TY  - GEN
A1  - Kühn, Michael
A1  - Schöne, Tim
T1  - Multivariate regression model from water level and production rate time series for the geothermal reservoir Waiwera (New Zealand)
T2  - Energy procedia
N2  - Water management tools are necessary to guarantee the preservation of natural resources while ensuring optimum utilization. Linear regression models are a simple and quick solution for creating prognostic capabilities. Multivariate models show higher precision than univariate models. In the case of Waiwera, implementation of individual production rates is more accurate than applying just the total production rate. A maximum of approximately 1,075 m3/day can be pumped to ensure a water level of at least 0.5 m a.s.l. in the monitoring well. The model should be renewed annually to implement new data and current water level trends to keep the quality.
KW  - geothermal reservoir
KW  - water management
KW  - data based model
KW  - multivariate regression
KW  - coefficient of determination
KW  - scenario analysis
Y1  - 2017
U6  - https://doi.org/10.1016/j.egypro.2017.08.196
SN  - 1876-6102
VL  - 125
SP  - 571
EP  - 579
PB  - Elsevier
CY  - Amsterdam
ER  - 
TY  - JOUR
A1  - Zimmermann, Alexander
A1  - Uber, Magdalena
A1  - Zimmermann, Beate
A1  - Levia, Delphis F.
T1  - Predictability of stemflow in a species-rich tropical forest
JF  - Hydrological processes
N2  - Numerous studies investigated the influence of abiotic (meteorological conditions) and biotic factors (tree characteristics) on stemflow generation. Although these studies identified the variables that influence stemflow volumes in simply structured forests, the combination of tree characteristics that allows a robust prediction of stemflow volumes in species-rich forests is not well known. Many hydrological applications, however, require at least a rough estimate of stemflow volumes based on the characteristics of a forest stand. The need for robust predictions of stemflow motivated us to investigate the relationships between tree characteristics and stemflow volumes in a species-rich tropical forest located in central Panama. Based on a sampling setup consisting of ten rainfall collectors, 300 throughfall samplers and 60 stemflow collectors and cumulated data comprising 26 rain events, we derive three main findings. Firstly, stemflow represents a minor hydrological component in the studied 1-ha forest patch (1.0% of cumulated rainfall). Secondly, in the studied species-rich forest, single tree characteristics are only weakly related to stemflow volumes. The influence of multiple tree parameters (e.g. crown diameter, presence of large epiphytes and inclination of branches) and the dependencies among these parameters require a multivariate approach to understand the generation of stemflow. Thirdly, predicting stemflow in species-rich forests based on tree parameters is a difficult task. Although our best model can capture the variation in stemflow to some degree, a critical validation reveals that the model cannot provide robust predictions of stemflow. A reanalysis of data from previous studies in species-rich forests corroborates this finding. Based on these results and considering that for most hydrological applications, stemflow is only one parameter among others to estimate, we advocate using the base model, i.e. the mean of the stemflow data, to quantify stemflow volumes for a given study area. Studies in species-rich forests that wish to obtain predictions of stemflow based on tree parameters probably need to conduct a much more extensive sampling than currently implemented by most studies. Copyright (c) 2015 John Wiley & Sons, Ltd.
KW  - stemflow
KW  - rainfall partitioning
KW  - multivariate regression
KW  - tropical lowland rainforest
KW  - ecohydrology
Y1  - 2015
U6  - https://doi.org/10.1002/hyp.10554
SN  - 0885-6087
SN  - 1099-1085
VL  - 29
IS  - 23
SP  - 4947
EP  - 4956
PB  - Wiley-Blackwell
CY  - Hoboken
ER  -