring model capacities and preliminary results for your region. From the process of crude matching and also the preliminary results obtained through this method, users can identify which land use and soil types of their study area are not currently well represented in LUCI and need to be better parameterised. Careful consideration should be given to whether the supported datasets span similar land use or soil types to those present in your region. Where they do not, direct user-specified parameterisation for at least those types dissimilar to any already supported should be immediately prioritised.

LUCI was run with these different parameterisation sets at two timeframes, 2010 and 2018. We then used the trade-off map of flood mitigation and agriculture/aquaculture productivity to identify areas suitable for Nbs implementation. Next, we mapped the ES value of Nbs by assigning economic values to the trade-off map’s parcels. Economic value was estimated using benefit transfer methods. The economic value of flood mitigation is the replacement cost of flood protection alternatives in the upstream VMD.

3.1. Land Utilisation and Capacity Indicator—LUCI Model Overview

The Land Utilisation and Capability Indicator model (LUCI) is a second-generation extension and software implementation of the Polyscape framework. Descriptions of LUCI can be found in Sharps et al. [21], Jackson et al. [25], Tomscha et al. [27], and Marapara [76]. LUCI is spatially explicit, respecting both the biophysical properties of individual landscape elements and their configuration when estimating ecosystem functions and services [26]. It explores the capability of a landscape to provide a variety of ES including flood mitigation, agricultural production, water quality (nitrogen and phosphorus), erosion risk and sediment delivery, carbon sequestration, and habitat provision. LUCI compares the services provided by the current utilisation of the landscape to estimates of its potential capability. It then uses this information to identify areas where change might be beneficial, and where maintenance of the status might be desirable.

The flood mitigation tool takes into account spatially explicit topographical routing with connectivity and configuration details that have not previously featured in any ES service model [21,22,25,26,76]. LUCI uses GIS functions to calculate flow direction and flow accumulation, among other types of background information, to support predictions. The tool then combines this information with storage and/or permeability capacity considerations based on soil and land use information to adapt mass (water, sediment, nutrients) accumulation using bespoke algorithms. Volumetric constraints on readily plant available water and plant available water, infiltration capacity, maximum drainage rate, and drainable water holding capacity are considered [76]. These soil hydraulic properties are usually not readily available and costly to measure. To support LUCI users and broader hydrological applications, guidance and an associated toolbox (LUCI_PTFs) to obtain required soil hydraulic properties in a cost-effective way are also available [77].

LUCI creates maps of “flood mitigating land”, which represents areas with high storage and/or high permeability to mitigate floods. Flood mitigating land acts as a “sink” for fast moving overland flow and near-surface subsurface flow. Typical flood mitigating areas in the VMD are melaleuca forests, rice fields, aquaculture fields, mangroves, and other natural wetlands. In contrast, “mitigated land” represents areas that receive mitigation, i.e., water originating in mitigated land later flows through mitigating areas before reaching a stream, lake, or river. Finally, areas where a large amount of unmitigated flow routes directly to waterways are treated as priority areas for change, called “flood concentration” areas. The tool also calculates the average flow delivery to all points in the stream network to estimate water supply services. Parameters to define thresholds for the “corrected” flow accumulation values (soil and land use storage and/or permeability capacity) are used to categorise priority areas for targeting change [25].

The spatially explicit hydrological routing algorithm of LUCI is valuable for the VMD. As the delta has a very dense stream network and wide floodplains, the flood mitigation capacity of stream networks and wide floodplains cannot be sufficiently recognised without considering spatially explicit hydrological routing. However, there are several characteristics of the LUCI flood mitigation tool that do not correctly represent the flood mitigation capacity and flood regimes of the VMD. Firstly, for flood mitigation features, LUCI only focuses on the storage capacity of soil and considers waterlogged areas as not providing flood mitigation. However, this is incorrect for the VMD, as the delta has a large above-ground flood water storage capacity in rice fields, which are frequently waterlogged. In addition, the flood tool evaluates flood mitigation potential based on physical principles of hillslope flow. The tool only considers flood water flow transfers from terrestrial environments (hillslopes, floodplains, etc.) to fluvial environments (streams, rivers, lakes, etc.). However, the VMD receives large upstream inflows annually (about 475 km³/year for the whole Mekong Delta) [78]; therefore, the flood water flow from fluvial environments to terrestrial environments (e.g., overbank flow) is significantly important and needed to be considered in the ES modelling process. It is noted that this study focused on adapting the most established parts of LUCI (ES tools) to a deltaic region. Although the LUCI framework has further relevant algorithms such as the flatwater inundation tool which accounts for overbank water flow [32,79], these algorithms have yet to be well documented or widely applied. Implementing these algorithms for new sites and contexts requires significant developer input and time; therefore, overbank flows could not be included in the ES modelling within the timeframe of this study.

The LUCI agricultural productivity tool evaluates potential agricultural productivity of land according to slope, fertility, aspect, and drainage. The model calculates predicted optimal agricultural utilisation based on soil type, using assigned values of fertility, waterlogging (permanent, seasonal, or negligible), and topographic data (aspect, slope, and elevation). Weights can be applied to increase or decrease the importance of these parameters [22,25]. Current agricultural utilisation is mapped according to the land cover data, ranking land use from highest productivity to lowest. Further agricultural productivity outputs consider differences between predicted productivity and “actual” productivity/land use.

Agricultural and aquacultural activities in the VMD are largely dependent on flood regimes. The LUCI agricultural productivity tool already considers the spatial interactions of hydrology routing with land and soil. These model features are key to representing the relationship between agriculture/aquaculture and floods in the VMD. However, the tool’s algorithms/mechanisms related to waterlogging dependency were developed for farming systems in temperate regions, and it considers waterlogged areas as not suitable for agricultural production. However, this principle is not correct for the farming systems in the VMD as well as other deltas where important crops are mostly waterlogged crops that can provide high productivity, e.g., rice, aquaculture, and lotus. Therefore, to provide ES maps for the VMD, waterlogging-relevant parameters were taken outside of their physically realistic settings for hilly temperate regions to represent productivity and flood mitigation capacity of waterlogged crops.

The carbon sequestration tool of LUCI is based on Tier I and Tier II of the IPCC to calculate carbon levels at the steady state. In this study, the Tier II approach was used. Values of carbon pools (above-ground carbon, below-ground carbon, deadwood carbon, litter carbon, and organic carbon) were assigned based on land use/land cover and soil. The total values for biomass carbon and soil carbon were then fed into the model to identify areas with significant carbon stocks which should be protected, and areas where there is potential for sequestration.

LUCI has a unique built-in trade-off tool to identify locations where multiple services might benefit from interventions, or where there may be a trade-off with one service benefitting from interventions while another is reduced [21]. This study includes analysis of a trade-off and synergy map of flood mitigation and agricultural/aquacultural productivity obtained from LUCI to examine areas where both flood mitigation potential and agricultural/aquacultural productivity potential exist. This information is then used to identify areas suitable for developing flood-based crops and prioritise Nbs implementation.

3.2. Data and Materials

The input data for this study together with their sources are presented in

Table 1. Data and data sources.

Required Data		Data Type	Data Source
DEM	5 m DEM	Raster	The Ministry of Natural Resources and Environment, Vietnam
	30 m SRTM DEM		United States Geological Survey (USGS)
Land use/land cover	LULC map 2010	Vector Polygon	Water Source University of Vietnam
	LULC map 2018		The National Institute of Agricultural Planning and Projection, Vietnam
Soil map		Vector Polygon	University of Can Tho (WISDOM project)
Study area mask		Vector Polygon	Ministry of Natural Resources and Environment, Vietnam
Stream network		Vector Polyline	Ministry of Natural Resources and Environment, Vietnam
Gridded annual rainfall (mm/year) 1 km		Raster	Worldclim.org
Gridded annual potential evapotranspiration (mm/year) 1 km		Raster	Worldclim.org
Hourly rainfall data of Tan Chau station		Tabulate	An Giang Hydro-Meteorological Station
Hourly rainfall data of Tan Chau station		Raster	ERA5 hourly data at single levels from 1979 to present
Hourly potential evapotranspiration of Tan Chau station		Raster	ERA5 hourly data at single levels from 1979 to present
Soil properties (sand, silt, clay, bulk density, organic carbon/organic matter, pH, CEC, ECE, etc.)			WISE—global soil property databases
Inundation maps 2000–2013		Vector	Obtained from Le et al. [91]
Agriculture and aquaculture production statistics 2000–2012 of the VMD’s 13 provinces		Tabulate	General Statistics Office of Vietnam

. Data scarcity is a big obstacle of ES modelling, especially in the Asia and Southeast Asia regions [19,80,81]. Besides the 5 m DEM, the 30 m SRTM DEM was used to investigate the capability of an open-access dataset with moderate resolution for ES modelling in the VMD, a data-sparse region. The land use/land cover (LULC) map of 2010 and 2018 contained detailed LULC classes, e.g., rice fields with detailed crop rotation and aquacultural lands with specific crop types. Detailed land use/land cover classes are needed to represent the spatial heterogeneity of their multiple ES. A stream network was used to generate a hydrologically and topographically consistent DEM which is important to improve the accuracy of flow routing. Global gridded annual rainfall and potential evapotranspiration data were renormalised using hourly rainfall and potential evapotranspiration data obtained from ERA5 (at the Tan Chau station location). Ideally, hourly climate data should be collected from local hydro-meteorological stations. However, local hydro-meteorological stations in the VMD only collect hourly rainfall data, not hourly potential evapotranspiration. In addition to the spatial data, information of soil and land evaluation was obtained from previous land and soil evaluation research conducted in the VMD [82,83,84,85,86,87,88,89,90]. This information is essential to understand soil and land characteristics in the VMD. Inundation maps obtained from the study on “Interplay between land-use dynamics and changes in hydrological regime in the Vietnamese Mekong Delta” by Le et al. [91] and agriculture/aquaculture production statistics of the VMD were used to validate the results of flood mitigation and agriculture/aquaculture productivity maps.

3.3. Parameterising LUCI for Mapping Biophysical Value of ES in the VMD

As discussed above, LUCI assumes some characteristics and mechanisms around flooding and soil waterlogging that are not directly applicable to the VMD. To obtain useful results for the VMD, we removed the waterlogging dependency and raised the productivity and flood mitigation values associated with waterlogged crops. We took two separate approaches to parameterising the soil and land cover information: in the first instance, through matching VMD soils and land covers to their closest counterparts in already supported datasets from other countries, and through user-defined parameterisation.

3.3.1. Matching VMD Soil and LULC Datasets to Supported/Already Parameterised Datasets

The aim of cross-referencing classes in new datasets to their closest matches in already supported datasets provides a way to explore LUCI capacities and preliminary results for regions where limited information to support parameterisation is readily available. Among the available LULC and soil products, the New Zealand Land Cover Databasev2.0 (NZ LCDB2) land use data and NZ’s Fundamental Soils Layer (FSL) soil data were selected to represent the VMD LULC classes and soil classes. The NZ LCDB 2 has more detailed LULC classes, especially with respect to wetlands, compared to other supported products. NZ’s FSL soil data are also more readily available than other soil data. Detailed LULC and soil descriptions make it easier to match the VMD’s LULC and soil classes to their closest counterparts from other countries. However, matching the VMD LULC classes with the available land databases is not very satisfactory in some VMD areas as none of the LULC datasets supported in LUCI to date have any classes similar to rice or agriculturally/aquaculturally productive wetlands in the VMD. Despite this shortcoming, details of the two preliminary crude matches (which we call link-code 1 and link-code 2) are presented in

Table 2. Preliminary exploration of rice and aquacultural LULC classes using LCDB2.

Land Table	The VMD LULC Classes	LCDB2
Link-code 1	“triple rice”	“short-rotation cropland”
	“double rice” and “single rice”	“herbaceous freshwater vegetation”
	Aquacultural LULC classes (rice-shrimp, shrimp, mangrove-shrimp)	“pond and lake”
Link-code 1	“triple rice”	“short-rotation cropland”
	“double rice” and “single rice”	“herbaceous freshwater vegetation”
	Aquacultural LULC classes (rice-shrimp, shrimp, mangrove-shrimp)	“herbaceous freshwater vegetation”

.

3.3.2. User-Defined Parameterisation

For the user-defined parameterisation, specific land and soil characteristics of the VMD (or any other region) can be brought into the modelling process. For the three selected ES, the required information for each of the LULC classes is as follows: flood mitigation ability, agricultural/aquacultural productivity, and carbon pools’ stock. Flood mitigation ability and agricultural/aquacultural productivity were defined based on the local knowledge of the study area and land evaluation research conducted in the VMD [82,83,84,85,86,87,88,89,90,91,92,93,94]. Given the high productivity and flood mitigation capacity of the VMD’s wetlands, the waterlogged dependency of LUCI’s flood mitigation and agricultural productivity tools was altered to raise the flood mitigation and productivity values associated with waterlogged crops of the VMD, e.g., rice and aquaculture (rice-shrimp, shrimp, mangrove-shrimp classes). By doing this, we partially take into account the flood mitigation capacity and agriculture/aquaculture productivity of wetlands and waterlogged crops in the VMD; however, the flooding environment and processes, i.e., overbank floods, tides and groundwater influences, and above-ground flood water storage capacity, need further work to be represented. Storage values of carbon pools were collected from carbon monitoring and estimation research conducted in the VMD. If the value could not be found in studies of the VMD, values of carbon pools of a similar LULC were collected from other studies in Vietnam or the guidance on global parameters from IPCC [95].

As for soil data, specific values of drainage ability, plant available water (PAW), and fertility were identified for the VMD soil classes. Information on organic carbon (OC)/organic matter (OM) was used to define soil fertility. For the VMD case study, we defined soil with more than 4% of OC as the most fertile soil, and soil with less than 1% of OC as the least fertile soil. Drainage was represented by the soil’s saturated hydraulic conductivity. Soil with higher saturated hydraulic conductivity had higher drainage. Information on soil hydraulic parameters was obtained using the guidance by Dang, Jackson, and Tomscha, et al. [77] and an associated toolbox (LUCI_PTFs). We developed three sets (Set 1, Set 2, and Set 3) of land and soil tables to understand how different model parameterisations produced different outcomes. The details of each land and soil table are presented in

Table 3. Three sets of soil and land tables for the user-defined parameterisation.

Set	Description
Set 1	Both soil and land tables were set following the instruction of “LUCI Factors Help Documentation” [96]. - Land table: aquacultural areas were assumed to have no productivity as with other water classes. - Soil table: the waterlogged nature of the VMD soil classes was set but led to LUCI not recognising that such soils can still have high agricultural value and flood mitigation capacity.
Set 2	Soil table was set following the instruction of “LUCI Factors Help Documentation” [96]. However, in the land table, productivity of aquacultural areas did not follow the instructions. - Land table: aquacultural areas were assigned as high-productivity areas. - Soil table: the waterlogged nature of the VMD soil classes was set but led to LUCI not recognising that such soils can still have high agricultural value and flood mitigation capacity.
Set 3	Altering the general parameterisation guidance for some parameters of both soil and land tables to help LUCI recognise flood mitigation capacity and productivity of waterlogged crops. - Land table: aquacultural areas were assigned as high-productivity areas. - Soil table: the flood mitigation and productivity values associated with waterlogged crops were increased so LUCI can recognise agricultural/aquacultural value and flood mitigation capacity of soils in waterlogged conditions.

. Details of the land and soil parameterisations are in

Table 4. Land parameterisations for the Vietnamese Mekong Delta application.

LULC Class	PFLOOD	PAGCLASS SET 1	PAGCLASS SET 2	PAGCLASS SET 3	CARBONABIO	CARBONBBIO	CARBONDEAD	CARBONLIT	SOILOC
Single Rice	2	2	2	2	4	1	0	0.6	20.05
Double Rice	2	1	1	1	6.67	1.1	0	0.6	20.05
Triple Rice	2	1	1	1	9.96	1.3	0	0.6	20.05
Rice-Shrimp	2	1	1	1	4	1	0	0.6	20.05
River	3	6	3	3	0	0	0	0	0
Melaleuca Forest	2	5	4	5	56.25	13.04	4	1.1	104.29
Orchard Farm	2	2	2	1	39.78	2.68	0	0.6	64.2
Shrimp	2	6	1	1	0	0	0	0	0
Residential Area	1	7	7	7	0	0	0	0	12.55
Ornamental Plant Garden	1	5	5	5	4	1	0	0.6	12.55
Sugarcane	2	2	2	2	4	1	0	0.6	12.55
Pineapple	2	2	2	2	4	1	0	0.6	12.55
Vegetable	2	2	2	2	4	1	0	0.6	12.55
Mangrove	2	5	3	3	125.8	32.92	3.4	0	691.35
Mangrove-Shrimp	2	5	2	2	125.8	32.92	3.4	0	691.35
Broadleaf Forest	2	5	5	5	107.7	15.1	4	1.1	75.54
Salt	2	5	4	5	0	0	0	0	0
Rice-Vegetable	2	1	1	1	6.67	1.1	0	0.6	20.05
Lake and Pond	3	6	3	3	0	0	0	0	0

PFLOOD: permeability/mitigation ability (value: 1–3); PAGCLASS: agricultural/aquacultural productivity (value: 1–7); CARBONBBIO (tonne/ha): below-ground carbon; CARBONDEAD (tonne/ha): deadwood carbon; CARBONLIT (tonne/ha): litter carbon; SOILOC (tonne/ha): soil carbon. Detailed instructions are in the “LUCI Factors Help Documentation” [96].

and

Table 5. Soil parameterisations for the Vietnamese Mekong Delta application.

SOILTYPE	LUCI_WLOG SET 1	LUCI_WLOG SET 2	LUCI_WLOG SET 3	LUCI_FERT	PAW
Alluvial Soils with Yellow-Red Mottles	2	2	1	1	150.90
Deposited Alluvial Soils	2	2	1	1	144.85
Eroded Soils	2	2	2	4	11.85
Gleyic Alluvial Soils	2	2	1	2	119.34
Grey Soils on Acid Magmatic Rocks and Sandy Stones	1	1	1	4	163.17
Grey Soils on Old Alluvium	2	2	1	3	169.01
Humic Grey Soils on Old Alluvium	2	2	1	2	155.09
Undeposited Alluvial Soils	2	2	1	1	131.44
Saline Mangrove Soils	3	3	1	3	134.40
Moderately Saline Soils	2	2	1	3	151.04
Peaty Acid Sulphate Soils	2	2	3	1	39.28
Saline–Acid Sulphate Soils–Sulfidic Horizon: 0–50 cm	3	3	1	2	186.23
Saline–Potential Acid Sulphate Soils–Sulfidic Material: 0–50 cm	3	3	1	2	186.23
Saline–Acid Sulphate Soils–Sulfidic Horizon: >50 cm	3	3	1	2	186.23
Saline–Potential Acid Sulphate Soils–Sulfidic Material: >50 cm	3	3	1	2	186.23
Raised Ridge Sandy Soils	1	1	1	4	78.19
Actual Acid Sulphate Soils–Sulfidic Horizon: 0–50 cm	2	2	1	2	186.23
Potential Acid Sulphate Mangrove Soils–Sulfidic Material: 0–50 cm	3	3	1	2	186.23
Potential Acid Sulphate Soils–Sulfidic Material: 0–50 cm	2	2	1	2	186.23
Strongly Saline Soils	2	2	1	4	134.40
Actual Acid Sulphate Soils–Sulfidic Horizon: >50 cm	2	2	1	2	37.60
Potential Acid Sulphate Mangrove Soils–Sulfidic Material: >50 cm	3	3	1	2	186.23
Potential Acid Sulphate Soils–Sulfidic Material: >50 cm	2	2	1	2	186.23
Slightly Saline Soils	2	2	2	3	134.40
Yellow-Red Soils on Acid Magmatic Rocks	1	1	1	4	167.12

LUCI_WLOG: drainage ability (value: 1–3); LUCI_FERT: soil fertility (value: 1–5); PAW: plant available water (mm). Detailed instructions are in the “LUCI Factors Help Documentation” [96].

, respectively.

3.4. Mapping Economic Values of Flood-Based Crops in Upper Streams of the VMD

The first step in mapping the economic values of flood-based crops is identifying areas that are most suitable for their establishment. The trade-off and synergy map of flood mitigation and agricultural/aquacultural productivity obtained from LUCI presents areas where both flood mitigation and agricultural/aquacultural productivity exist. These areas are mapped into two classes, namely, “excellent service provision” and “moderate service provision”, in LUCI’s trade-off and synergy map. Flood protection infrastructure is important to ensure the safety of crops and people when growing flood-based crops in the flood season. Therefore, the most suitable areas to implement flood-based crops in the VMD should be the “excellent service provision” and “moderate service provision” locations within dike rings. Using ES modelling results to map ES economic values results in greater spatial heterogeneity and precision of ES economic value maps compared to the use of commonly used land use/land cover-based proxy methods.

After identifying the most suitable areas for flood-based crops, economic valuations were placed on these areas using simple benefit transfer methods, transferring the value estimated from previous research to our study. Replacement cost methods are widely used to estimate economic value of flood regulation services [97]. In this approach, the cost for planned flood control alternatives (low dike and high dike management cost) is assigned as the economic value of flood mitigation services. The management costs of low dikes and high dikes were estimated at USD 2345.7/ha for high dikes and USD 118/ha for low dikes by Tran et al. [98]. We applied this value in our study as the economic value of flood mitigation services in the upstream VMD. We assumed that the management cost is the minimum compensation that should be paid to farmers for implementing flood-based crops in the upstream of the VMD. These values were then placed on the flood mitigation and agriculture/aquaculture productivity trade-off and synergy map to produce an economic value map. Each cell therefore has an economic value based on its capacity to provide synergies of flood mitigation and agriculture/aquaculture productivity. We assumed that the “moderate service provision” areas can provide 80% of the function of “excellent service provision” areas. Weights equal to 1 and 0.8 were applied when estimating the economic value for the “excellent service provision” and “moderate service provision” areas, respectively. A map of the ES economic value based on only the dike system was also developed for comparison with the map that was developed using ES modelling results. The comparison demonstrates how ES modelling can improve ES economic value mapping.

A proposed ecosystem service-based framework to support Nbs implementation in the VMD is presented in Figure 2. This framework was adapted from the framework of the ecosystem service-based approach for decision making by Daily et al. [99] and the circle for PES implementation by Marino and Pellegrino [100]. In this framework, ecosystem service modelling plays an important role in providing information on ecosystem services and in supporting economic evaluation of ES. ES values are then used in payment for the ecosystem services scheme which compensates farmers who implement nature-based alternatives by letting their rice fields flood during the flood season.

4. Results and Discussion

4.1. Ecosystem Service Biophysical Mapping

LUCI was run with different input data combinations and model parameterisations to explore the applicability of the model to map flood mitigation, agriculture/aquaculture productivity, and carbon sequestration in the VMD. The modelling computation time for this study was accomplished in ~1500 computer hours. For each ES of a single run, ~1500 million elements were computed. No such fine-resolution ES modelling has been previously conducted in the VMD.

Maps of flood mitigation services, agriculture/aquaculture, and carbon sequestration are presented in Figure 3, Figure 4 and Figure 5, respectively. LUCI delineates existing flood mitigation features of the VMD including rice fields, aquacultural areas, mangroves, and other wetlands. These areas are shown as flooding mitigation land in green colour (Figure 3). High-flood concentration areas in red and moderate-flood concentration areas in orange (Figure 3) are the areas with non-mitigated features. These areas could be modified to improve water holding capacity. Areas highlighted as having a negligible flood concentration have local characteristics which promote flooding but have been protected by upstream and/or uphill mitigating soil and LULC combinations.

Figure 4 presents agriculture/aquaculture productivity maps obtained using the three sets of land and soil tables described in the methodology. When using Set 1, the productivity of aquacultural areas could not be mapped properly. All aquacultural areas are shown in red (Figure 4, Agriculture Productivity—Set 1). Areas with high ES provision (optimum utilisation) are in dark green. Moderate-ES provision areas including near-optimum utilisation and non-optimum utilisation are in light green and orange, respectively. Using Set 2 (productivity value of agricultural land was raised), the productivity of aquacultural areas was better recognised compared to when using Set 1, but not fully represented (Figure 4 Agri-Aquaculture Productivity—Set 2). In Set 3, the productivity value of agricultural land was also raised, and, additionally, the waterlogging dependency in LUCI was taken out for ES modelling in the VMD. Using Set 3, the productivity of typical agriculture/aquaculture lands of the VMD was much better represented compared to the previous sets (Agri-Aquaculture Productivity—Set 3).

In the carbon sequestration maps (Figure 5), the dark green areas have a very high carbon stock (≥90 tonnes C/ha). These areas are broadleaf forests, melaleuca forests, mangrove forests, and orchards. Areas with a high carbon stock (50–90 tonnes C/ha) and moderate carbon stock (27–50 tonnes C/ha) are triple rice fields, double rice-vegetable, other crops, and other wetlands. Rice-shrimp land has a low carbon stock (21–217 tonnes C/ha), and aquacultural land has a very low carbon stock. Modelling carbon sequestration for the VMD is advantageous because, to date, there has been a large amount of research on carbon stock measurements/estimations conducted in the VMD. These are valuable sources of information to create carbon tables for mapping carbon sequestration services.

Figure 3, Figure 4 and Figure 5 also show the change in ES between 2010 and 2018. The year 2011 marks the completion of large-scale high-dike systems and important water control systems in the VMD, especially the Vam Nao project to connect the Bassac River and the Mekong River [91]. The year 2011 also marks the change in economic development strategies of the delta. The Resolution 21-policy guidelines for economic development in the VMD (2001–2010) were replaced in 2011 by the Conclusion 28-policy guidelines for economic development in the VMD (2011–2020) [101]. ES maps of 2010 and 2018, therefore, can provide information on the influences of flood protection infrastructure and economic development policies on ES in the VMD. However, the comparison of ES maps between 2010 and 2018 only provides the relative trend of ES change during the study period, not the exact change in areas, as the LULC maps of 2010 and 2018 are different in source of origin and resolution.

A clear decrease in flood mitigation services can be seen in the riparian areas between the Mekong River and the Bassac River (area within the black square, Figure 3). This reduction in mitigating land could be due to the development of dike systems in the upstream and urban/residential expansion in the riparian areas. From 2010 to 2018, agriculture/aquaculture productivity for the VMD increased mostly in the southern part of the delta, in Ca Mau Province (area in the black square, Figure 4). In recent years, large areas of mangroves in the province have been converted to aquaculture. However, the development of aquaculture has led to a reduction in carbon stocks in this area (Figure 5, carbon stocks and flux).

The flood mitigation map produced by LUCI using the 2010 data was compared with an inundation map from 2010 of the VMD. Similarly, the agriculture productivity map of 2010 was compared with the VMD agriculture/aquaculture production statistics at the district level (2010). The results demonstrate that LUCI provided reasonable information on the spatial heterogeneity of flood mitigation and agriculture/aquaculture productivity in the VMD. As for carbon sequestration, all local carbon measurements/estimations available were used to develop the carbon table for the modelling process. There is no other source of information that can be used for comparison with our carbon sequestration map. Field visits are ideal to validate the modelling results; however, we could not carry out field visits due to COVID-19 travel restrictions during the time spent conducting this study.

Maps of trade-offs and synergies between flood mitigation and agricultural/aquacultural productivity provide information on areas where both flood mitigation and agricultural/aquacultural productivity exist. Those areas that have excellent service provision and moderate service provision should be preserved to assure the co-existence of flood mitigation and agricultural/aquacultural productivity in the VMD (Figure 6). The excellent service provision areas and moderate service provision areas within dike systems are the most and second most suitable areas to implement flood-based crops.

The preliminary exploration provided information on the suitability of the model in the VMD context. Flood mitigation services of the “double rice” and “single rice” classes were partially acknowledged in the LUCI outputs (Figure 7). However, the coastal rice-shrimp areas and other aquacultural areas were considered to be water bodies, meaning LUCI did not identify any flood mitigation (or agriculture productivity) values in these places. Furthermore, the differing capabilities of alternate rice cropping systems to differently mitigate floods are not fully represented (Figure 7).

As Figure 8 shows, the productivity of the VMD orchards is also not sufficiently recognised using either preliminary parameterisation due to the differences in fruit species or their annual growth cycles between Vietnam and NZ. The productivity of mangroves cannot be highlighted either because NZ does not have integrated mangrove aquaculture as in the VMD. The differences in vegetation types and crops in Vietnam and New Zealand also led to poor carbon stock mapping using crude matching. Carbon sequestration of rice fields, mangroves, and other wetlands was mapped with a very low carbon stock (Figure 9).

The SRTM 30 m DEM was also used to map multiple ES of the VMD to explore the applicability of global data in ES mapping in a data-limited context. Figure 10 shows maps of flood mitigation (Figure 10a) and agriculture/aquaculture productivity (Figure 10b) using the SRTM 30 m DEM compared with results from the 5 m DEM. The results show that at the regional scale (the whole VMD), ecosystem service maps obtained using the SRTM 30 m DEM can provide comparable information to ecosystem service maps obtained from the 5 m DEM. However, the SRTM 30 m DEM cannot map ES properly at a fine scale/local scale. Figure 10a zooms in on Can Tho city. SRTM 30 m contains many artefacts and elevation errors. Therefore, using SRTM 30 m can only provide reasonable ES maps for decision making at a regional scale. ES maps obtained from SRTM 30 m are rarely suitable for supporting decision making at local scales (e.g., small wetlands or farms). Therefore, deltaic regions should be prioritised for collection of high-resolution topographic information to improve the representation of small-scale features that contribute to ES.

4.2. Mapping ES Values to Support PES Schemes in the Upper Part of the VMD

Maps of the economic value of flood mitigation were developed based on the trade-off and synergy mapping of flood mitigation and agricultural/aquacultural productivity, a dike system map, and dike management (investment and maintenance) costs. Figure 11a presents an ES economic value map using only dike system map information, and Figure 11b presents an ES economic map using both modelling results (synergy and trade-off map) and the dike system map. It can be seen that Figure 11b presents more detail and spatial heterogeneity of ES values compared with Figure 11a.

The most suitable areas to implement flood-based crops are in the two most flood-prone areas of the VMD: Dong Thap Muoi or “the Plain of Reeds” (within the purple boundary), and the Long Xuyen Quadrangle (within the pink boundary), as well as in the areas between these two particularly flood-prone regions. The areas with the highest value (USD 2345.6/ha) provide “excellent service provision” within high dikes. The areas with “moderate service provision” within high dikes have the second highest value (USD 1876.8/ha). The areas within low dikes providing “excellent service provision” and “moderate service provision” are estimated to have a flood mitigation economic value of ~USD 118/ha and ~USD 94.4/ha, respectively. The most cost-effective places to develop flood-based crops are the areas providing “excellent service provision” within low dikes. These areas can provide excellent services with lower flood management costs. Figure 11c presents the ranking for the Nbs implementation priority based on ES provision and flood management costs. Although the economic valuation methodology behind the map is simple, it can still provide useful guidance to spatially target payment amounts for farmers who implement flood-based crops. We hope that seeing economically beneficial places located in the low-dike areas can help to improve investment in these areas to encourage farmers not to increase dike heights. Investment in flood-based crops can help to improve the livelihoods of local people in the flood season while preserving the important benefits provided by the natural floodplains. Note that weights of 0.8 and 1.0 were placed evenly across the “excellent service provision” and “moderate service provision” areas; therefore, these weights may cause overestimation or underestimation in some areas. Highly detailed ES economic value mapping would require further information such as questionnaire surveys, which could be implemented in future studies.

The implementation of flood-based crops and species, such as lotus, water lilies, floating rice, flood-based giant freshwater prawns, or snakehead fish, can improve the economic viability of flooding. In the upper VMD, these crops and species were recommended as the “no-regret” measures for flood control in the short and medium terms [43]. These no-regret measures are assessed as efficient and effective in all potential scenarios, and flexibility can be maintained for future developments [43]. The development of flood-based crops aligns with Resolution 899 (2013) [102] and Resolution 120 (2017) of the Vietnamese Government, which resolved to carry out policy changes around the agriculture sector of the VMD to avoid the consequences of the current focus on maximising rice production and instead move towards the development of more climate-resilient agriculture and food production in Vietnam.

A previous study advised that sustainable hydrological production in the Mekong River basin should attain conservation goals [103]. The ecosystem service-based framework, the FOR-POWER framework, and PES for forest conservation were developed and applied to a proposed dam in Cambodia (Pursat 1) [103]. We hope we have argued that PES schemes to protect wetlands in the downstream parts of the VMD will significantly augment the existing efforts to protect forests in upland areas.

4.3. Recommendations to Improve LUCI to Better Adapt to the VMD and Delta and/or Tropical Geoclimatic Regions More Generally

Although reconceptualising some of LUCI’s parameters and mechanisms can result in reasonable ES maps of flood mitigation and agriculture/aquaculture productivity for the VMD, the model can be further improved for deltaic regions, which would likely improve the applicability and utility of LUCI in the VMD. These improvements would not only be valuable for the VMD but also for broader coastal flat areas and other areas with waterlogging-tolerant and/or waterlogging-dependent crops.

The current hydrological routing of LUCI primarily considers flow directed from hills through floodplains to rivers. However, as a large receptor of the Mekong River, water flow exchanges back from the river to the floodplains as overbank flow are an important process in the VMD. Overbank flows are also the main sources of sediment in the floodplains of the VMD as well as other deltaic regions. While sediment retention is an important ES of flood control, sediment retention was not modelled in this study because the LUCI sediment retention tool was developed to predict terrestrial erosion and delivery to waterways and does not incorporate the sediment deposition processes from overbank flows coming from the river. In future developments, the hydrological routing algorithm of LUCI should include connections from rivers back to the floodplains. These improvements will enable LUCI to better represent the flood environment and sediment retention processes of the VMD and other deltaic regions. In addition, water flow in the VMD is greatly influenced by tidal regimes. Furthermore, the VMD also has huge areas of wetlands which have close interactions with groundwater movement. As explained earlier, some algorithms in the LUCI framework considering such processes do exist but are not yet mature enough to be easily implemented without significant developer input; our study highlights the need for prioritising efforts to make these more widely applicable and user friendly to improve LUCI’s applicability in wider environment and data contexts.

The agriculture productivity ES algorithms within LUCI should be adapted to recognise the potential food and fibre productivity of waterlogged areas: for example, the rice field and aquaculturally utilised areas in the VMD.

5. Conclusions

To date, this is the first fine-scale multiple ecosystem service modelling study in the Vietnamese Mekong River Delta. With each modelling step processed at 5 by 5 m for an area of 39,000 km², this study dealt with big data and high computational requirements. The Land Utilisation and Capability Indicator (LUCI) model was used to map flood mitigation, agriculture/aquaculture productivity, and climate regulation (carbon sequestration), as well as their synergies and trade-offs, in the VMD at two timeframes, 2010 and 2018. To obtain meaningful results for the VMD, the productivity and flood mitigation values associated with waterlogged crops were adapted to help LUCI recognise these ES provided by waterlogged crops. However, LUCI’s parameters and algorithms could be further improved to better represent the often waterlogged environment and other unique processes within tropical delta regions. Sensitivity analysis could also be conducted in future studies to improve the understanding of how key parameters can affect the modelling results. We are currently working on the next generation follow-up to LUCI, the Nature Braid (www.naturebraid.org) which is including innovations along these lines, more temporal analysis, better cultural, economic and social considerations and availability within a freely available open source GIS platform (QGIS). This will make the Nature Braid more accessible than the current LUCI platform, both as relates to costs and transparency of results, important in policy making contexts.

The modelling results provide useful information on multiple ecosystem services and their distribution in the VMD, as well as their changes between 2010 and 2018. Comparing ES maps between these two timeframes reveals how flood control infrastructure accomplishment and regional development strategy updates in 2011 have affected ES across the delta. There are decreases in flood mitigation in the lower stream of the VMD and increases in agriculture/aquaculture productivity in the southern parts of the delta. The increased agriculture/aquaculture productivity has led to a reduction in the carbon stock in the southern VMD. In addition, this study used the biophysical ES modelling results combined with a benefit transfer approach to produce spatially explicit economic value mapping of flood mitigation benefits occurring in the upper part of the delta. The information on ES economic values shown at the farm scale can help farmers see the benefits of Nbs more clearly. The spatially explicit economic value map also shows where areas should be prioritised for Nbs implementation. COVID-19 has thus far prevented us from carrying out planned participatory work with VMD farmers, other stakeholders, and policymakers to evaluate their thoughts on LUCI predictions. We hope this information can be useful for informing designs of PES schemes for Nbs implementation which both preserve multiple ES relating to floods and sustain the livelihoods of local people during the flood season.