GISMO (Gap Interpolation and quality Screening of Meteorological Observations) has been developed to preprocess meteorological observations prior to an application as input for land surface models. This preprocessing of meteorological data includes data quality checking as well as gap closure in data time series.

GISMO analyzes recordings of the 5 meteorological parameters (precipitation, temperature, humidity, wind speed and global radiation) that are commonly used as input for the forcing of land surface models (e.g. snow models, hydrological models). Three types of screening rules have been implemented for quality checking of hourly station recordings:

  1. high/low range limits
  2. rate-of-change limits
  3. continuous no-observed-change in meteorology

Quality screening rules in GISMO.

Missing values in the data series that might either result from an elemination of identified measurement errors or that have already existed in the time series before the quality checking process are interpolated by an application of different interpolation algorithms, depending on the length of the gap considered.

Gap interpolation in GISMO:
1 h 2 – 24 h > 24 h

value (h) =
[value (h-1) + value (h+1)] * 0.5

value (h) =
[value (h-24) + value (h+24)] * 0.5

value (h) =

To evaluate the performance of GISMO for a given station location, the evaluation tool EvaluateGISMO has been developed that allows easy access to different performance criteria. The program creates a user defined number of gaps in a continuous time series of meteorological observation. Beginning at randomly created starting points, gap length is continuously increased in order to show the performance for different gap lengths.

The evaluation tool EvaluateGISMO.

As the line plots created by EvaluateGISMO indicate, GISMO performs well in the interpolation of short gap length (length < 3 days) for the parameters temperature, humidity and global radiation. Wind speed can not be interpolated with good accurancy.

Average performance of GISMO for station Kühroint in the National Park Berchtesgaden (Germany).

A part from itʼs utility in the processing of past meteorological conditions, GISMO can be combined with a stochastic climate generator. In this setup, the temporal interpolation and the associated closure of gaps in the meteorological time series strongly enhances the meteorological data available for a statistical reorganisation (e.g. on a weekly time basis) within the climate generator.