A comparison of three models used to determine water fluxes over the Albany Thicket, Eastern Cape, South Africa
Introduction
Terrestrial biomes are strongly linked with climatic factors, soil types, land-form and disturbance regime (Smart, 2016). The ecophysiological characteristics of the plant functional types, interaction with soil effects and climate constraints, give each biome a unique set of climate interactions with respect to the nature and strength of the interactions with the atmosphere. South Africa's Sub-tropical Thicket was recognised as a biome in mid-1990s, before then, it was seen as mixture of different biomes (Smart, 2016). The regional climate varies from subtropical to warm-temperate through frost-free or semi-arid to sub-humid with a bimodal rainfall pattern (Cowling et al., 2005). The thicket can be distinguished by the presence or absence of succulent species and scrutinized by species evolutionary lineage (Smart, 2016). It comprises evergreen and weakly deciduous trees and shrubs of 2–5 m height (Cowling et al., 2005).
It is obvious that more attention has been given to carbon fluxes than water fluxes in historical studies in the AT. Even in semi-arid region of South Africa where water is scarce, little is known about the water flux across the Sub-tropical Thicket (Gwate et al., 2018). Some water related studies conducted on thicket in South Africa include the assessment of annual ET of some native thicket by Meijninger and Jarmain (2014). It was noted that annual ET for dune and spekboom thicket in the Western Cape ranges between 515 and 660 mm respectively while valley thicket in KwaZulu-Natal is 755 mm. Dye et al. (2001) noted that removal of riparian wattle, an invasive alien tree will lead to significant reductions in annual ET which in turn will enhance streamflow. So far, studies on ET over AT include: measurement of ET using an EC system (Gwate et al., 2016); development of ET predictive model to be applied at biome scale (Gwate et al., 2018); validation of improved simple single layer ET model (Penman-Monteith-Palmer (PMP)) using large aperture scintillometer (LAS) and EC system data (2015-2016) (Gwate et al., 2019). Gwate et al. (2016) noted that during the experiment period between 2015-2016, the observed ET was more than the amount of rainfall received, possibly because the vegetation may be accessing groundwater in addition to the high water storage capacity of the vegetation. Contrarily, Cleverly et al. (2006) found that ET from an invasive riparian shrub (Tamarix chinensis) did not decline with increasing groundwater depth; instead, ET increased by 50%, from 6 mm/day to 9 mm/day, as the water table receded at nearly 7 cm/day. Using frequency distribution analysis, Meijninger and Jarmain (2014) demonstrated that the variation in ET of the thicket vegetation can be large and studies over AT referenced above were based on relatively short periods of observed data which can result in more uncertainties for seasonal analysis. With long-term data, assessment of seasonal or annual variability in water fluxes can be carried out with greater certainty. Therefore, this study aims to evaluate water fluxes over AT covering a longer time period, ultimately aiming to improve understanding of the long-term water use and the variation in water use of this biome. Daily ET (from October 2015 to April 2018) from an eddy covariance system was compared with ET generated from a remotely-sensed product (MOD16), BGC-MAN and a biophysical model, namely the Penman-Monteith-Leuning model.
These three models have the capability of simulating ET over the thicket ecosystem. The BGC-MAN model is a process-based ecosystem model with capability of simulating forest management activities (Song et al., 2019). The model operates on a daily time step and provides a mechanistic description of the movement of water, energy, nitrogen and carbon in an ecosystem (Pietsch, 2014). The BGC-MAN is an extension of the Biome-BGC 4.2 model (Thornton et al., 2005; Akujärvi et al., 2019) with management options. The model is driven by meteorological, site specific and ecophysiological data to simulate fluxes of a given area of land (Hidy et al., 2012; Pietsch and Hausenauer, 2005; Pietsch et al., 2005). In addition, the remotely-sensed Moderate Resolution Imaging Spectroradiometer (MODIS) product uses sensors on Terra and Aqua satellites to provide global observations of the Earth's land, atmosphere and oceans in the visible and infrared regions of the spectrum. MOD16 is a MODIS global ET product that provides continuous ET datasets (both actual ET and potential ET) at 0.5 km spatial resolution and 8-day, monthly, and annual temporal scale using Priestley-Taylor, Penman-Monteith and land surface model approaches (Aguilar et al., 2018; Khan et al., 2018). Also, the PML model is based on the Penman-Monteith equation and remotely-sensed leaf area index (Morillas et al., 2013). It uses three methods to improve the soil evaporation influence in total ET estimates on daily time step.
Section snippets
Study area
The study area is located within eZulu Game Reserve situated approximately 70km north west of Grahamstown, Eastern Cape Province (33° 01′ 08.929′' S 26° 04′ 47.860′' E). The vegetation is described as the Great Fish Thicket of the AT biome and is mainly restricted to relatively warmer sub-escarpment coastal plains (Mucina and Rutherford, 2006; Duker et al., 2015). The AT is known as a biodiversity hotspot with succulents, deciduous and semi-deciduous woody and dwarf shrubs (Hoare et al., 2006;
Environmental conditions
Variations in ET from EC and average daily environmental conditions during the study period are presented in Fig. 2 (a–c) and Table 3 respectively. Total rainfall amount received during the study for 2015-2016, 2016-2017, and 2017-2018 were 280.4, 256.3 and 232.9 mm respectively and daily maximum rainfall recorded were 25.1, 20.2 and 36.7 mm respectively. Among the environmental parameters, rainfall was the most varied parameter with high CV of 336.78%, 346.79% and 370.55% for 2015-2016,
Discussions of results
The discrepancies in the results could be attributed to some level of uncertainties associated with ET methods and input data especially remote-sensed product like MODIS that has some unresolved issues of spatial scale mismatch among coarser meteorological forcing and cloud-free images, which results to higher uncertainties in ET estimations (Long et al., 2014). Although, the differences in ET estimates is not peculiar to these models used in this study. Chen et al. (2014) found that the ET
Conclusions
Water fluxes over AT were assessed using rainfall and ET data as well as other environmental variables. Among the environmental parameters, rainfall was the most varied parameter in the region and ET trend mostly followed that of rainfall and SWC. Among the three hydrological years in this study, total ET for 2016-2017 exceeded rainfall received by about 7% which shows that Albany Thicket is likely to be supported by groundwater at some point but not all the time as claimed by Gwate et al.
Declaration of Competing Interest
None.
Acknowledgements
We acknowledge Red Meat Research Development-SAand the National Research Foundation of South Africa (National Equipment Programme) for funding the equipment and data collection. In addition, we thank Rhodes University for providing financial support to GIE.
References (60)
- et al.
Comparison of satellite-based evapotranspiration models over terrestrial ecosystems in China
Remote Sens. Environ.
(2014) - et al.
On the origin of Southern African subtropical thicket vegetation
S. Afr. J. Bot.
(2005) - et al.
Actual evapotranspiration in drylands derived from in-situ and satellite data: assessing biophysical constraints
Remote Sens. Environ.
(2013) Evapotranspiration evaluation models based on machine learning algorithms-a comparative study
Agric. Water Manag.
(2019)- et al.
Stand-alone uncertainty characterization of GLEAM, GLDAS and MOD16 evapotranspiration products using an extended triple collocation approach
Agricultural and Forest Meteorology
(2018) - et al.
Improvements to a MODIS global terrestrial evapotranspiration algorithm
Remote Sensing of Environment
(2011) - et al.
Incorporating forest growth response to thinning within Biome-BGC
Forest Ecol. Manag.
(2007) - et al.
BGC-model parameters for tree species growing in central European forests
Forest Ecol. Manag.
(2005) - et al.
Application of BIOME-BGC model to managed forests: sensitivity analysis
For. Ecol. Manag
(2006) - et al.
An improved algorithm for estimating incident daily solar radiation from measurements of temperature, humidity, and precipitation
Agric. For. Meteorol.
(1999)