Floods are considered to be natural processes [1
]; however, in recent years, the global increase in flooding incidents has been associated with climate change [2
]. Higher number and frequencies of floods increase the risk of damaging properties, destroying infrastructures, and reducing the overall well-being of people. Many communities who are perhaps ‘accustomed’ to flooding and once felt that occasional flooding was part of their lives are at greater risk [1
]. This narrative stands true for many small and rural communities living in the Canadian Prairies.
The landscape of the Canadian Prairies, which is part of the greater North American Prairie Pothole Region (PPR), is characterized by millions of wetland depressions, also known as potholes [5
]. While many of these depressions are isolated, they can occasionally connect at times of high overland water flow through a mechanism known as ‘fill and spill’ [5
]. These wetland depressions have high water storage capacities and capture the majority of runoff generated from snowmelt [8
]. The extensive land use change in the PPR and uncertainties arising from changing climate change, however, have increased flood magnitudes in the Prairies [6
]. These increases intensify vulnerabilities in rural communities. While in the past, rain-on-snow incidents were rare in the Prairies, they are becoming more frequent. The flooding between 2011 and 2016 in Saskatchewan and Manitoba was a result of a rain-on-snow contribution [9
]. There is, therefore, a need to create modernized flood risk management tools to support community preparation for future events brought on by the climate emergency.
Flood risk management involves identifying and managing hazards and inundated areas as well as the particular population’s vulnerability when confronted with unexpected inundated areas [4
]. Hydrological and hydraulic models are often used to produce flood inundation maps that provide information on the spatial extent of floodwater. These maps can be useful for first responders, risk managers, policy-makers, and engineers [10
]. There are, however, several challenges to using conventional hydrological models for flood mapping in the PPR. First, the complex and dynamic nature of the PPR hydrology presents challenges in using well-developed hydrological and hydraulic models which often rely on rainfall-runoff data as input and are not capable of reproducing the spring runoff processes dominated by snowmelt and ‘fill and spill’ hydrology [11
]. Additionally, PPR basins may not contain any naturally-formed streams. Local attributes such as wetlands and weather conditions, can lead to small-scale, localized flooding [5
]. Second, hydrological models which rely in part on historical data often lack spatially relevant information that is required to identify hazard areas and implement locally-relevant flood management strategies [3
]. Third, effective flood risk management requires improved communication between experts and stakeholders [13
]. Improved communication does not only involve disseminating model findings to the public, it should additionally focus on encouraging participation in identifying flood concerns that will help produce information relevant to local contexts [14
]. The three needs exemplify recent calls for transdisciplinary methodologies that allow the integration of different methods and knowledge systems for accurate and relevant flood mapping [15
Public participation in flood-related spatial modeling and analysis has seen increasing appreciation and acceptance in the last decade, and led to advances in disaster awareness [16
] as well as in risk planning and management [17
]. Innovative tools, data gathering methods, and processing techniques have provided opportunities for researchers to engage with local people and establish processes that create benefits for local communities [19
]. New participatory processes for flood mapping and planning promote inclusivity and empower communities to develop their management plans [21
]. The level of public participation in spatial analysis can vary. Stakeholders and community members can measure or provide data to be fed directly into geoprocessing software; help in the interpretation of data or information; provide context, experience, and knowledge of historical events; put forward opinions and needs; and assess methods, tools, or results [20
]. Public participation in spatial modeling provides ways to integrate public knowledge, including local spatial knowledge and indigenous knowledge with conventional scientific approaches [22
In this study, we detail one such integrative participatory approach for flood mapping for an indigenous community in the PPR using spatial data and modeling techniques. We evaluate the utility of different spatial data and modeling outputs for community flood preparedness and management. First, we discuss the opportunities for using light detection and ranging digital elevation model (LiDAR DEM) and wetland DEM ponding model (WDPM) for flood mapping in the PPR. Then, using an actual LiDAR DEM combined with a spatially distributed model wetland digital elevation model (DEM) ponding model (WDPM), we display our locally relevant flood map. The map provides an initial overview of flood extent rather than a detailed analysis of flood dynamics, which was beyond the scope of this paper. Instead, we responded to the community’s need to have an accessible and up-to-date spatial tool that supports community decision making for flood resilience. The three objectives of the study were to:
Assess the ability to combine data from LiDAR, a physical survey of culverts, and WDPM to accurately portray flood extent in PPR;
Model flood extent through community-driven flood scenarios and produce meta-data to support community decision making for flood resilience;
Evaluate the utility of the modeling pathway and meta-data using satellite imagery and community reflection.
Given the complexity of hydrological processes in the PPR, the WDPM modeling approach for flood mapping presented in our case study suggests utility in other regions in the PPR to demonstrate its practicality and feasibility on a larger scale. With finer precision DEMs such as LiDAR DEM, it is possible to produce detailed water extent map for flooding scenarios, providing information such as connectivity of water bodies, water accumulation zones or ‘hotspots,’ and accurate water depth in each cell [24
]. While in previous work [11
], coarser DEM datasets did not produce accurate flooded areas compared to LiDAR DEM, in our study, we found ALOS DEM produced the most accurate flooded areas. Findings from earlier work and this study indicate that without the information of the surrounding contributing areas extending to the hydrosheds, and in the best case, watershed boundaries, the use of LIDAR DEM with WDPM leads to an exaggerated inundation area. This may be because the software does not recognize what to do with water at an edge interface. Future versions are planned that can better manage this situation. Nevertheless, we have found that having the DEM for basin or even sub-basin areas improves the performance and accuracy of WDPM.
Various modeling approaches have been proposed and reviewed for flood mapping, such as hydrological models [50
], hydrodynamic approaches [51
], and integrated modeling [53
]. In the context of Prairies, however, the combined WDPM and DEM process provided a diagnostic approach to assess flood extent with minimum data requirements. WDPM-derived water depths, in our study, were useful for calculating total inundated areas, qualitatively describing the flood extent, assessing the hazard areas based on accumulation zones, and simulating different runoff events. With land-use information for the community, it may also be possible in the future to calculate percentage area covered with water for different land-use types (e.g., agricultural, residential, and administrative), but this was not within the scope of the current study.
There are, however, some limitations to the model. Although WDPM provides a simple modeling approach to improve flood mapping for the PPR, the model’s execution time is slow. The runtime depends on the DEM size and tolerance, and the processor used, which can make WDPM computationally expensive at present. In our case study, we found that running WDPM with a powerful processor could reduce the runtime for the model at the expense of computations. Increasing the tolerance can help reduce the runtime; however, doing so affects the output from the model as we found from the running simulations at both 1 mm and 100 mm elevation tolerances. In addition, preprocessing the DEM can take some time (e.g., breaching roads at culvert points) depending on the DEM size. Because the model cannot establish the initial water distribution in the depressional storages when the DEM is created, it is important to establish the initial water distribution before estimating the flooding extent. The process requires access to an aerial photo of the community from the fall or late summer before the LiDAR survey is done, and adding and removing water from the DEM by ‘trial and error’ so that the water distribution more or less matches with the aerial photo [32
]. It can, however, be challenging to acquire aerial images for specific times of the year, which was the case in this study, and can take a lot of time trying to match the water distribution.
Additionally, at the moment, runoff depths are generated using straightforward approaches, such as either using arbitrary reference depth or rainfall frequency. For future works, runoff estimates for WDPMs which consider a range of PPR hydrological processes can be used. The cold regions hydrological model (CRHM) can produce such runoff estimates which can be used for assigning runoff depth in WDPM to produce an accurate and detailed spatial representation of flooding in the PPR [54
]. There is an ongoing effort to integrate the CRHM’s outputs, including runoff for different watershed classification and evaporation changes, with WDPM. Doing so would account for the loses in the reference water depth due to processes such as evaporation, infiltration, and snow redistribution. However, currently this integration is not feasible.
Having up-to-date information on flood inundation extents and hazards is beneficial in rural and indigenous communities in the PPR for their spatial planning and emergency preparedness. Producing flood maps in most cases is a technical process using hydrological and hydraulic models [12
]. The evaluations of flood maps by the end users, however, are rarely done. In our study, we found that community participation in flood mapping could be inclusive of local experiences and memories which can be valuable in evaluating the model-derived flood maps and providing direction for future works. Furthermore, explanation of data and modeling limitations also helped with establishing transparency in the process, which is a vital aspect of engagement [19
]. Similar findings have been confirmed by other participatory modeling studies [55
]. Interestingly, the discussions on the flood maps also led to the understanding that the community is keen towards collecting more data in the future. Our results also provide opportunities for improving flood maps using transdisciplinary methods to combine local and indigenous knowledge of flooding and blend them into model-derived flood maps (e.g., identifying high impacted areas, paying attention to risk perceptions, and locating control structures) [21
Doong et al. [57
] highlighted the importance of stakeholder engagement for improving flood mapping. In our study, engagement processes in flood mapping were initiated in early stages. Participants were involved in collecting data, determining scenarios for the model, and evaluating the utility of WDPM-generated flood maps for community flood preparedness. Other studies have also described the importance of engaging stakeholders in preliminary stages of any modeling process to increase the trust and legitimacy of modeling outputs [19
]. The feedback from the participants in the workshop shed some light on their preferences for flood maps which could help with the selection of modeling scenarios and risk mitigation strategies in the future [21
]. Key community feedback included having more data for the surrounding areas in the future for detailed flood mapping; using historical events in other communities in the PPR as scenarios; and modeling the effect of having control structures versus no control structures. Furthermore, others have noted that integrating community-specific spatial flood information can empower local and indigenous communities to take actions, develop locally relevant adaptation strategies, and build resilient communities in the era of climate change [21
Lessons from our study draw on the significance of the participation of local and indigenous communities in flood modeling and mapping studies as contributing both to better science and to reconciliation by scientists. Based on this we provide four recommendations:
Participation of public, local and indigenous communities is possible in otherwise traditionally top-down modeling practices and contributes to good practice in doing research. It also meets the calls of others doing community-engaged research or participatory research with local and indigenous communities [58
Engagement with communities facilitates the in-filling of some data gaps, overcoming unideal or incomplete data, and uncertainty in modeling. In our case, we overcame data deficiencies by being gifted access to the community-held data, co-collecting culvert points and co-validating flood maps. Furthermore, engagement can lead to the creation of innovative modeling approaches, the generation of new knowledge, and ultimately, the practicing of science that is relevant to greater society [12
Use of spatially focused tools in small, rural, and indigenous communities in the PPR can provide valuable information for identifying vulnerable regions, better spatial plans, and accordingly, better response or management strategies for floods [11
LiDAR, although an expensive tool, is a worthwhile investment, particularly in relatively flat areas such as the PPR. Investment and access to LiDAR at the catchment scale would improve the estimation of flood extent and flood risk. It would help to plan efficient management strategies and reduce the cost of flood damage and recovery in long run.
The results and findings from our study are based on only one community in the PPR which makes it difficult to say that the accuracy and utility of WDPM in other regions will be equally high. There are cases, however, where WDPM has been used in community planning [11
]. Given the growing importance of LiDAR, communities across the Prairies need access to it. In communities where spatial data is available, WDPM can provide quick initial overviews of potential flood extents and assist communities in assessing their vulnerabilities to extreme flood hazards in the future. Furthermore, the model outputs can be easily verified against aerial photographs, satellite images, and lived experiences. Community feedback is valuable for developing scenarios for future modeling works and for creating locally relevant information. While in our study we only evaluated the land covered with runoff estimated from WDPM, future, work can include economic evaluation of flood damage (e.g., roads, buildings, risk to people, etc.). However, for this, more rigorous modeling work may be needed. In our research, we demonstrated the application of WDPM and LiDAR by creating flood extent maps for an indigenous community in the PPR and used community feedback to evaluate the accuracy and utility of flood maps. We hope the methodology we have used in our work could contribute to supporting flood resilience and management in rural and indigenous communities in the PPR.
Increasing climate uncertainty and land use changes have led to an increase in extreme flooding events across the PPR, leaving many rural and Indigenous communities vulnerable to the negative impacts. Because of the unique PPR hydrological processes, conventional modeling approaches which are often focused on flooding in rivers and streams, can become insufficient and invalid. In this work, we used a spatially-focused modeling tool, WDPM, to develop flood hazard maps at a local scale. Furthermore, the use of LiDAR DEM provided a detailed estimation of flood hazards (i.e., drainage to trace pathways of runoff over the landscape, and impacts of roads on water pooling) compared to coarser DEM. We also found that without having sufficient information of the contributing areas, the total inundated areas can be exaggerated. Despite the limitations in accuracy, community members found the information to be useful. We also found community engagement to be valuable for co-producing data, providing feedback, and guiding future work.
In general, up-to-date spatial datasets, flood simulations, and accurate and detailed flood hazard maps will be important for designing flood management strategies for many communities across the PPR. Our study demonstrates the feasibility of using LiDAR datasets and WDPM approaches to identify flooding hazards in a small community. In future, more technical rigor can be applied to generate reference depths for simulations using physically-based hydrological models to improve the runoff estimates from WDPM. In addition, with the addition of information such as population, land use types, and infrastructure, WDPM generated runoff maps also have the potential to provide economic evaluations of flood damage in such communities.