HR: 1340h
AN: A53B-0898    [Abstracts]
TI: Complementary Dynamical and Statistical Downscaling from a GCM: Maha rainfall over Sri Lanka
AU: * Zubair, L
EM: lareef@iri.columbia.edu
AF: International Research Institute for Climate Prediction, The Earth Institute at Columbia University, POB 1000, Palisades, NY 10964 United States
AU: Qian, J
EM: jqian@iri.columbia.edu
AF: International Research Institute for Climate Prediction, The Earth Institute at Columbia University, POB 1000, Palisades, NY 10964 United States
AU: Ward, N
EM: nward@iri.columbia.edu
AF: International Research Institute for Climate Prediction, The Earth Institute at Columbia University, POB 1000, Palisades, NY 10964 United States
AU: Ndiaye, O
EM: ousmane@iri.columbia.edu
AF: International Research Institute for Climate Prediction, The Earth Institute at Columbia University, POB 1000, Palisades, NY 10964 United States
AU: Perera, R
EM: ruvini@iri.columbia.edu
AF: International Research Institute for Climate Prediction, The Earth Institute at Columbia University, POB 1000, Palisades, NY 10964 United States
AU: Chandimala, J
EM: janaki@iri.columbia.edu
AF: International Research Institute for Climate Prediction, The Earth Institute at Columbia University, POB 1000, Palisades, NY 10964 United States
AU: Chandimala, J
EM: janaki@iri.columbia.edu
AF: Natural Resources Management Services, Environment and Forest Conservation Division, Mahaweli Authority of Sri Lanka, Dam Site, Polgolla, CP 20250 Sri Lanka
AU: Ralapanawe, V
EM: vidhura@yahoo.com
AF: Foundational for Environment, Climate and Technology,, 76/2, Matale Road, , Akurana, CP 20850 Sri Lanka
AU: Blumenthal, B
EM: benno@iri.columbia.edu
AF: International Research Institute for Climate Prediction, The Earth Institute at Columbia University, POB 1000, Palisades, NY 10964 United States
AB: There are two approaches to downscaling the results of the coarse scale of Global Climate Model (GCM) to fine-scales that are needed for applications. One may use a regional climate model that captures the fine scale details within a limited domain (dynamical downscaling) or use statistical relationships between GCM outputs and historical observations (statistical downscaling). Both approaches have relative strengths. Dynamical downscaling captures the physics explicitly and it provides complete solutions for the evolution of the atmosphere. It does not need extensive historical records. However, it is computationally expensive at fine scales. Statistical downscaling is simpler, inexpensive and can provide results that may be more skillful but it may be vulnerable to sampling error. We may also introduce details into the forecasts that have no physical basis. Sri Lanka is 224 km wide and 450 km long and has a mountain range with a narrow peak of over 2 km. The topography changes drastically over small distances and high-resolution downscaling is needed. Here, we attempt to downscale to a grid of 10-20 km. Sri Lanka receives 45-70% of its rainfall between October and December at the start of the main Maha cultivation season. The El Nino / Southern Oscillation and Indian Ocean Dipole phenomena modulate the Maha rainfall. These large-scale physical mechanisms are likely to be captured by GCM predictions leading to skillful predictions over Sri Lanka. The dynamical downscaling from the ECHAM4.5 GCM using the RegCM3 regional climate model was undertaken for Sri Lanka by Joshua Qian. The RegCM3 simulations yielded reasonable spatial distribution of precipitation at a resolution of 20-km. The GCM low-level wind fields from ECHAM5.4 GCM were used as a predictor to build statistical relations with observations. This downscaling work led to skillful predictions for Eastern Sri Lanka. These predictions are consistent with the reported mechanisms; the anomalous zonal wind brings preferential skill to the eastern windward side. The use of dynamical downscaling approach provided fine scale results that improve considerably upoun the GCM output. The use of a fine scale of 20 km was essential to obtain reasonable results. Statistical downscaling provided skillful predictions of seasonal rainfall for Eastern Sri Lanka. This skill was obtained at a finer scale than for the dynamical approach. The use of both techniques is complementary in that the dynamical downscaling provides physical insight that can be used to investigate statistical relationships between observations and GCM fields.
UR: http://iri.columbia.edu/~mahaweli/
DE: 9320 Asia
DE: 9340 Indian Ocean
DE: 3329 Mesoscale meteorology
DE: 3374 Tropical meteorology
DE: 1620 Climate dynamics (3309)
SC: Atmospheric Sciences [A]
MN: 2004 AGU Fall Meeting