Satellite Technology

Using Satellite Technology

Satellites have been around for several decades and have proven to be an invaluable tool and source of information for studying the conditions and phenomena taking place at the earth surface. Examples are vegetation mapping and land use mapping, sea ice monitoring, weather forecasting etc.

Satellite data have many advantages. They provide repeated, full coverage of the earth surface, which is impossible with measurements on the ground. They are not only continuous, but also uniform and objective. There are no issues of inter-calibration and related corrections, as is usually the case with ground measurements.

Satellite sensors measure incoming radiation in a specific part of the electromagnetic spectrum, which is either reflected or emitted by the earth-atmosphere system. The spectral reflectance in the visual wavelength range may characterize the object at the surface. The emission in the thermal infrared is determined by the emissivity and temperature of the earth surface. Both signals are subject to atmospheric influences. Therefore the extraction of useful and valuable information has become an art as such, which however is very rewarding when successful.

Meteorological satellites have been launched for the purpose of large scale weather analysis and forecasting. However, already since the early 1980’s, other applications, focusing on the exchange of energy and water at the earth surface, have emerged. In this development EARS has played a major role. With its Energy and Water Balance Monitoring System, the company was the first of such climatic data from satellite.

This Meteosat 2nd Generation (MSG) Noon Composite image has been created from hourly images on 24-06-2010 and shows the surface reflection (albedo) in the visible part of the electromagnetic spectrum.

Energy and Water Balance Monitoring System

System Overview

The Energy and Water Balance Monitoring System includes two Meteosat data processing lines, one for rainfall and one for the energy balance components. Rainfall mapping is based on a statistical approach, using the empirical relation between the dwelling time of clouds and the precipitation amount. Monitoring of the energy balance components (radiation, sensible and latent heat) is based on the physics of energy and mass exchange at the earth surface.


The Meteosat thermal infrared and visual data are converted to planetary temperature and albedo by applying the appropriate calibration. The thermal channel calibration is based on simultaneously observed space and blackbody references and is available in the header or footer of the data files. For the visible channel a vicarious calibration approach is applied, based on known albedo’s of references: cumulonimbus, desert, forest, sea.

Rainfall processing

Different cloud levels are identified on the basis of their cloud top temperatures. Their presence is counted, leading to the cloud durations (CD) for different cloud height classes during a day or a ten day period. These cloud durations are then related to near real time rainfall observations, obtained through the WMO-GTS, by means of ´local´ multiple regression.

The regression coefficients (a) and the residual are determined for each rainfall station and its 11 nearest stations, and are then interpolated for each pixel. Hereafter the rainfall in each pixel is estimated by means of the regression equation, using the interpolated coefficients and residual.

R = a0 + a1 . CD1 + a 2 . CD2 + ai . CDi + … + residual

Energy Balance Processing

For energy balance mapping, noon and midnight data are most useful. Since in a single satellite image the time is changing from west to east, multiple hourly images are used to create visible and thermal infrared noon and midnight composites from the banana shaped segments that correspond to noon and midnight in each image. Appropriate temporal interpolation is applied to prevent possible rims between these segments. Subsequently, atmospheric correction procedures are applied, so as to obtain actual noon and midnight surface temperature (Tn, Tm) and noon albedo (Ao). The daily average surface temperature (To) is obtained by averaging the noon and midnight surface temperature. In addition the air temperature at the top of the atmospheric boundary layer (Ta) is derived by means of an analytic model of the daily temperature cycle.

The surface albedo, surface and air temperature are input to the calculation of the net radiation (In) and the sensible heat flux (H) at the earth surface:

Ln = (1 - Ao) Ig + ɛasTa4 - ɛosTo4

H = a (To - Ta)

The incoming solar ‘global’ radiation (Ig) is calculated from the solar constant, coordinates, time of the year, atmospheric optical depth and cloudiness. A correction for photosynthetic light use is applied. The calculation of the longwave radiative fluxes involves an assumption on the surface emissivity (ɛo ≈ 0.9) and the use of an empirical formulation of the emissivity of the atmosphere (ɛa). Hereafter the latent energy flux (LE), i.e. the heat used for evaporating water, is obtained as the difference between the net radiation and the sensible heat flux:

LE = In - H

The latent energy flux, expressed in W/m2 is finally converted to the actual daily evapotranspiration expressed in mm/day (1 mm/day corresponds roughly to 29 W/m2). The potential evapotranspiration, which occurs when the surface is wet, is determined by the available energy:

LEp ≈ 0.8 ∙ In

The relative evapotranspiration (RE) is finally obtained as:


The relative evapotranspiration is most important information in relation to agricultural drought monitoring and crop yield forecasting. Relative evapotranspiration is known to be a good measure of plant available soil water and is proportional to plant growth (Doorenbos and Kassam, 1979, “Yield Response to Water”, FAO Irrigation and Drainage Paper 33).

Dealing with Clouds

When a pixel is cloudy, the radiation calculations are different. The transmissivity (tc) of the cloud is derived from the cloud albedo. The long wave radiation fluxes approximately cancel. The net radiation at the earth surface is then determined with:

Inc = (1 - Ao) tc Ig

The last measured cloud free albedo is used. The final step is to estimate the distribution of the net radiation between sensible and latent heat. To this end we use the relative evapotranspiration (similar to the Bowen ratio, which tends to be conservative) of the last cloud free day and estimate the actual evapotranspiration with LE = 0.8 ∙ RE ∙ In. However if cold clouds (Cumulonimbus) are observed over the area, the relative evapotranspiration is corrected upward on the basis of the cold cloud duration (CCD):

LE = 0.8 ∙ RE ∙ In + CCD ∙ LEP

The energy balance model calculations are schematically represented in the diagram below.

Drought Products

Relative Evapotranspiration

The relative evapotranspiration is the daily actual over potential evapotranspiration, expressed in percent. It characterizes the crop water availability and crop growth level.

Evapotranspiration Drought Index

The Evapotranspiration Drought Index (EDI) is defined as the actual evapotranspiration (LE) over the potential evapotranspiration (LEP) on a timescale of one or several months. It is comparable to the relative evapotranspiration product (RE) but then applied to a longer period, and suitable to characterize the crop growth conditions during a significant part of, or the entire growing season:

EDI = LE / LEP      [1-3 monthly]]

Difference Evapotranspiration

The difference evapotranspiration (DE) product compares the relative evapotranspiration (RE) in the current period with the relative evapotranspiration for the same period in one or more reference years (RE*):

DE = (RE - RE*) / RE*       [%]

The reference period is usually the previous year or the average of the previous 5 years. DE is expressed in percent decrease or increase relative to the reference period. When applied to a period of the growing season, DE is used as an indicator of crop production relative to other years.

The image on the right shows an example for West Africa. Note that difference maps are usually best interpreted in conjunction with the product that it is based on, in this case the monthly relative evapotranspiration, also called Evapotranspiration Drought Index (EDI).

Precipitation Drought Index

The Precipitation Drought Index (PDI) is defined as the precipitation (PRC, in mm) over the potential evapotranspiration (EP, in mm) expressed as a percentage on a timescale of one or several months:

PDI = PRC / EP       [x-monthly]

A PDI lower than 100% implies that precipitation was lower than potential evapotranspiration. Consequently, precipitation could not supply enough water for optimum plant growth. A PDI higher than 100% however, does not necessarily mean that optimum plant growing conditions are met, whereas an unknown part of the rainfall is lost to runoff and deep percolation.

Crop Yield Forecasting

Crop yield estimation and forecasting

The ECGM is a crop growth model that utilizes the EWBMS radiation and evapotranspiration outputs to generate biomass estimates. It is a relatively simple model that considers the plant as a single organ, generating and storing carbohydrates (‘dry matter’). It assumes the economic yield to be proportional to the dry matter production. The simulation of biomass production is done on a daily basis and involves the calculation of photosynthetic active radiation, photosynthetic light use efficiency, light interception, relative growth level due to water limitation, gross photosynthesis, maintenance respiration and finally net biomass production and accumulation. An example of such simulation is shown in the graph below. However it is difficult to estimate the actual biomass accurately in this way, as there are many additional factors (seed quality, soil type, fertilizer, crop management) that play a role. Therefore a differential approach is usually followed, where the deviation of the expected yield relative to a (series of) reference years or “difference yield” is presented.

For practical applications, it is sufficient to estimate the relative crop yield (RY) directly from the relative evapotranspiration on the basis of a relation that was developed by Stewart (1973) and Doorenbos and Kassam (1979), and which states that the relative yield deficit is proportional to the relative evapotranspiration deficit:

(1 - RY) = k (1 - RE)

This relation results from the fact that both CO2 uptake and transpiration of the plants are regulated by the plant stomata and are therefore proportional. Here the relative evapotranspiration applies to the entire growing season, usually a period of 3 months. The factor k is related to the drought sensitivity of a crop. For example for maize k = 1.25 (drought sensitive) and for sorghum k = 0.9 (drought resistant). For an intermediate crop (e.g. wheat) k = 1. In that case the relative yield is equal to the relative evapotranspiration (RY = RE). Finally the difference yield (DY) follows from:

DY = (RY - RY*) / RY*

Where RY* is the reference relative yield. Difference yields are usually presented relative to the previous year and/or relative to the previous 5 year average.