Kr. Sperber et al., SIMULATION OF THE NORTHERN SUMMER MONSOON IN THE ECMWF MODEL - SENSITIVITY TO HORIZONTAL RESOLUTION, Monthly weather review, 122(11), 1994, pp. 2461-2481
The ability of the ECMWF model (cycle 33) to simulate the Indian and E
ast Asian summer monsoons is evaluated at four different horizontal re
solutions: T21, T42, T63, and T106. Generally, with respect to the lar
ge-scale features of the circulation, the largest differences among th
e simulations occur at T42 relative to T21. However, on regional scale
s, important differences among the high-frequency temporal variability
serve as a further critical test of the model's ability to simulate t
he monsoon. T106 best captures both the spatial and temporal character
istics of the Indian and East Asian monsoons, whereas T42 fails to cor
rectly simulate the sequence and development of synoptic-scale milesto
nes that characterize the monsoon flow. In particular, T106 is superio
r at simulating the development and migration of the monsoon trough ov
er the Bay of Bengal. In the T42 simulation, the development of the mo
nsoon occurs one month earlier than typically observed. At this time t
he trough is incorrectly located adjacent to the east coast of India,
which results in an underestimate of precipitation over the Burma-Thai
land region. This early establishment of the monsoon trough affects th
e evolution of the East Asian monsoon and yields excessive preseason r
ainfall over the Mei-yu region. EOF analysis of precipitation over Chi
na indicates that T106 best simulates the Mei-yu mode of variability,
which is associated with an oscillation of the rainband that gives ris
e to periods of enhanced rainfall over the Yangtze River valley. The c
oarse resolution of T21 precludes simulation of the aforementioned reg
ional-scale monsoon flows.