ESA Earth Home Missions Data Products Resources Applications
EO Data Access
How to Apply
How to Access
Site Map
Frequently asked questions
Terms of use
Contact us



MERIS AOD and PM10 in-situ measurements: data fusion in an operational air quality forecast model

Stijn Janssen(1), Peter Viaene(1), Frans Fierens(2), Gerwin Dumont(2) and Clemens Mensink(1)

(1) VITO, Boeretang 200, 2400 Mol, Belgium
(2) IRCEL, Kunstlaan 10-11, 1210 Brussel, Belgium


The availability of a reliable air pollution forecast model is of crucial importance to environment agencies in their assignment to inform the general public in case of expected air pollution episode and to trigger air quality action plans. At IRCEL, the Belgian Interregional Environment Agency, the PM10 forecast model OVL is operational and produces forecasts up to four days in advance on a daily basis. Besides meteorological input these forecasts are driven by the most recent PM10 in-situ measurements available from the regional telemetric networks. OVL is a statistical forecast model relying on a neural network approach. The computational cost of OVL is extremely low and the accuracy of the forecasts can compete to those of deterministic air quality models. The most important draw back of the statistical approach is that the forecast values are only representative for the point locations of the monitoring stations of the telemetric network. After all, the neural networks are trained on the historical time series collected at the monitoring sites. To tackle this shortcoming a dedicated spatial interpolation model RIO is developed alongside, suitable for the production of PM10 air quality forecast maps.

RIO can be classified as a detrended Kriging interpolation model. The core of the interpolation model is based on the Ordinary Kriging interpolation schema. However, before the standard Kriging methodology is applied, all PM10 input values are detrended to remove the local character of the sampling sites. After the detrending, the PM10 values are transformed into location-independent artificial quantities. It is shown that this detrending procedure is crucial to obtain spatial homogeneity of the set input values. This is an essential requirement of the Kriging methodology which is often disregarded in air pollution mapping tools. After the interpolation step, a local bias is introduced in a re-trending procedure to arrive at realistic PM10 air pollution maps with local patterns at places where no monitoring data is available.

For the detrending of the PM10 values Aerosol Optical Depth (AOD) measurements of the MERIS sensor on board of the Envisat satellite are used. Daily AOD observations over Belgium are collected for the 2002 – 2006 period. Since AOD is a columnar value (# particles/m²), each daily measurement is modified with the appropriate Boundary Layer Height (BLH) to convert it to a concentration quantity (# particles/m³). In a subsequent step BLH modified AOD values are aggregated over space and time to increase the overall statistics. Spatial aggregation ranges between 1.2 km and 10 km whereas the time aggregation is implemented as a climatology for a set of predefined periods (a season, a month,…). Finally aggregated AOD result are compared to corresponding aggregations of PM10 in-situ measurements and the optimal correlation is used to set up a relation between PM10 and (modified) AOD that is usable in the detrending step of the RIO interpolation tool.

In this paper different AOD aggregation levels both in space and time are presented and their correlation with corresponding PM10 in-situ values is discussed. The application of the data fusion process between in-situ and earth observation aerosol measurements is demonstrated in the operational PM10 forecast model OVL-RIO. Besides this, the paper illustrates other opportunities for PM10 mapping (historical records, annual averages…) making use of satellite retrieved AOD values.


Workshop presentation

Keywords: ESA European Space Agency - Agence spatiale europeenne, observation de la terre, earth observation, satellite remote sensing, teledetection, geophysique, altimetrie, radar, chimique atmospherique, geophysics, altimetry, radar, atmospheric chemistry