3 February 1987 The humidity pre-processing was modified.
10 February 1987 SATEM precipitable water content was included to analysis.
7 April 1987 Forecast model cycle 29. The surface and subsurface parametrisation scheme has been revised. Each grid box is now divided into vegetated and bare ground parts which concerns the evaporation over land surfaces. The time evolution of the soil water content takes root uptake, interruption of precipitation and collusion of dew by a skin reservoir, surface run off due to sloping terrain and gravitational drainage into account.
The use of specific thermal properties of snow modifies the surface temperature evolution over snow covered ground.
The convective Kuo scheme was modified. The accumulated convective precipitation now includes convective snowfall. Over sea surface convective precipitation is allowed to fall as shown when the sea surface temperature is above 0°C. For both land and sea points the air temperature at the first model level is required to be colder than minus 3°C for snowfall.
The post-processing method to compute the 10m winds, 2m temperature and dew point has been reformulated. The calculations of 10m wind components and the 2m temperature base on realistic profiles of wind speed and temperature gradients within the atmospheric boundary layer which is assumed to be a Constant flux Layer (CFL). The variables (at any height) are obtained by integrating their vertical derivatives.
The 2m dewpoint depression is computed by assuming that the relative humidity is constant in the CFL.
The modifications of the near surface temperature give a more realistic simulation of the diurnal temperature variation under clear sky conditions. The old 2m dew point calculation suffered from a surface layer which was too moist which results in a too narrow dew point spread. The new scheme corrects this deficiency to a large extent.
In stable conditions the new post-processing give lower wind speeds u and v at 10m height. The reduction which is of the order 1-3 ms-1 is in better agreement with locally observed winds.
Over sea the Charnock constant of 0.032 was replaced by the lower value of 0.018.
Soil moisture analysis is not being done any more. The initial soil temperature and moisture content are taken from the first-guess.
13 April 1987 An error in the computation of 10m wind and 2m T and Td was fixed. The data of u10, v10, 2m T and 2m Td are incorrect within the time period from 7-13 April 1987
16 June 1987 Land wind data in the Tropics were used and the wind direction check was tightened up.
7 July 1987 An error in post-processing of low cloud and total cloud amount was fixed. The total cloud cover is incorrect from 15 July 1986 to 6 July 1987 as a result.
21 July 1987 A number of changes relating mainly to the use of SATEMs was implemented. Now 7 SATEM layers are used in the vertical instead of 11, i.e. 1000/700 hPa, 700/500 hPa, 500/300 hPa,300/100 hPa, 100/50 hPa, 50/30 hPa, 30/10 hPa.
The modifications allow better use to be made of satellite sounding data in agreement with the vertical resolution given by the satellite instruments.
The satellite observation statistics and quality control were revised.
11 August 1987 A problem with the stratospheric SATEMs during early August caused the analysis to develop an erroneous warm dome in the 50-30 hPa thickness field over the Antarctic which was fixed on11 August.
27 October 1987 Observations at North Pole were included in data selection. The humidity analysis data selection criteria were made consistent with mass and wind analysis.
8 December 1987 The first-guess rejection limit for winds was tightened and an asymmetric first-guess check on extra-tropical cloud track winds was introduced.