GR5J hydrological model code

In a river basin, the mass balance equation links total water storage (TWS), precipitation (P), actual evapotranspiration (E) and river discharge (Q). Though, a focused portion of the catchment, located between two river discharge stations, can be studied. Indeed, TWS in a downstream sub-catchment can be estimated based on the mass balance equation dTWS/dt = P + Qin - E - Qout - Qgw; where Qin, Qout, Qgw are respectively incoming river inflow, outcoming river discharge, and potential groundwater import/export in a surrounding basin. Among the different water fluxes, P, Qin and Qout can be measured, whereas actual evapotranspiration and Qgw should be estimated with a model. E and Qgw can be estimated with the lumped parameter rainfall-runoff hydrological model GR5J (Pushpalatha et al., 2011), which allow to quantify daily TWS at the scale of single hydrological basins.
Here we provide a MATLAB code of the GR5J model.
This model is forced with precipitation, temperature and potential evapotranspiration and computes actual river discharge.
The model is based on two storage compartments - production and routing stores - which mimic the typical response of soils and groundwater to antecedent precipitation and evapotranspiration. Snow is considered and it is estimated following the method described in the HBV (Hydrologiska Byrns Vattenbalansavdelning) model (Lindstrom et al., 1997): mean catchment temperature defines both rainfall/snowfall partitioning and snow melt events.
It is worth noting that GR5J does not need the specific knowledge of any intrinsic structure/property of the basin. The model is parsimonious and designed to model river discharge.
Although the GR5J model is a simplified conceptual model, where only five mathematical parameters define the dynamics of the two stores and their relations, it has proven skillful in predicting river discharge better than more complex models (de Lavenne et al., 2016) and has been successfully applied to represent groundwater storage changes in Nepal rivers (Andermann et al., 2012).
GR5J parameters are calibrated using a Marquard-Levenberg least squares regression analysis using root mean square error on the logarithm of observed river discharge to limit the impact of floods and promote the description of the whole water cycle.

在流域范围内，质量平衡方程将总储水量（TWS）、降水（P）、实际蒸散量（E）和河流径流量（Q）相互关联。尽管如此，可以针对位于两个河流径流站之间的特定流域区域进行深入研究。实际上，下游子流域的总储水量可以根据质量平衡方程 dTWS/dt = P + Qin - E - Qout - Qgw 进行估算；其中，Qin、Qout 和 Qgw 分别代表周边流域的入境河流补给、出境河流径流和潜在地下水流进出口。在不同的水流通量中，P、Qin 和 Qout 可以进行测量，而实际蒸散量和 Qgw 则需要通过模型进行估算。E 和 Qgw 可以利用集总参数降雨径流水文模型 GR5J（Pushpalatha 等人，2011年）进行估算，该模型能够对单一流域尺度的每日总储水量进行量化。在此，我们提供了 GR5J 模型的 MATLAB 代码。该模型以降水、温度和潜在蒸散量为强迫项，计算实际河流径流量。模型基于两个存储室——生产室和路由存储室——来模拟土壤和地下水对前期降水和蒸散量的典型响应。模型考虑了降雪，并采用 HBV（Hydrologiska Byrns Vattenbalansavdelning）模型（Lindstrom 等人，1997年）中描述的方法进行估算：平均流域温度既定义了降雨/降雪分配，也定义了融雪事件。值得注意的是，GR5J 模型无需了解流域的特定结构/属性知识。该模型简洁高效，旨在模拟河流径流量。尽管 GR5J 模型是一个简化的概念模型，其中仅用五个数学参数定义了两个存储室及其相互关系，但它已被证明在预测河流径流量方面优于更复杂的模型（de Lavenne 等人，2016年），并且已成功应用于代表尼泊尔河流中的地下储水变化（Andermann 等人，2012年）。GR5J 参数通过使用均方根误差对观测到的河流径流量对数进行 Marquard-Levenberg 最小二乘回归分析进行校准，以限制洪水的影响并促进整个水循环的描述。