Elsevier

Journal of Hydrology

Volume 348, Issues 3–4, 15 January 2008, Pages 480-495
Journal of Hydrology

Modelling interception in coastal and montane rainforests in northern Queensland, Australia

https://doi.org/10.1016/j.jhydrol.2007.10.019Get rights and content

Summary

This paper reports a comparison of measured and modelled interception for three different forest types at six rainforest locations in northern Queensland. The Gash interception model was able to reproduce cumulative interception at the sites accurately, provided an appropriate value of canopy storage capacity (S) was used, 2.0–3.6 mm. These values are significantly higher than S values generally reported in other rainforest studies (∼1 mm) and the reason may be that Australian rainforests contain many epiphytes and mosses, which can trap significant quantities of water within the canopy. There is also some evidence of a seasonal variation in S and wet canopy evaporation rate (E), both being lower in the dry season than the wet season. However, although the rainfall rate (R), S and E all affect the seasonal value of interception, the changes in these three parameters tend to compensate and so the biggest factor affecting seasonal variations in interception is the number of small storms. The consequence of this is that it is still possible to get good estimates of seasonal and annual interception using R, S and E values that are fixed for the entire year. Values of E fell in the range 0.35–0.81 mm h−1, which are 1.4–9 times the concurrent rates estimated using the Penman–Monteith equation. This implies that either our rainforests received very large amounts of advected energy during rain storms, or the Penman–Monteith E values are too low. Some advection of energy to our sites is quite feasible given their proximity to the ocean and generally well exposed locations. However, most of the above discrepancy is probably due to underestimation of the Penman–Monteith values of E, because of errors in the estimation of the above canopy relative humidity, due to the use of weather data adjacent to rather than above the forests and inherent difficulties of measuring the very high humidity’s that occur during rainfall.

Introduction

The interception of water on rainforest canopies and its subsequent evaporation back into the atmosphere (i.e. the interception loss) has been shown to be an important factor influencing both local climate (e.g. Salati and Vose, 1984) and downstream river flows (Bruijnzeel, 1996). In northern Queensland, tropical rainforests cover large areas of the coastal fringe in the Wet Tropics region and play a major role in the generation of runoff across the coastal plains and into the Coral Sea. The amount of runoff from these tropical forests is highly influenced by the canopy interception loss, yet until recently this has not been well quantified. Measurements around three single rainforest trees in the sub-tropical rainforest in south east Queensland indicated that up to 40% of the total water input to the canopy (rain plus cloud interception) was lost as interception (Hutley et al., 1997). More recently McJannet et al., 2007a, McJannet et al., 2007b have reported comprehensive, long term stand level interception losses in six rainforest locations in the wet tropics that span the altitudinal gradient from the coastal plains to the second highest mountain. Interception losses ranged from 22% to 29% of total water input (rainfall and cloud interception) at all sites except the highest altitude site on Bellenden Ker, where interception was only 6% of the total water input.

The Australian rainforest interception losses span the complete range of values reported for other rainforests around the world. For example, low interception losses are associated with continental rainforests, such as in Amazonia, e.g. ∼9% (Lloyd et al., 1988). Higher interception losses have been reported for mountain rainforests in other humid tropical regions, e.g. 19–23% in Puerto Rico (Holwerda et al., 2006), 35% in Uganda (Hopkins, 1960) and 37% in Panama (Cavelier et al., 1997). The lower, continental rainforest interception losses can be explained using an appropriate value of the canopy storage (usually ∼1 mm) and wet canopy evaporation rates calculated using local weather data (e.g. see Roberts et al., 2005). At these sites the energy used in the interception process is less than the total available energy, the remainder being used to heat the atmosphere (Shuttleworth, 1989). The higher interception losses occur in rainforests in coastal or island locations, and are usually explained by the addition of advected energy from surrounding areas of dry vegetation or from nearby oceans. However, the amount of advected energy required is often several times the incident net radiation and the origin of this additional energy is still largely a matter of speculation (Schellekens et al., 1999, Murakami, 2006).

The current paper presents an analysis of interception at six rainforest sites in northern Queensland. We have used a physically based interception model to explore the different processes contributing to canopy evaporation losses to identify the key differences between coastal and mountain rainforests in Australia.

Section snippets

Site and measurements

A full description of the six rainforest sites in the Wet Tropics region of north Queensland, Australian is given by McJannet et al., 2007a, McJannet et al., 2007b, hence only a brief description is given here. The locations and altitudes of the sites, which are all World Heritage listed rainforests, are given in Fig. 1. Two of the sites, Hutchinson Creek (HC) and Oliver Creek (OC), are lowland coastal rainforests located in the Daintree National Park. They are at an altitude of 20 m and only 3 

The Gash rainfall interception model

We have used the revised Gash interception model (Gash et al., 1995), with canopy cover values in the range 94–97%, estimated using a fish eye lens camera (see McJannet et al., 2007a). Table 2 (reproduced from Wallace and McJannet, 2006) summarises the five terms in the Gash model, which assumes that rainfall is intercepted in a series of discrete storms, with sufficient time between each storm for the canopy to dry. Each storm can have up to three sequential phases; (i) a wetting phase during

Canopy storage capacity

Two examples of the relationship between (Tf + Sf) and Pga (at ML1 and UB) are shown in Fig. 2. The linear regressions through all the points gave Smean as 3.6 (±0.25) mm at Mount Lewis and 2.7 (±0.78) mm at Upper Barron. In contrast, the minimum estimates of the canopy storage capacity, Smin, for these two sites were much lower at 0.13 (±0.46) mm and 0.56 (±0.71), respectively. As Wallace and McJannet (2006) have already pointed out, the value of Smin is highly dependent on the points chosen to

Conclusions

This study has demonstrated that the most reliable methods for estimating canopy storage (S) is the ‘mean’ method described by Klaassen et al. (1998). The more conventional ‘minimum’ method consistently gave canopy storage values that were so low it was not possible to reproduce measured interception values. The optimisation method for estimating gave more plausible results, but there is evidence that the values derived in this way may have been uncertain due to the RMS minima being very

Acknowledgements

The authors would like to thank Mark Disher, Andrew Ford, Peter Richardson, Trevor Parker, Adam McKeown, Trudi Prideaux, Jenny Holmes and Pepper Brown for their help with field installation and maintenance. We are also grateful to the internal CSIRO reviewers, Peter Hairsine and Helen Cleugh, and the two anonymous external reviewers for their helpful comments on the manuscript and the thought provoking ideas on wet canopy evaporation rates. Funding for this research was provided by the

References (37)

  • J. Schellekens et al.

    Modelling rainfall interception by a lowland tropical rainforest in northeastern Puerto Rico

    Journal of Hydrology

    (1999)
  • D. Sharon

    The distribution of hydrologically effective rainfall incident on sloping ground

    Journal of Hydrology

    (1980)
  • M.K. van der Molen et al.

    Climate is affected more by maritime than by continental land use changes: A multiple scale analysis

    Global and Planetary Change

    (2006)
  • A.I.J.M. van Dijk et al.

    Modelling rainfall interception by vegetation of variable density using an adapted analytical model. Part 2. Model validation for a tropical upland mixed cropping system

    Journal of Hydrology

    (2001)
  • A.I.J.M. van Dijk et al.

    A two-parameter exponential rainfall depth-intensity distribution applied to runoff and erosion modelling

    Journal of Hydrology

    (2005)
  • J.S. Wallace et al.

    On interception modelling of a lowland coastal rainforest in northern Queensland, Australia

    Journal of Hydrology

    (2006)
  • Björk, L., Mareby, J., 1992. Epiphyte mass and water storage capacity of epiphyte vegetation in a mossy montane rain...
  • L.A. Bruijnzeel

    Predicting the hydrological impacts of land cover transformation in the humid tropics: the need for integrated research

  • Cited by (56)

    • Testing three rainfall interception models and different parameterization methods with data from an open Mediterranean pine forest

      2022, Agricultural and Forest Meteorology
      Citation Excerpt :

      The selection of the time separation interval is often subjective; and durations in the literature vary from 2 h to 12 h (Hassan et al., 2017; Klaassen et al., 1998). The selection of a short storm event separation interval will result in more storms of short duration, thus mean rainfall and evaporation rates will be overestimated (van Dijk et al., 2015; Wallace and McJannet, 2008). Even though an event separation interval is used for model parameter derivation, many studies subsequently compute interception losses on a daily basis (e.g., Hassan et al., 2017; Ghimire et al., 2017), following the assumption of Gash (1979) that the canopy is dry at the start of each daily rainfall event.

    View all citing articles on Scopus
    View full text