37 Radar and Satellite Meteorology

Prabir Kumar Das

1. Learning outcomes
2. Introduction
3. Types of meteorological satellites
4. Applications of satellites in deriving different meteorological
parameters
4.1. Rainfall
4.1.1. Visible and infrared radiometry (VIRS) technique
4.1.2. Passive microwave technique
4.1.3. Active microwave technique
4.2. Air temperature
4.3. Solar radiation
4.3.1. Regression approach
4.3.2. Physical approach
4.4. Evapo-transpiration
5. Radar meteorology
6. Summary

1. Learning outcomes

After studying this module, you shall be able to know about:

• The types of meteorological satellites
• The basic principles and methodology for detection/estimation of different meteorological parameters like cloud, rainfall, temperature, solar radiation, evapo-transpiration etc.
• The basics of radar system and its applications

2. Introduction

Forecasting of weather and climate or regional crop yield monitoring requires multi temporal input data on a number of parameters and over a large area. The sparse ground based observation network for meteorological and other parameters and the logistics difficulties in maintaining a large number of ground stations, mostly in the developing countries are the major issue for availability of near-real time and reliable spatial information. With the advent of satellite technology, frequent availability of reliable information on many meteorological parameters such as cloud, solar radiation, surface air temperature, rainfall, soil moisture and evapo-transpiration have become possible using both meteorological and earth observation satellites. The meteorological parameters are generally retrieved from satellite by establishing algorithm or empirical relationship between the estimated radiance of satellite and the ground based physical state variables. The optical, infrared, thermal and/or microwave sensors of the satellite can be utilized for deriving the required meteorological observations.

3. Types of meteorological satellites

The meteorological satellites can be broadly classified into two types, i.e. (i) polar orbiting, and (ii) geostationary (Fig.1). The polar orbiting satellites move around the earth at a height of about 750-900 km, by passing approximately the poles during its each orbit. As the satellite moves around the earth, simultaneously the earth is also rotating beneath the satellite, as a result the entire earth surface can be observed by polar orbiting satellites. The satellites generally follow orbits nearly fixed in space and the areas scanned on each pass (swath) are nearly adjacent at the equator with overlapping areas further pole-ward. By completing 14 cycles per day, this type of satellite can provide observations for  entire globe twice in a day. Some of the major polar orbiting meteorological satellites used globally are NOAA, IRS, ERS, TRMM, DMSP, Oceansat-1 etc.

The geostationary satellites move around the earth over the equator at a height of about 36000 km. The height of the satellite is estimated to fix the angular velocity of the satellite such that it can complete one orbit around the earth synchronized with the earth’s rotation around its own axis. Hence, the satellites remain virtually stationary over the earth over a given location on the equator and are capable to provide images in every 30 minutes. Instead of its high temporal resolution, the main limitation of the geostationary satellite is its spatial resolution as compared to polar orbiting satellite because of its high distance from earth surface. Some of the geostationary satellites which have been used extensively for agro-meteorology studies are GMS, GOES-W, GOES-E, INSAT Series, GEOS, METEOSAT series etc. The differences between geostationary and polar orbiting satellites are given in table 1.

1. Rainfall

1.1. Visible and infrared radiometry (VIRS) technique

Visible (VIS) and infrared (IR) techniques are simple to interpret but generally show a relatively lower accuracy. There are several methods for estimating the rainfall using visible and infrared imageries,(a) some of them are: cloud-indexing,(b) thresholding, life history and cloud model. Each method has its unique interpretation towards sensing the physical properties of clouds.

In ‘cloud indexing’ technique, the cloud types are identified from the satellite images and the corresponding rain rate is being applied to the identified cloud type. Different methods have been developed to calibrate the indices for rainfall estimation; (a) The Earthsat method estimates the 6-hour rainfall using the cloud temperature and empirical information for major crop growing regions, (b) The manual Bristol method is a family of algorithms related to cloud indexing, developed at the University of Bristol. Though the algorithms had been originally developed for polar orbiting NOAA satellites, those were successfully adopted for geostationary satellites also. This method has been modified into an interactive method, known as Bristol/NOAA Interactive System (BIAS). In BIAS, the 6-12 or 24 hour rainfall (R) is a function of several cloud indices, like rain cloud type (Ct), fraction of rain cloud-  type area (Ca), cloud duration (Cd), as well as the pixel characteristics in terms of climate category (Cc) or surface measurements (Sw).

Thresholding method is based on the relationship of cold-bright clouds with associated high probable rainfall, like in cumulonimbus cloud. It is a very simple approach, but it may not be always true. The probability of rainfall is lower for ‘cold but dull’ thin cirrus and ‘bright but warm’ stratus clouds. This approach has been implemented in TAMSAT (Tropical Applications of Meteorology using SATellite) and PERMIT (Polar Orbiter Effective Rainfall Monitoring Interactive Technique) methods.

In TAMSAT, the ‘Cold Cloud Duration’ is used as surrogate for rainfall estimation, where the fractional coverage of cloud with temperature below 235 K is related to tropical rainfall. This approach led to development of GPI (GOES precipitation index) which is widely accepted and used by the meteorologist.

Where, GPI is the mean rain rate estimate, ‘f’ is the fraction of area covered with cold clouds (Temperature < 235K) and ‘∆t’ is the time interval between two successive images (in hours). In GPI, the effect of land surface characteristics was not considered, as a result though GPI could provide good rainfall estimates over ocean, the estimates over land tend to underestimate.

The PERMIT method uses the infrared thresholds for detecting the rain-cloud areas. In this approach the rainy days are identified instead of cold cloud duration.

Life-history method generally provides rainfall estimates for any kind of convective clouds by considering the growth and disappearance of individual cloud over time. This approach is totally based upon the temporal geo-stationary satellite imageries, where the time-series imageries are analyzed to study the clouds’ life cycle. The most adopted and acknowledged life-history based rainfall estimation approach is Griffith-Woodley technique (Griffith et al., 1978).

In ‘cloud model’ technique, the cloud physics is introduced for better understanding the rain forming process and for quantitative improvement in the retrieval process as such. A cumulus convection parameterization was used to estimate the rain rate from fractional cloud cover. A one-dimensional cloud model relates cloud top temperature to rain rate and rain area in the Convective Stratiform Technique (CST).

4.1.2.   Passive microwave technique

In visible and infrared imageries, clouds appear as opaque and the rainfall is estimated using cloud top temperature and structures. In passive microwave, the precipitation particles are the main source of attenuation, hence microwave techniques are more direct technique than visible or infrared. The emitted radiation from the atmospheric particles increases the signal whereas the scattering due to hydrometeors decreases the radiation. Based on the upwelling radiation, the types and size of the hydrometeors can be detected. Above 60 GHz the scattering due to ice dominates, hence only ice can be detected but not rain. Whereas below 22 GHz the ice layer above rain becomes virtually transparent as the absorption is the primary mechanism affecting the microwave radiation transfer. The microwave frequency between 19.3 and 85.5 GHz mainly interacts with different types of hydrometeors like water particles or cloud droplets. The major limitation of the passive microwave technique is its poor spatial and temporal resolution. Due to the differential radiative characteristics of different earth surface features, the interpretation of this type of data becomes more difficult.

Special Sensor Microwave Imager (SSM/I) and Tropical Rainfall Measuring Mission Microwave Imager (TMI) are the major passive microwave sensors used for rainfall estimation. This is a scanning type instrument with four microwave frequencies, i.e. 19.35, 22.235, 37.0 and 85.5 GHz, with 1400-km wide swath. The distribution of the rainfall averages and its anomalies derived from TRMM are given in Fig.2.

4.1.3.   Active microwave technique

The Precipitation Radar (PR) is the best space-based precipitation measuring instrument till now. It operates at 13.8 GHz and a critical part of the Tropical Rainfall Measuring Mission (TRMM).

This radar can provide quantitative measurements of rainfall with better spatial resolution. It is also capable of providing the vertical profile of the rain along with layer thickness.

The comparative and complementarity among different sensors of TRMM, i.e. TMI, PR and Visible and Infrared radiometry System (VIRS), are given in table 2.

4.2 Air temperature

After  successful  representations  of  the  earth  surface  and  atmosphere,  the  focus  of  the   meteorologists  shifted  to  the  creation   of  vertical distribution  IF air temperature for  improved  initialization of the global numerical weather models. Several studies showed that the atmospheric  temperature is a function of atmospheric pressure and by using the multiple tangential viewing angles the information of changing temperature with altitude can be obtained

initialization of the global numerical weather models. Several studies showed that the atmospheric temperature is a function of atmospheric pressure and by using the multiple tangential viewing angles the information of changing temperature with altitude can be obtained initialization of the global numerical weather models. Several studies showed that the atmospheric temperature is a function of atmospheric pressure and by using the multiple tangential viewing angles the information of changing temperature with altitude can be obtained.

The GOES-VAS (Geostationary Orbit Environmental Satellites-Vertical Atmospheric Sounder) and TOVS (TIROS Operational vertical sounder) are two major satellite based instruments that are used for estimation of the spatial and diurnal variability of air temperature. These are also  capable to provide information on vertical temperature profiles. The collocated and coincident data of satellite soundings and shelter temperature observations are used to develop regression coefficients which are further deployed to estimate the air temperature. This method is capable to provide shelter temperature with an accuracy of about 2°C. During winter season due to the existing temperature inversion over cold earth surface, the satellite may overestimate the shelter temperature with a bias of around 1-2°C. Moreover, due to its poor vertical resolution the temperature estimates in the lowest 1 km of atmosphere is not accurate.

The radiometers cannot directly measure the air temperature, but by using land surface temperature and the regression parameters the same can be retrieved. Several coarse resolution sensors, like Advanced Very High Resolution Radiometer (AVHRR) and Moderate Resolution Imaging Spectro-radiometer (MODIS) have been extensively used for deriving the land surface temperature. Though strong relationship between land surface temperature and air temperature has been studied by many researchers, the major limitation of this method is the cloud cover. In some cases, the satellite derived vegetation index is used to estimate the air temperature with the assumption that surface temperature of a closed canopy is equal to the air temperature.

4.3. Solar radiation

Insolation measurements in agricultural field or weather stations are generally carried out using pyranometers, but due to the sparse distribution of the ground based meteorological measurements like cloud cover, cloud type, cloud height, bright sunshine hour, temperature etc., the spatial representation of the insolation is not proper. For derivation and mapping of the insolation surface using satellite data, several methods can be followed like (a) regression, (b) physically and digital Elevation Model (DEM) based approach. These techniques can be applied both on geostationary satellites (e.g. Meteosat, GOES) and polar orbiting satellites (e.g., Landsat TM, NOAA-AVHRR).

4.3.1. Regression approach

In this approach, the coincidental ground as well as satellite data is used to develop the regression coefficients. The amount of cloud is estimated for a given area from the satellite derived visible imageries, followed by deployment of regression equation for different categories of cloud to estimate solar radiation. By this method, the solar radiation can be estimated at daily level using geostationary satellite data from 0800 to 1600 hours.

4.3.2.   Physical approach

In this approach, the interaction of the incident and reflected radiation with the land surface features and atmosphere is treated in physical manner. Radiative transfer equations are used to estimate the solar radiation, by taking into account the concentration profiles of different atmospheric components, cloudiness and atmospheric water vapour. Though this approach could be able to provide higher accuracy, the regression approach remained the choice for operational use due to its simplicity.

Along with the satellite imageries, several other inputs, e.g. Julian day, time, solar zenith angle, cloud attenuation coefficients for different cloud categories, ozone content, atmospheric moisture content etc. are used for successful estimation of surface insolation.

4.4. Evapo-transpiration

With the availability of different resolution satellite imageries, remote sensing is becoming a useful tool for providing evapo-transpiration information at various scales. Methods for estimation of evapo-transpiration can be broadly categorized into (a) empirical direct method; (b) residual methods of the energy budget; and (c) indirect method.

In empirical direct method, satellite imageries are directly introduced in semi-empirical models to estimate the evapo-transpiration. Generally, empirical equations are developed using the surface temperature derived from satellite derived infrared imageries. This approach assumes that there exists

a direct relationship between daily evapo-transpiration (ETd) and instantaneous temperature difference (Ts-Ta)i between surface (Ts) and air temperature (Ta).

Where, Rn is the available net radiation. A & B are constant and depend on location.

The major hypothesis of this approach is that the ratio H/Rn remains constant all along the day and soil heat flux (G) is equal to zero (Gd = 0). The surface temperature (Ts) can be estimated from satellite derived infrared imageries after atmospheric corrections. Similarly, the Rn can be computed from satellite derived incident solar and atmospheric radiations, whereas, the A & B can be derived as a function of several parameters, like wind speed, roughness, atmospheric stability etc. But the problem of spatial representativity of air temperature is more arguable and particularly acute for regional studies.

The residual methods are an intermediate approach using both empirical relationships and physical parameterization. There are several models available for estimation of evapo-transpiration using satellite data, among them SEBAL, SEBI, S-SEBI, SEBS etc. have become popular. These models have been designed to compute the energy partitioning with minimum ground based data. The incoming solar radiation, surface temperature, albedo and vegetation indices maps are the requirements for these models. In these models, the emissivity, roughness length and soil heat flux (G) are retrieved from vegetation indices using semi-empirical relationships. The sensible heat flux (H) is computed by the flux inversion at dry non-evaporating land units and at wet surface types, whereas latent heat flux (LE) is estimated as the residual of energy balance

In indirect methods, the models describe the exchanges between soil, plant and atmosphere accordingly to the physical processes occurring in each compartment with generally finer time step, like hour or second (Courault et al., 2003). Based on the process description the complexity levels are introduced in the model, e.g. the evaporation and transpiration can be computed separately if the soil and plant behave differently. In this scenario, the infrared imageries acquired at different period of day and angle may be very useful. These models can also be executed for single layer or multi-layered vegetation, based upon the detail complexity involved and the expected accuracy. With increasing fineness of the surface and details of the process, the requirement of the parameters increases. Some of the parameters required by these models, e.g. soil moisture, surface temperature, emissivity, albedo, vegetation fraction, leaf area index etc. can be derived from satellite or remote sensing data. Though the data related to surface and stomatal resistance, roughness etc. is difficult to estimate through satellite data, can be retrieved based on the satellite derived canopy type or phonological information.

5. Radar meteorology

Radar, i.e. Radio Detection and Ranging, was mainly developed to detect and monitor air craft when they are out of visual range. Radar operates in microwave region of the electro-magnetic spectra for detection of location, altitude and movements of any stationary or moving object. Though the radar has been extensively used for war purpose, sometimes the blobs of precipitation hindered the detection of air craft. In February, 1941 the radar was first intentionally utilized for looking at rain and thus the limitation of radar was converted into its advantage.

A high-power radiation beam is transmitted by radar, whereas a small portion of the same is received back after interaction with objects. The emitted beam can be characterized by three fundamental properties; (i) pulse repetition frequency (PRF), (ii) transmission time, and (iii) beam width. The number of pulses transmitted per second is known as PRF. The transmission time is represented by the duration of each pulse. As the velocity of the beam is equal to the speed of light, by knowing the transmission time the pulse length can be calculated. The angular width of the beam is known as beam width. For any typical weather radar the PRF value is 325, pulse length is around 1 km and beam width is about 1°. By combining the pulse length and beam width information, the pulse volume can be calculated. The pulse length, band width and pulse volume are illustrated in Fig. 4.

While interacting with several atmospheric particles, e.g. raindrop, atmospheric gases, aerosols etc., the major parts of the beam energy are attenuated in terms of both scattering and absorption. Hence, a small fraction of the incident radiation could return back to antenna. The beam attenuation increases with length of atmosphere or the atmospheric particles. The heavy thunderstorm may result in around 5% reduction in backscattered power, whereas it may be upto 80% in case of rain and hailstorms. The frequencies used in the radar vary based on the kinds of applications. For detection of clouds and aerosols high frequency (W-band) whereas for heavy rain and hailstorms low frequency (L-band) are used. The high frequency beam is readily attenuated by smaller particles; on the contrary low frequency beams cannot even see the smaller targets. In major cases, S-band radar is used for weather based studies as it is the trade- off between high sensitivity and minimal attenuation. The radar frequencies along with their major applications are given in table 3.

Doppler radar can provide information on the speed and direction of a moving target. Due to the relative motions of the source of signal and observer, there will be a change in frequency observed by the observer. As the source and observer come closer, the frequency increase, and vice-versa. The change in frequency is proportional to the target velocity. This was named after by Christian Doppler, who first discovered this principle. By comparing the transmitted and received signal frequency, Doppler radars compute the frequency difference. The reflectivity (Z), radial velocity (V) and spectral width (w) are the three major base parameters that are available from Doppler Weather Radar (DWR).

These parameters are generated based on the “moment of power distribution”, arises due to the target and signal interaction. The 0th moment is the reflectivity (Z), i.e. the amount of power received by radar after scattering from the targets. In absence of any target the Z value will be zero, as no power will be reflected back. The unit of reflectivity is decibels (dBZ). More the power received by the radar, higher the values of reflectivity, e.g. 5-20 dBZ for light snow due to inefficient reflection, 30-45 dBZ for moderate rain and 60-75 dBZ for hailstorms.

The 1st moment, i.e. radial velocity (V) can be computed using the time variation of reflectivity. Two kinds of primary information can be gathered from the V, first is the (a) rotation and another is the (b) scan angle. The entire vortex, e.g. hurricane, tornado, mesocyclone etc., cannot be visualized by a single radar, but some of its parts moving towards or away from the radar can be captured. The component of the wind moving tangential to the radar will appear as very low radial velocities (http://www.mcwar.org). The second component, i.e. scan angle, can provide a vertical profile of rainfall or wind.

The 2nd moment of power distribution is spectrum width (w), which measures the time variation of the radial velocity. Generally, the turbulence like gust fronts, tornadoes, updraft etc. can be described by spectrum width. In case of the turbulence the values of ‘w’ become higher.

For last 40 years, the most significant application of radar is to monitor the real time weather condition and its related studies. The knowledge pertaining to structure of precipitating clouds, thunderstorms and cyclones are provided by Doppler radars. Based on the radar type, it is also possible to estimate the number and size of droplets in a cloud and the probability of presence of water or ice droplets. The Doppler radar can provide improved estimates of possible rainfall by eliminating the effects of different vertical structures on ground, i.e. buildings, hill etc., and minimizing the anomalous propagation echoes generated by birds, insects and other atmospheric condition etc. Though the radar is capable to provide information on detecting and monitoring the dynamic weather system along with prediction, care should be undertaken to eliminate the noise generated through atmosphere, earth curvature, permanent structure on ground etc.

6. Summary

• The availability of near-real time and reliable spatial information on meteorological and vegetation parameters through satellite technology contributes a large in weather and climate forecasting and large scale crop production monitoring.
• The meteorological satellites are of two types, i.e. polar orbiting and geostationary. The entire earth surface can be observed by polar orbiting satellites from a height of ~750-900 km, whereas the geostationary satellites (~36000 km) remain virtually stationary over the earth at a given location.
• Cloud-indexing, thresholding, life history and cloud model are some of the major rainfall estimating procedures using visible (VIS) and infrared (IR) satellite imageries whereas SSM/I and TRMM are the major microwave sensors used for rainfall estimation.
• AVHRR and MODIS have been extensively used for land surface temperature estimation, which can be further utilized for retrieval of air temperature using regression approach.
• The solar radiation parameters can be retrieved from both polar and geostationary satellites using regression and physical aaproaches.
• The evapo-transpiration can be estimated using the incoming solar radiation, surface temperature, albedo and vegetation indices information from satellite imageries. Few of the popular models are SEBAL, SEBI, S-SEBI, SEBS etc.
• With the advent of radar technology, the structure of precipitating clouds, thunderstorms and cyclones can also be estimated with high accuracy.