Introduction

The number and diversity of water-related challenges are large and are expected to increase in the future (Wagener et al. 2010; Lall 2014). Even today, the ideal condition of having the appropriate amount of good-quality water at the desired place and time is most often not satisfied (Biswas and Tortajada 2010; Droogers and Bouma 2014). It is likely that climate variability and change will intensify food insecurity by water shortages (Wheeler and Braun 2013), loss of access to drinking water (Rockström et al. 2012), and increased risk of natural hazards from floods and soil erosion. Current and future water-related challenges are location and time specific and can vary from impact of glacier dynamics (Immerzeel et al. 2011), economic and population growth (Droogers et al. 2012), floods or extended and more prolonged droughts (Dai 2011), amongst others.

In response to these challenges, hydrologists and water resource specialists are developing modeling tools to analyze, understand and explore solutions to support decision makers and operational water managers (Pechlivanidis et al. 2011). Despite difficulties in connecting the scientific advances in hydrological modeling with the needs of decision makers and water managers, progress has been made and there is no doubt that modeling tools are indispensable in what is called good “water governance” (Droogers and Bouma 2014; Liu et al. 2008).

The strength of hydrological models is that they can provide output at high temporal and spatial resolutions, and for hydrological processes that are difficult to observe on the large scale that they are generally applied on (Bastiaanssen et al. 2007). The most important aspect of applying models is in their use in exploring different scenarios, expressing for example, possible effects of changes in population and climate on the water cycle (Droogers and Aerts 2005). Models are also applied at the operational level to explore interventions (management scenarios) to be used by water managers and policy makers. Examples of this are changes in reservoir operation rules, water allocation between sectors, investment in infrastructure such as water treatment or desalination plants, and agricultural and irrigation practices. In other words: models enable hydrologists and water managers to change focus from a re-active towards a pro-active approach.

Over the past decades, the land surface and hydrologic communities have made substantial progress in understanding the spatial presentation of fluxes of water and energy (Abbott et al. 1986; Wigmosta et al., 1994; Van der Kwaak and Loague 2001; Rigon et al., 2006). Their efforts have led to the development of well-known hydrological models, such as, e.g., VIC (Liang et al. 1994, 1996), SWAT (Neitsch et al. 2009), TOPKAPI-ETH (Finger et al. 2011; Ragettli and Pellicciotti 2012; Ragettli et al. 2013; Ragettli et al. 2014), LISFLOOD (Van Der Knijff et al, 2010), SWIM (Krysanova et al. 2015; Krysanova et al. 2000; Krysanova et al., 1998), HYPE (Lindström et al. 2010), mHM (Samaniego et al., 2010), PCR-GLOBWB (Beek and Bierkens 2008; Bierkens and Beek 2009; Wada et al. 2010; Sperna Weiland et al. 2010), MIKE-SHE (Refshaard and Storm 1995; Oogathoo et al. 2008; Deb and Shukla 2011) and GEOtop (Rigon et al., 2006; Endrizzi et al. 2013; Endrizzi et al. 2011), amongst others. The number of existing hydrological models is probably in the tens of thousands (Droogers and Bouma 2014). Some existing model reviews cover a substantial number of models: IRRISOFT (Irrisoft 2014): 114; USGS (2014): 110; EPA (2014): 211; USACE (HEC 2014): 18.

All these hydrological models are different with respect to (i) the number and detail of hydrological processes that are integrated, (ii) their field and (iii) scale of application, and (iv) the way they are implemented. Whereas, for example, the SWIM (Krysanova et al. 2015; Krysanova et al. 2000; Krysanova, Müller-Wohlfeil, and Becker 1998) and HYPE (Lindström et al. 2010) models both include all major hydrological processes, the SWIM model is typically developed for large-scale (large river basins to continental) applications, and the HYPE model operates on the sub-basin scale. Therefore, these models contain less detail, in contrast to fully distributed models operating at grid level, such as, e.g., GEOtop (Rigon et al., 2006; Endrizzi et al. 2013; Endrizzi et al. 2011) and TOPKAPI-ETH (Finger et al. 2011; Ragettli and Pellicciotti 2012; Ragettli et al. 2013; Ragettli et al. 2014). Models like, e.g., MIKE-SHE (Refshaard and Storm 1995; Oogathoo et al. 2008; Deb and Shukla 2011) and LISFLOOD (Van Der Knijff, Younis, and De Roo 2010) have the advantage of being flexible in terms of the spatial and temporal resolutions, but their disadvantages are that they do not include glacier processes and that they are not open source and therefore not available to the larger community.

It is clear that all these models have their pros and cons in terms of (i) processes integrated, (ii) field of application, (iii) scale of application, and (iv) implementation. Table 1 shows the pros and cons of some well-known hydrological models, including the Spatial Processes in HYdrology (SPHY) model. Over the last couple of years, we have developed the SPHY model, and improved its usefulness by applying the model in various research projects. SPHY has been developed with the explicit aim of simulating terrestrial hydrology under various physiographical and hydroclimatic conditions by integrating key components from existing and well-tested models: HydroS (Droogers and Immerzeel 2010), SWAT (Neitsch et al. 2009), PCR-GLOBWB (Beek and Bierkens 2008; Bierkens and Beek 2009; Wada et al. 2010; Sperna Weiland et al. 2010), SWAP (Dam et al. 1997) and HimSim (Immerzeel et al. 2011). Based on Table 1 it is clear that SPHY (i) integrates most hydrologic processes, including glacier processes, (ii) has the flexibility to study a wide range of applications, including climate and land use change impacts, irrigation planning, and droughts, (iii) can be used for catchment- and river-basin-scale applications as well as farm- and country-level applications, and has a flexible spatial resolution, and (iv) can easily be implemented. Implementation of SPHY is relatively easy because (i) it is open source, (ii) input and output maps can directly be used in GIS, (iii) it is set up modular in order to switch on/off relevant/irrelevant processes and thus decreases model run time and data requirements, (iv) it needs only daily precipitation and temperature data as climate forcing, (v) it can be forced with remote sensing data, and (vi) it uses a configuration file that allows the user to change model parameters and choose the model output that needs to be reported.

Assessing the impacts of environmental change on soil erosion and sediment yield at the large catchment scale remains one of the main challenges in soil erosion modelling (Poesen, 2018; de Vente et al., 2013). Most soil erosion and sediment yield models adopt simplified model formulations, are applied at low temporal resolutions, and often only partly represent the impacts of changes in land use or climate conditions. This often leads to unreliable results that do not sufficiently increase process understanding or support decision-making (de Vente et al., 2013). From the available soil erosion models, process-based models aim to incorporate the most relevant processes driving soil detachment, sediment transport and deposition, see (Morgan, 2005) for an overview of process-based models. Process-based models often run at small spatial (hillslope to small catchment) and temporal scales (sub-hourly to daily time steps) and require detailed input data, such as (sub-)hourly precipitation and topographic data, and incorporate a large number of calibration parameters (Govers, 2011). At larger scales, soil erosion is often assessed using empirical erosion models, see de Vente et al. (2013) for an overview. These models are derived from field studies where soil erosion has been observed under different land use, management, soil, climate, and topographical conditions. The best-known and applied empirical model is the Universal Soil Loss Equation (USLE; Wischmeier & Smith, 1978) and its derivatives RULSE (Renard et al., 1997) and MUSLE (Williams, 1995). While the empirical formulations of the USLE were obtained at plot-scale, the model is often applied at much larger scales, sometimes in combination with a sediment transport capacity equation or a sediment delivery ratio to assess sediment yield (e.g. Van Rompaey et al., 2001; de Vente et al., 2008). Due to its simplicity, the USLE can be applied with a relatively limited amount of input data. However, their main restriction is the limited number of processes accounted for (e.g. the USLE and RUSLE based models only consider sheet and rill erosion) and the limited potential to evaluate the impacts of changes in climate and land management (Govers, 2011; de Vente et al., 2013). Furthermore, these models are typically applied at annual time steps, largely neglecting intra-annual variation of climate and vegetation conditions.

Most current soil erosion models have a limited potential for application at larger temporal and spatial scales (i.e. process-based models) or lack sufficient representation of the underlying soil detachment and sediment transport processes and sensitivity to changes in land use or climate (i.e. empirical models), making them of limited use for scenario studies and process understanding. We have extended the SPHY model with a soil erosion module based on the process-based Morgan-Morgan-Finney erosion model (MMF; Morgan & Duzant, 2008) that allows evaluating the impacts of land use, land management and climate conditions on erosion and sediment yield from local to regional scales (Eekhout et al., 2018).

The objectives of this manual are:

• Introduce and present the SPHY model (v3.0)

• Present the SPHY model (v3.0) theory and demonstrate some typical applications

• Provide the steps that are required to install the SPHY model as a standalone application

• Learn how-to prepare model data for a SPHY model for your own area of interest

The model source code is in the public domain (open access) and can be obtained from the SPHY model website free of charge (www.sphy-model.org). The two peer-reviewed open-access publications of the SPHY model can be found at https://doi.org/10.5194/gmd-8-2009-2015 (Terink et al., 2015) and https://doi.org/10.5194/esurf-6-687-2018 (Eekhout et al., 2018).

Table 1: Pros (+) and cons (-) of some well-known hydrological models, including the SPHY model. A categorization is made between (i) processes that are integrated, (ii) field of application, (iii) scale of application, and (iv) implementation.