The impact of land cover change on storms in the Sydney Basin, Australia

Abstract

This study has used a numerical model (RAMS) at 1 km horizontal grid intervals over the Sydney Basin to assess the impact of land cover change on storms. Multiple storms using the National Center for Environmental Prediction (NCEP) reanalysis data were simulated with pre-European settlement land cover then re-simulated with land cover representing Sydney's current land use pattern. While all simulated storms did not respond to the change in land cover consistently, storms of similar types responded in comparable ways. All simulated synoptically forced storms (e.g. those triggered by cold fronts) were unresponsive to a changed land surface, while local convective storms were highly sensitive to the triggering mechanism associated with land surface influences. Storms travelling over the smoother agricultural land in the south-west of the Sydney Basin experienced an increase in velocity, and in a special case, the dense urban surface of Sydney's city core appears to trigger an intense convective storm. It is shown that the dynamical setting predominantly triggers storm outbreaks. This is seen most clearly in the isolated convective storm category where the sea breeze front often dictates the location of storm cell initiation.

Introduction

The land surface affects the lower atmosphere via the surface energy budget and the surface water budget (Verstraete and Dickinson, 1986). Several recent review papers have addressed the issue of the role of the land surface in weather and climate (Avissar and Verstraete, 1990, Betts et al., 1996, Pielke et al., 1998, Pitman, 2003, Kabat et al., 2004). Changes in the albedo, the surface roughness, leaf area index, root depth and a range of other biophysical characteristics affect the surface energy balance both via changing net radiation and via the partitioning of net radiation between sensible and latent heat (see Sellers, 1992, Betts et al., 1996). A change in the partitioning of net radiation can affect boundary layer depth (Pielke et al., 1998) and thereby clouds and incoming solar radiation (Sellers, 1992). The surface also affects the exchange of momentum and the fluxes of carbon and other trace gases.

There is substantial recent literature focusing on how the land surface affects weather and climate (see Pielke et al., 1998 and references therein; Kabat et al., 2004 and references therein). This literature includes modeling evidence from global climate models, regional climate models and observational campaigns. Almost all this literature focuses on the modification of natural landscapes to grassland or agriculture which is reasonable given that humans have altered close to 50% of continental surfaces (Vitousek et al., 1997) via reforestation, deforestation, overgrazing or agriculture. For example, Copeland et al. (1996), Pielke et al. (1999) and Marshall et al. (2004) found simulated changes in precipitation resulted from altered surface vegetation at a regional scale. In addition to the biophysical characteristics of the surface, the specific distribution of land cover types (i.e. landscape heterogeneity) can also influence weather patterns at scales of tens of kilometers (Chen and Avissar, 1994, Lynn et al., 1995, Avissar and Liu, 1996, Shao et al., 2001). Thus, accurate representation of land surface properties and landscape patterns in regional climate studies across a range of scales is important.

While the impact of land cover modification from natural vegetation to crops or grasslands has been shown to affect local weather, the impact of urbanization on the atmosphere has been less well studied by the regional climate modeling community. If land cover change (LCC) can affect the partitioning of net radiation, the depth of the boundary layer and the momentum flux, then urbanization has the potential to affect weather patterns over urban and surrounding areas. There is extensive evidence that urban areas can affect climate (Arnfield, 2003) as cities are capable of altering natural weather patterns through the urban heat island effect, the disruption of air flow, initiation of mesoscale circulations and through the discernable influence on storm occurrence. Pielke (2002) gives a thorough summary of both modeling and observational studies concerned with urban impacts since their inception in the 1970s and 1980s. An important early study was conducted by Hjelmfelt (1980), simulating the response of wind flow over the urban area of St Louis. A recent study by Kalnay and Cai (2003) investigated the impact of urbanization on climate, highlighting the significance of cities in terms of regional warming and the reduction in diurnal temperature fluctuations. The precise impact of urbanization on weather and climate warrants further research as cities continue to draw higher populations and expand in their spatial reach exerting a potentially greater influence on atmospheric processes.

Observational evidence dating back over 30 yrs also suggests an urban influence on weather and climate, with some studies concluding increases in warm season rainfall result downwind of cities (e.g. Changnon, 1968, Landsberg, 1970, Huff, 1986). There is also evidence for cities causing decreased precipitation by altering cloud microphysics (Rosenfield, 1999, Ramanathan et al., 2001). An observational study by Shepherd and Burian (2003) demonstrated the significance of the sea breeze and coastline curvature in addition to the urban surface for the meteorology of coastal cities, highlighting the complexities in urban-atmosphere studies. Thus, specific features of cities are important in determining meteorological outcomes, and generalized assessments of urban influences on weather and climate must be reviewed critically (Shepherd and Burian, 2003). Other specific storm studies by Balling and Brazel (1987), Jauregui and Romales (1996), Bornstein and Lin (2000) and Baik et al. (2001) have shown urban areas to directly influence storm initiation, intensity and motion.

In Australia, LCC has been extensive since European settlement in 1788. Initially, this LCC was deforestation for a variety of types of agriculture. More recently, urban expansion in the Sydney region (mid-coast of New South Wales) has been extensive to accommodate an influx of 50,000 people per year (EPA, 2003). The Sydney Basin is a relatively flat area, bounded to the north, west and south by areas of high relief to create a basin. This study explores whether the changes in land cover over the Sydney Basin affect the characteristics of storms occurring in the region (see Section 3 for a description of storm types). The Sydney Basin is frequently affected by such events, predominantly in summer during the early afternoon and evening (Potts et al., 2000). Case studies of several of Sydney's severe storms can be found in Spillane and Dixon (1969) and Bureau of Meteorology, 1993, Bureau of Meteorology, 1995, while Matthews and Geerts (1995) provide a study of the spatial distribution of Sydney's storms according to synoptic type. The possible interaction of these storms with Sydney's urban surface is a growing concern due to the associated financial implications of storms. For example, Sydney's hailstorm of April 1999 caused the most insured damage of any natural disaster in Australia's history with insurance costs exceeding AU$1.7 bn (Insurance Disaster Response Organisation, 2002). Whether the urban surface played any part in the intensity of this storm is worthy of further study.

This study tests the hypothesis that urbanization over the Sydney Basin affects the nature of storms. The aim is to determine whether the LCC over the Sydney Basin has led to an intensification of storms, a change in their preferred paths or velocities, or the time at which they occur. To explore the land surface influence on storms in the Sydney Basin, multiple simulations using a high resolution numerical model were performed (described in Section 2). Section 3 displays the results obtained using the different land cover regimes. Discussion of results is found in Section 4, followed by conclusions in Section 5.