Introduction

It is widely recognized that hydrological processes and water balances between mountain and lowland watersheds are significantly different. Mountain watersheds with sharp wet-dry seasonal transitions and steep gradients in temperature and precipitation with elevation make the difficulties lie in quantifying the hydrologic change by integrating the temporal and spatial distribution of water resources (Bales et al., 2006; de Jong et al., 2005). Previous studies simulated this relatively natural regions based on the supply (precipitation)-demand (potential evapotranspiration)-storage (water storage in the soil) hypothesis (Milly, 1994). However, their adjacent downstream areas where slope are flat or gentle (≤25) have dissimilar hydrological processes relative to upstream. Streamflow shows crossed, looped and human intervening, needing another hydrological regime to explain. Few studies have been conducted to quantify these differences of hydrological characteristics between mountain and lowland watersheds (Berihun et al., 2019; Weingartner et al., 2007).
Climate and landscape characteristics are the primary factors determining watershed hydrological processes (Berihun et al., 2019). Their influences vary with different watershed characteristics and the agro-ecological settings. For example, 1) climate and farming controls water sources.Precipitation is the main source for runoff generation in mountain watersheds, whereas irrigation is another important water source for lowland watersheds (e.g. accounting for 27% in Lake Taihu Basin, China (Huang et al., 2018a). 2) Slope is not the driving force for streamflow. In lowland watersheds, the target water levels human required regulate the flow direction. 3)Alternative combination of land use options generates huge discrepancy in hydrological processes . Polder is the main geographic unit of lowland watersheds. More than half of the total area are covered by farmland and surface water (Vermaat & Hellmann, 2010). The water storages such as ponds can increase the retention time of surface runoff by 6 months within polders (Cui et al., 2019). Moreover, large area of paddy lands will produce irrigation-related evaporation enhancement. 4) Water conservancy facilities hinder the natural flow exchange . Pumping stations commonly existed in lowlands interfere with the hydrological connections between polders and their receiving water body (Hesse et al., 2008), for example, inducing 8.6% reduction of annual runoff to surrounding rivers (Yan et al., 2018). In conclusions, predicting the responses of hydrological processes under future climate change and land use scenarios is favorable for applying sustainable land and water managements to adapt these impacts, in return resulting to an increase in available water. However, the research of individual and synergetic influence of climate change and land use on hydrological processes with various watershed characteristics is still limited (Gusarov, 2019).
The process-based models were widely employed to predict contributions of driving factors on streamflow variation via physical process simulations (Chen et al., 2019; Li et al., 2015; Wang et al., 2019). However, most models (e.g. SWAT, HSPF, Xin’anjiang and MIKE-SHE) were suitable for freely draining areas with sloping surfaces, and not designed for lowland polders with shallow groundwater (Hesse et al., 2008; Yan et al., 2016). They failed to reflect the complicated water management operation, especially lacking the irrigation and drainage processes simulation in crop fields (Salmon et al., 2015). For example, water conservancy facilities like dikes dramatically influence in simulating the quick return flow process through drainage systems (Tsuchiya et al., 2018). Moreover, pumping stations result in multiple drainage outlets within polders. More importantly, many existing models fail to accurate consider the characteristics of paddy field and ponds. For example, Soil Water Atmosphere Plant Model (SWAP) treats paddy fields as upland areas, which actually are the depression areas (Utset et al., 2007). Others set ponds and paddy lands as reservoirs (Xie & Cui, 2011). However, water balance of reservoir is calculated in volume, while paddy fields calculates water balance in depth (Tsuchiya et al., 2018). Up to now, the simulation methods aimed to modeling the hydrological processes in lowland artificial watersheds are the simplifications of real situations, inducing that not all the vital processes are taken into account (Lai et al., 2016; Su & Luo, 2019).
In our study area, there are two typical hydrological systems: mountain watersheds and lowland artificial watersheds. Aimed to identify the relative contributions of climatic and land use to streamflow varying with different watershed characteristics, we used the developed raster-based Xin’anjiang model to clarify the mechanism of runoff generation within mountain areas, and polder areas were treated separately which used Nitrogen Dynamic Polder (NDP) model to simulate the artificial drainage and natural flow components (Huang et al., 2018a). As changes in land use and climate are expected to intensify in future, it is a task and challenge to identify further hydrological responses considering these changes, for devising sustainable land and water management strategies. The modelling approach in this study is transferable to other watersheds.