DKDeniz Kilic

Deniz Kilic

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Deniz Kilic has a total of 18 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Inverse Modelling of Shallow Land Subsidence using a Hybrid Differentiable Physics-Machine Learning Approach

    by Deniz Kilic, Oriol Pomarol Moya, Gilles Erkens, Derek Karssenberg, Madlene Nussbaum, Kim M. Cohen, Esther Stouthamer

    Shallow subsidence is a key problem in many coastal plains, such as those of the Mississippi (U.S.A), the Mekong (Vietnam), Po (Italy), and Rhine deltas (Netherlands). Managing it is as important as managing anticipated sea-level rise. Accurate prediction of shallow land subsidence into the coming century, requires robust parameterization of a series of complexly interacting physical and biochemical processes (leading to consolidation, oxidation, shrinkage), that operate across heterogeneous shallow subsurface conditions. Traditional physics-based models depend on parameters that are difficult to constrain spatially, while purely data-driven approaches might give physically inconsistent results and often lack physical interpretability. We present a hybrid modeling framework that balances this tradeoff by combining a fully differentiable version of the process-based subsidence model Atlantis with neural network components for learning spatial parameter heterogeneity, enabling gradient-based parameter optimization directly from observational data.Our approach converts established isotach consolidation model and peat-oxidation calculation methods into a differentiable computational graph, allowing automatic differentiation to propagate observational constraints through the physics model. This enables joint inversion of spatially distributed InSAR-derived observations and vertical extensometer profiles to constrain process parameters at voxel scale. Crucially, we incorporate observation uncertainty (the full variance-covariance structure) of InSAR-derived measurements through a statistically rigorous loss function (Mahalanobis distance), properly accounting for spatial and temporal correlations that traditional calibration approaches neglect. First results confirm that the framework can recover peat oxidation parameters from synthetic subsidence observations; integration of InSAR-derived data with full uncertainty characterization is underway.The differentiable architecture offers several advantages: 1) principled uncertainty quantification by accounting for the error structure of input observations, 2) efficient optimization through gradient descent rather than computationally expensive sampling methods, and 3) flexible integration of heterogeneous data sources within a unified modeling framework. We demonstrate the approach spatially, using various observations for a long-managed mainly agricultural peat meadow polder (Krimpenerwaard, The Netherlands; current average elevation 1.5 m below MSL and sinking). Our methodology bridges geodetic remote sensing with process-based geotechnical modeling, contributing to improved projections of coastal relative sea-level rise by constraining subsurface processes at operationally relevant scales. The approach is computationally efficient and can be scaled to larger areas and longer timeframes, depending on the availability of novel InSAR-derived observations and subsurface data.

  • Open Access

    A Process-Based Modelling Approach to Evaluate Alternative Sustainable Land Subsidence Adaptation Pathways in the Netherlands

    by Deniz Kılıç, Gilles Erkens, Kim M. Cohen, Esther Stouthamer

    Land subsidence is a slowly progressing phenomenon that often goes unnoticed due to its gradual nature, yet it can significantly compromise long-term sustainability if left unaddressed. This challenge is particularly pronounced in coastal and deltaic regions with limited fluvial sediment input – e.g. the Netherlands, the Po River Basin, the Mekong Delta, the Mississippi Delta – where anthropogenic activities altered water tables, sediment dynamics and ecosystem health, and further such impacts with climate change and sea-level rise are expected. Any robust, future-proof adaptation strategy and spatial planning must therefore account for ongoing land subsidence, if human presence is to be viable.In the Netherlands, the situation is already severe: approximately 50% of its coastal-deltaic plain now lies below mean sea level (Koster et al., 2018) owing to soft soil consolidation, peat oxidation and mining, accumulated over centuries and never technologically halted. Even more, progressive subsidence has increasing economic costs (Van den Born et. al., 2016). Recognizing the urgency of this problem, the Dutch government and related authorities pay attention and resources at regional and national scale (e.g. platforms, knowledge centres, incentives, directives, regional deals), and several cross-disciplinary research programs have been prompted (e.g. NWA-LOSS, NOBV, DeepNL). Within NWA-LOSS (nwa-loss.nl) our work focuses on the numerical modelling. With partners, we develop and operate the land subsidence model Atlantis (Bootsma et al., 2020) that captures the interplay of soft soil consolidation, peat oxidation, climate change, and human interventions (e.g. agricultural drainage) to predict future spatial and temporal evolution of the Dutch landscape. Employing global sensitivity analyses (Morris screening and Sobol’ indices), we identify the most influential parameters and processes and integrate uncertainty quantification to ensure robust subsidence predictions.Our results reveal how shallow subsidence evolves under various climate scenarios, pinpointing ‘hotspots’ for targeted adaptation and nature based solutions (e.g. peat regeneration). Critically, our findings underscore the role of subsidence in shaping relative sea-level rise, a driver of coastal vulnerability that can profoundly influence coastal and deltaic biogeochemistry, biomorphodynamics, and hydrodynamics. Incorporating land subsidence into long-term adaptation measures is therefore essential for mitigating climate change impacts and improving the resilience of coastal and estuarine environments worldwide.References:Bootsma, H., Kooi, H., Erkens, G. (2020). Atlantis, a tool for producing national predictive land subsidence maps of the Netherlands. Proceedings of the International Association of Hydrological Sciences, 382, 415-420.Koster K., Stafleu J., Stouthamer E. (2018). Differential subsidence in the urbanised coastal-deltaic plain of the Netherlands. Netherlands Journal of Geosciences. 2018;97(4):215-227. doi:10.1017/njg.2018.11Van den Born, G. J., Kragt, F., Henkens, D., Rijken, B., Van Bemmel, B., Van der Sluis, S. (2016). Dalende bodems, Stijgende kosten, Report Planning Agency for the Environment (PBL), report nr. 1064, 93 pp., 2016.

  • Open Access

    Managing Land Subsidence in the Netherlands: A Process-Based Modelling Approach to Evaluate Alternative Sustainable Pathways

    by Deniz Kılıç, Gilles Erkens, Kim M. Cohen, Esther Stouthamer

    The Netherlands confronts significant land subsidence challenges, primarily in the low-lying, densely populated soft soil regions. Land subsidence results from drainage or loading of the clayey and peaty shallow subsurface, extraction of groundwater or hydrocarbons, and local salt mining at deeper levels. These activities are compounded by natural subsidence processes such as local to regional tectonics, glacial isostatic adjustment, and autocompaction. The impacts of land subsidence are further exacerbated by global warming and rising sea levels. While Dutch institutions acknowledge the escalating economic costs of land subsidence (e.g. Van den Born et al., 2016), the prevailing land management practices, essential for current economic activities, contribute to a policy lock-in (Seijger et al., 2017), necessitating sustainable alternatives.Our research aims to quantify explicitly the individual shallow processes of land subsidence (i.e. compression, peat oxidation) and establish their relative contributions using Atlantis, a process-based 3D land subsidence model integrated with the GeoTOP geological framework that represents the subsurface build-up and properties. This model serves as a tool for developing regional land subsidence projections up to 2100 across the Netherlands. By leveraging a combination of advanced local and regional observation data, including novel processed InSAR products, and recent findings from key national research initiatives like NWA-LOSS, NOBV, DeepNL, and Regiodeal Groene Hart, we gain a more refined understanding of subsidence processes which will be implemented in the model. Here we present the initial results of our approach by calculating the extent of human-induced shallow subsidence in the Netherlands for different future land subsidence scenarios. These scenarios were developed via a back casting method, encompassing extremes like no subsidence, minimal damage, and cessation of shallow drainage. We highlight the importance of combining numerical modelling with policy development methods such as back casting to test and evaluate optimal management pathways.Building on previous land subsidence projections (Erkens et al., 2017), our approach integrates novel data and improved process understanding. A significant advancement of our study is the incorporation of an impact quantification module. This module translates the land subsidence projections (mm.yr-1) and greenhouse gas emissions from peatlands (tonnes.yr-1) into economic terms (EUR.yr-1). This approach enables us to link the regional land subsidence to economic implications under various future projections. This tool could also help policymakers to gain insights via societal cost-benefit analyses to assess the socio-economic impacts of different adaptation and mitigation strategies. This research is pivotal in exploring and finding alternative pathways on managing land subsidence in the Netherlands, therefore providing new perspectives on how to break the current policy lock-in. ReferencesErkens, G., Stafleu, J., and Van den Akker, J. J. H. (2017). Bodemdalingvoorspellingskaarten van Nederland, versie 2017, Deltares rapport klimaateffectatlas, 2017. Seijger, C., Ellen, G. J., Janssen, S., Verheijen, E., & Erkens, G. (2017). Sinking deltas: trapped in a dual lock-in of technology and institutions. Prometheus, 35(3), 193-213.Van den Born, G. J., Kragt, F., Henkens, D., Rijken, B., Van Bemmel, B., and Van der Sluis, S. (2016). Dalende bodems, Stijgende kosten, Report Planning Agency for the Environment (PBL), report nr. 1064, 93 pp., 2


  • Open Access

    Evaluation of a high-resolution high-performance routing scheme for regional to global scale applications within Earth System Models

    by Deniz Kilic, Yann Meurdesoif, Agnès Ducharne, Jan Polcher, Josefine Ghattas

    River routing is a critical component of land surface models (LSMs), as it plays a significant role in the closure of the water balance at global scale, linking the estimation of river discharge to the one of sea level. The estimation of river discharge within LSMs also enables researchers to use widely available discharge observations to evaluate their models, and to study the human impact on discharge. However, river discharge calculation is often simplified in continental to global scale applications, as topography varies at a much smaller scale than the one of LSM grid-scales, which requires sub-grid parameterizations.In this study, we present a revision of the current routing module of the ORCHIDEE LSM (Nguyen-Quang et al., 2018) that aims to improve the accuracy of discharge estimation in an agile way. We propose a simple, parallelized, conservative routing module that is based on principal flow directions at the digital elevation model (DEM) scale and independent of the LSM grid cell, enabling the use of routing maps at various resolutions. The influence of topography is factored in by the topographic index, i.e. a product of slope and pixel length in each routing cell. The routing can be solved directly on DEM native grid using conservative interpolations of run-off and drainage computed from LSM. To reduce the computational cost, we developed an upscaling method by aggregating DEM pixels into irregular hydrological transfer units (HTUs), which respect the basin hierarchy by construction, therefore making the computation of effective topographic index straightforward, and which is constrained by validity of the numerical stability criteria. This upscaling method drastically reduces the computational cost, by a factor depending on the targeted resolution, without compromising the discharge estimation.We test this new routing module via offline simulations, to evaluate the discharge within 10 of the world's largest river basins, at the outlet and in upstream sub-catchments. To this end we use a 2km version of the MERIT global DEM to derive information on flow direction, slope and pixel length; and the GRDC global river discharge observation dataset to evaluate the simulated river discharge. First, using the routing at the DEM pixel scale, we will tune the lag time of reservoirs to improve the discharge estimation. Then, we will test the stability of performances when upscaling the routing over a range of HTU lengths. This work is pivotal for the use of ORCHIDEE and its routing scale at various spatial scales, either off-line or coupled to the IPSL climate model, especially with its scalable atmospheric dynamical core, which is based on a quasi-uniform icosahedral-hexagonal mesh, and can be used for both global or limited-area simulations (Dubos et al., 2015).References:Dubos, T., Dubey, S., Tort, M., Mittal, R., Meurdesoif, Y., and Hourdin, F.: DYNAMICO-1.0, an icosahedral hydrostatic dynamical core designed for consistency and versatility, Geosci. Model Dev., 8, 3131–3150, https://doi.org/10.5194/gmd-8-3131-2015, 2015.Nguyen-Quang, T., Polcher, J., Ducharne, A., Arsouze, T., Zhou, X., Schneider, A., and Fita, L.: ORCHIDEE-ROUTING: revising the river routing scheme using a high-resolution hydrological database, Geosci. Model Dev., 11(12), 4965-4985, https://doi.org/10.5194/gmd-11-4965-2018, 2018.

  • Assessing water and energy fluxes in a regional hydrosystem: case study of the Seine basin

    by Deniz Kilic, Agnès Rivière, Nicolas Gallois, Agnès Ducharne, Shuaitao Wang, Philippe Peylin, Nicolas Flipo

    While it is well accepted that climate change and growing water needs affect long-term sustainable water resources management, performing accurate simulations of water cycle and energy balance dynamics at regional scale remains a challenging task. Traditional Soil-Vegetation-Atmosphere-Transfer (SVAT) models are used for numerical surface water and energy simulations. These models, by conception, do not account for the groundwater lower boundary that permits a full hydrosystem representation. Conversely, while addressing important features such as subsurface heterogeneity and river–aquifer exchanges, groundwater models often integrate overly simplified upper boundary conditions ignoring soil heating and the impacts of vegetation processes on radiation fluxes and root-zone uptakes. In this paper, one of the first attempts to jointly model water and energy fluxes with a special focus on both surface and groundwater at the regional scale is proposed on the Seine hydrosystem (78,650 km 2 ), which overlays one of the main multi-aquifer systems of Europe. This study couples the SVAT model ORCHIDEE and the process-based hydrological–hydrogeological model CaWaQS, which describes water fluxes, via a one-way coupling approach from ORCHIDEE toward CaWaQS based on the blueprint published by [de Marsily et al., 1978]. An original transport library based on the resolution of the diffusion/advection transport equation was developed in order to simulate heat transfer in both 1D-river networks and pseudo-3D aquifer systems. In addition, an analytical solution is used to simulate heat transport through aquitards and streambeds. Simulated ORCHIDEE surface water and energy fluxes feed fast surface runoff and slow recharge respectively and then is used as CaWaQS forcings to compute river discharges, hydraulic heads and temperature dynamics through space and time, within each of the hydrosystem compartments. The tool makes it possible to establish a fully consistent water and energy budget over a period of 17 years. It also simulates temperature evolution in each aquifer and evaluates that river thermal regulation mostly relies by order of importance on short wave radiations (109.3 W · m - 2 ), groundwater fluxes (48.1 W · m - 2 ) and surface runoff (22.7 W · m - 2 ).


  • Open Access

    Integrated modeling of water and heat fluxes in the Seine hydrosystem, France

    by Deniz Kilic, Agnès Rivière, Nicolas Flipo, Agnès Ducharne, Philippe Peylin, Patrick Goblet

    Given the current climate and anthropogenic evolution, water management becomes one of the greatest challenges of the 21st century. Number of studies have analyzed observed hydrologic trends and their connections with the changing climate. Impacts include changes in runoff, river discharge and groundwater recharge. Water quality is also impacted, through its many facets including the water temperature. Despite the important progress made in climate modelling, the impact of the predicted global warming on hydrological processes remains uncertain; particularly, in large hydrosystems. The Seine River basin has a surface of 78,650 km², it includes the Seine River and its 50 tributaries, it is populated by 30% of France inhabitants. The Seine River basin crosses 14 departments and 4 regions, including the Paris metropolitan area. Climate change poses a vulnerability due to its potential political, social, and economic consequences in the Seine basin. The agricultural activities and number industries depend on water resources or are located on the river sides. Our ability to adapt water resource management strategies to the climate change depends on our ability to understand and estimate the actual evolution of water resource. The terrestrial water budget is now considered as a single continuum. This integrated conceptualisation needs to simulate the spatial and temporal dynamics of water exchanges between the surface and groundwater. Here we propose to improve the representation of the surface water budget with the goal to decrease the uncertainty of the whole water budget of the Seine hydrosystem. We used the process-based physical land surface model ORCHIDEE (tag 2.2) to estimate surface water budget and heat balance for the period 1980-2018. This application takes advantage of high resolution land-use and albedo maps from ESA-CCI database, and various soil map databases. The model was satisfactorily able to reproduce the discharges of each sub-catchment, the actual evapotranspiration fluxes and LAI. With these results, we are able to estimate the the partitioning of the surface water balance of each catchment of the Seine basin. These results have wide ranging implications such as the estimation of energy balance in the basin, the estimation of spatialisation of the aquifer recharge, and the feedback between aquifers and the surface.