Improved LAI estimation for weather forecasts

Short Description

Vegetation plays an important role in the field of weather forecasting, as their spatially and temporally varying characteristics significantly influence energy and water balance of the soil, as well as the near-surface atmosphere, respectively boundary layer.

Currently, this important aspect is only inadequately taken into consideration by weather models. Reasons for this are limited computational capacity as well as lacking availability of reliable monitoring data of vegetation and land cover. Therefore, it is still common practice to use climatological LAI data and coarsely resolved global land cover data. The use of LAI climatologies might be sufficient for global models due to their coarse grid resolution. However, such a simplification becomes insufficient when dealing with convection resolving models with low km grid distances, leading to sub-optimal forecasts of small-scale phenomena, especially in complex terrain.

Within the proposed project, an innovative approach is used to counterbalance this deficiency and to quantify achievable improvements. In particular, Sentinel-2 satellite data provided by Copernicus will be analysed by means of physically based radiative transfer models to obtain spatially and temporally continuous near-real-time LAI time series, using all ten spectral bands of the satellite.

Subsequently, the obtained LAI data will be integrated into a (simplified) Extended Kalman Filter assimilation process within a high-resolution physical soil model. To maximise the benefit of the high resolution in combination with available land cover data that are mainly based on Sentinel-2 data (CORINE, LISA), the soil model is supposed to run decoupled from the atmospheric model with very high resolution (< 1 km) to provide best possible analyses of soil parameters and state of vegetation. This approach is new – also from an international perspective.

Apart from the fact that analyses of soil properties themselves provide valuable information, through an innovative coupling with the convection-resolving atmospheric model AROME additional benefit through an improvement of the weather forecasts can be expected. To assess the (expected positive) effect of LAI assimilation on predicting near-surface parameters in weather models, meteorological measurements are used for the quantification. So far, such kind of analysis has not yet been carried out, neither in Austria nor in an international environment.

The proposed project pursues several goals. From an observational point of view, the focus lays on improving the quality of LAI values, which are based on Sentinel-2 data. Radiative transfer models are used for the derivation of LAIs at high spatial and temporal resolution. Compared to empirical methods, these models have the advantage of being robust and furthermore, they do not depend on local in-situ data for calibration purposes. This guarantees widespread global applicability.

From a modelling point of view, the optimized description of vegetation and the related exchange processes between soil and boundary layer is paramount. Data assimilation, as proposed within LAETITIA, is used to correct known weaknesses in model prediction of the LAI. At the end of the project, optimized soil analyses and weather forecasts will be available, whereas the quantification of the optimization will promote reliable statements regarding a subsequent operational implementation.

Project Partners


Central Institute for Meteorology and Geodynamics (ZAMG)

Contact Address

Central Institute for Meteorology and Geodynamics (ZAMG)
Hohe Warte 38
A-1190 Vienna