The pilot phase of the Join Air-Sea Monsoon Interaction Experiment (JASMINE) was completed in 1999. One of its objectives was to study the evolution of tropical cloud clusters (TCC) over the North Indian Ocean. Long-term evolution (1 week to 10 days) of TCCs have been studied in the past by employing mesoscale models initialized with the operational European Center for Medium-Range Weather Forecasts (ECMWF) analysis. The objective of the upper air measurement program of the Indian Ocean Experiment (INDOEX) was to diagnostically estimate the surface energy budget at the ocean surface using the ECWMF analysis. In the past, several diagnostic studies have been based on the ECWMF operational and reanalysis datasets. However, given that the 40-year ECMWF reanalysis data will be available for the period between 1957-1996, there is a need to evaluate the quality of the operational ECMWF analysis during the 1999 field phase of the two experiments.

Vertical profiles of wind speed, wind direction, mixing ratio, and temperature were measured with radiosondes during INDOEX (between January to March 1999) and JASMINE (between April to May 1999). The same variables from the ECMWF operational analysis are horizontally interpolated to the location of the radiosonde stations (ship-based). The quality of the analysis was evaluated in terms of bias (observation minus analysis), and root mean square (RMS) of the bias.

The ECMWF humidity analysis is performed only upto the 250 hPa and therefore the the bias and RMS profiles between 1000 - 250 hPa will be reported. The profiles of temperature bias shows that the analysis exhibits a cold (warm) bias at low-levels (high levels, 800-200 hPa). The maximum cold (warm) bias is 0.5 C (1 C). The RMS temperature bias is about 1 C and does not vary significantly with height. The wind speed bias profile shows that the analysis wind agrees to the observed within 1 m/s between 1000 - 250 hPa. The RMS wind bias lies between 1 and 3 m/s, and increases with height above 800 hPa. The mixing ratio bias profile suggests that the analysis is too moist at the low-levels (900 - 650 hPa) and too dry at the midlevels (650 - 300 hPa). The RMS mixing ratio profiles decrease from a maximum of 1.7 g/kg to near zero at 250 hPa.

Our preliminary results suggest that the analyzed winds at low- and high-level are underestimated. Excess humidity is evident at low levels and is accompanied by colder air, while humidity deficit in the mid and upper troposphere is accompanied by warmer air. The analysis suffers from deficiencies that are similar to those found over the western Pacific ocean (during the Coupled Ocean Response Experiment 1992-93) despite improvements effected in the ECMWF model since then. Currently the study is being extended to include data between January-April 1999, the results of which will be reported at the conference.