Cloud cover, cloud liquid water and cloud attenuation at Ka and V bands over equatorial climate

Transcription

1 METEOROLOGICAL APPLICATIONS Meteorol. Appl. : (4) Published online 4 October in Wiley Online Library (wileyonlinelibrary.com) DOI:./met.47 Cloud cover, cloud liquid water and cloud attenuation at Ka and V bands over equatorial climate T. V. Omotosho, a,b,c * J. S. Mandeep b,c and M. Abdullah b,c a Department of Physics, College of Science and Technology, Covenant University, Ota, Nigeria b Institute of Space Science, Universiti Kebangsaan Malaysia, UKM Bangi, Malaysia c Department of Electrical, Electronic and System Engineering, Universiti Kebangsaan Malaysia, UKM Bangi, Malaysia ABSTRACT: Cloud cover statistics and their diurnal variation have been obtained from in situ and satellite measurements for three equatorial locations. Cloud liquid water content, C isotherm height and cloud attenuation have also been obtained from radiosonde measurement using the so-called Salonen model at Kuala Lumpur (Malaysia). The results show a strong seasonal variation of cloud cover and cloud liquid water content on the two monsoon seasons. The Liquid Water Content (LWC) obtained from radiosonde and the Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI) is higher during the Northeast Monsoon season, which corresponds to the period of higher percentage cloud cover and high rainfall accumulation. The International Telecommunication Union Region (ITU-R) model underestimates the cumulative distribution of LWC values at the present station. The relationship of the cloud attenuation, derived from the profiles of liquid water density and temperature within the cloud, shows an underestimate by the data obtained from the ITU-R model. The cloud attenuation at Kuala Lumpur is somewhat underestimated by the ITU-R model up to about. db at Ka ( GHz) and.4 db at V (5 GHz) bands. The results of the specific attenuation can be used for the estimation of cloud attenuation at microwave and millimetre wave over earth-space paths. The present data are important for planning and design of satellite communications at Ka and V bands on the Earth space path in the equatorial region. KEY WORDS cloud cover statistics; cloud temperature; cloud liquid water contents; C isotherm height; cloud attenuation Received January ; Revised 6 May ; Accepted 7 May. Introduction Technological advances in remote sensing and satellite missions in recent times, such as NASA (A-Train) Atmospheric Infrared Sounder (AIRS), Moderate Resolution Imaging Spectroradiometer (MODIS), CLOUDSAT, Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO), TRMM Microwave Imager, Cloud and Earth radiation sensor Cloud and Earth radiation sensor (CERES), and weather stations making direct measurements and recording observations such as NASA Student Cloud Observation Online, (SCOOL) have contributed greatly to the wealth of cloud data available to the scientific community for new research in satellite to ground radiowave propagations studies. The properties of clouds most important for cloud attenuation modelling on Earth-space communication links at the Ka and V bands are statistics of cloud cover, cloud base height, cloud thickness, total cloud liquid water content, cloud temperature, cloud horizontal extent and variability. Cloud climatologies have been developed in recent times from two kinds of data; () using radiance measured by satellites in geostationary and polar orbits, and, () using visual observation of clouds from the Earth s surface as coded in weather reports from stations on land and ships at sea. Satellites detect clouds principally at visible and * Correspondence: T. Omotosho, Department of Physics, College of Science and Technology, Covenant University, PMB Ota, Nigeria. tomotosho@covenantuniversity.com thermal-infrared wavelengths: during the daytime cloud can be detected in both wavelength regions, but at night only in the thermal infrared. Satellite observations of clouds are made more frequently than surface observations and cover many land and ocean areas compared to surface observations, but ground observation offers a useful adjunct to satellite observations. The surface observer views clouds from below, thus observing low clouds that are often hidden from satellite views by higher clouds. Multiple cloud layers often occur together, so the views from above and below are complementary (Warren et al., 7). The present work is concerned with reporting cloud occurrence statistics and estimating cloud attenuation in the Ka ( GHz) and V (5 GHz) bands based on cloud climatology data from ground and satellite measurements with a view to understanding the diurnal, monthly and seasonal impact of cloud at these two frequency bands. This is very important because of the rapid growth in new satellite services using very small aperture terminals (VSATs) and ultra-small aperture terminals (USATs). The abundance bandwidth available at Ka and V bands for various applications of radio systems such as higher data transmission rate, fast internet access, multimedia applications, local area network (LAN) connections, supervisory control and data acquisition (SCADA) and point of sales verifications will be increasingly important in the coming years. At these frequency bands, radio signals are expected to experience losses due to cloud and cloud noise temperature in clear-sky and non-clear sky conditions at any time of the day, depending on the type of season and the climate of the location (Bouchard, 8). Cloud and rain are the most Royal Meteorological Society

2 KL A B C 778 T. V. Omotosho et al. Figure. Map of Malaysia and the position of Kuala-Lumpur (**KL), with the three stations, Tampin (**A), South China Sea (**B) and Mukah (**C). Table. TRMM sensors performance characteristics (Kummerow and Barnes, 998). Microwave radiometer (TMI) Radar (PR) Visible and infrared radiometer (VIRS) Frequencies.7, 9.,., 7., and.8 GHz.6,.6,.75,.8, and μm 85.5 GHz (dual-polarized except for. vertical only) Resolution km 8km field of view at 5 km footprint and 5 m vertical.5 km resolution 7 GHz resolution Ground resolution 4. km at nadir 4.4 km at 85.5 GHz. km at nadir Scanning Conically scanning Cross-track scanning Cross-track scanning Swath width 88 km swath 5 km swath 8 km swath important factor in planning high availability satellite communication systems in equatorial and tropical regions because of higher relative humidity, greater frequency of rainfall, and higher percentage of cloud cover. Although the effect of rain on radio waves is more than the effect due to clouds, the occurrence of cloud is always more than rain (Mattioli et al., 9) and rain can be traced to the formation of certain types of clouds. A lot of work has been done on the statistics of rain and its effects on radio wave propagation in Malaysia. Cloud impairment related work on radiowave propagation is sparse. There are a few results on cloud morphology and cloud occurrence statistics in relation to radio wave propagation studies at Ku band only in Malaysia (Mandeep and Hassan, 8;Mandeep et al., 9). In the present paper radiosonde and cloud cover data were retrieved from ground and satellite observations for three locations covering the western and eastern parts of Malaysia. The data will be useful to estimate and predict the impact of clouds on radio wave signal at Ka and V bands in Malaysia.. Data sources.. Database of cloud reports Ground observations of low cloud data from 97 to 996 were retrieved online for three ground stations in Malaysia: A(Tampin,.5 N,.5 E) land area 68%, B (an island in the South China Sea,.5 N, 7.5 E) land area %, and C(Mukah,.5 N,.5 E) land area 65% (Figure ). Also, satellite observations from the International Satellite Cloud Climatology Project (ISCCP, ) from 98 to 6 were retrieved online for the three stations for comparison. The cloud cover climatology over the land surface by Warren et al. (7) is a 5 by 5 latitude/longitude grid. Ground observing sites used to form the long term (97 996) cloud climatology data for stations A, B, and C are 7, and 4 daily-reporting sites, respectively. Station A has more reporting sites (seven), than the other two stations. One of the reporting sites close to station A is Kuala Lumpur (. N,.55 E). Each station was equipped to measure cloud base height, cloud types and cloud cover, and cloud amount for both day and night. The low cloud particle temperature was also retrieved from My NASA Data Live Access Server (Durre et al., 6) for five cloud cover types: cumulonimbus (Cb), nimbostratus (Ns), cumulus (Cu) stratus (St) and stratocumulus (Sc) over Kuala Lumpur. The monthly average of cloud coverage from 994 to 7 was used in these work... Database of radiosonde reports Daily radiosonde measurements from 97 to 998 were retrieved online from Integrated Global Radiosonde Archive (Geleyn, 98) for Kuala Lumpur. Consistent radiosonde (Sonde Model I VAISALA RS8) measurements, launched twice daily at and GMT from 97 to 97, were used to estimate cloud-liquid-water statistics over Kuala Lumpur. Radiosonde data can indicate the presence of cloud liquid water depending on whether the relative humidity exceeds the critical humidity as define by Salonen and Uppala (99). The phase of water within the cloud is determined by the temperature. If the temperature is greater than C, the water is completely in the liquid phase, and if the temperature is less than C only ice exists. The temperature between and C yields a mixture of liquid water and ice. Salonen proposed a model for cloud water density in terms of temperature and height of cloud base that can be used to obtain the liquid water density profile within the cloud. There can be a significant variation of water density within the cloud. The pure ice does not cause any attenuation of radio signal, but super-cooled water can cause very significant attenuation.

3 Cloud liquid water and cloud attenuation at Ka and V bands 779 (a) (b) (c) Figure. Average cloud cover (a) type for day (b) and night (c), TRMM sensors performance characteristics The Tropical Rainfall Measuring Mission (TRMM) is a joint project between the National Aeronautics and Space Administration (NASA) and the National Space Development Agency (NASDA) of Japan that was used to measure rainfall and energy in tropical and sub-tropical regions. The primary rainfall instruments on TRMM are the TRMM Microwave Imager (TMI), the precipitation radar (PR), and the Visible and Infrared Radiometer System (VIRS). Table summarizes the characteristics of the TRMM sensors performance characteristics. The TMI has nine passive channels of microwave radiometer based upon the Special Sensor Microwave/Imager (SSM/I). The TMI has five separate frequencies, which are.7, 9.,., 7 and 85.5 GHz used to measures the radiation intensity. The TMI scanned a 88 km wide swath in a conical manner with a viewing angle of 49 off nadir. PR is a rain radar that provides a three-dimensional structure of rainfall based on obtaining the quantitative rainfall measurements over and ocean (Kummerow and Barnes, 998). The PR is equipped with a 8-element active phased array system operating at.8 GHz. The PR scanned a 5 km wide swath in cross-track scanning manner with a ground resolution of 4.4 km at nadir. The VIRS is a five-channel imaging spectroradiometer with bands in the wavelength range from.6 to μm. The VRIS scanned a 8 km wide swath with cross track scanning with a ground resolution of. km at nadir.. Model descriptions.. Salonen model The Salonen model identifies clouds when the relative humidity exceeds the critical humidity function RH c : RH c = ασ ( σ ) [ + β (σ.5)] () where s = P(i)/P(), andp(i) and P() are the pressures (hpa) at the considered atmospheric i th level and at the ground, respectively. The two empirical parameters in Equation () are

4 78 T. V. Omotosho et al. Figure. Day and night monthly average cloud cover α =. and β =. The total cloud water content within each cloud layer is a function of the height above the cloud base and of the temperature in the layer, as given in Equation (): ( ) a w h hb o h r ( + ct ) T C TWC (h, T ) = ( ) a w h hb o h r [exp (ct )] T C () where w =.7 kg m, h and h b are the heights (km) above the surface and of the cloud base, h r =.5 km, a =, and c =.4 C. Liquid Water Content (LWC) and Ice Water Content (IWC) are given by Equation (): LWC = TWC (h, T ). [ fw (T ) ] () where the fraction of cloud liquid f W is given by Equation (4): T C f w = + T T C (4) T < C The specific attenuation due to the cloud can be expressed in terms of the frequency, f, and the complex dielectric permittivity, ɛ, (ITU-R 849, 9) as: γ c (h) =.89 f w (h) 8 s ( ii + η ) db km (5) where ɛ = ɛ i + jɛ ii. The constant η is given by: η = + si s ii (6)

5 Cloud liquid water and cloud attenuation at Ka and V bands 78 (a) 4.8 C Isotherm height (km) jan feb mar apr may jun jul aug sep oct nov dec (Months) Monthly Average of C Isothermal height 97 to 98 (b) C Isotherm height (km) DJF MAM JJA SON Seasons (months) Seasonal pattern of C isotherm height Figure 4. Monthly and seasonal rain height for Kuala Lumpur, Monthly (a) and seasonal (b) average of C isothermal height The complex permittivity, ɛ, can be obtained by using a double-debye model, which is a function of frequency and temperature (Manabe et al., 987). Hence, the specific attenuation, γ c (h), being a function of temperature and liquid water density, varies within the cloud yielding a height profile. The total cloud attenuation can, therefore, be obtained by integrating γ c (h) along the Earth-space path within the cloud as follows: A = ht γ c (h) dh (7) sin θ h b Here, h b and h t are, respectively, the heights of the base and top of the cloud layer. Two Matlab programs were written for Equations () (7) for the computation of cloud-liquid-water profile and cloud attenuation, taking into consideration parameters such cloud base height, cloud height and temperature profile. By using the expression provided by Slobin (98) the specific attenuation of radio waves due to cloud at different wavelengths was calculated. The expression for attenuation of radio wave, (σ ) due to the cloud is given by: σ = 4.4M.(9 T ).6 λ (8) where σ is the specific attenuation due to clouds (db km ); M, is the cloud water particle density (g m ); T, is the cloud temperature ( K) and λ, is the wavelength of radio wave (cm). 4. Results and discussion Cloud cover characteristics have been derived for five types of low cloud: cumulonimbus (Cb), nimbostratus (Ns), cumulus (Cu) stratus (St) and stratocumulus (Sc) averaged monthly and yearly for the measurement period The long term yearly average ( years) is presented in Figure for both day and night at the three stations. Figure presents the monthly average ( ) of low cloud cover for the three stations. It is seen that low cloud occurrence over the three stations in Malaysia is very significant during October to January for both day and night. The monthly cumulative distribution of rainfall is influenced by the seasonal monsoons, the Northeast Monsoon from October to March and the southwest monsoon from April to September. The Northeast Monsoon is marked by heavy rainfall while the Southwest Monsoon season (April to September) has lower cloud cover due to less rainfall (Mandeep et al., 9). Station B, which is largely surrounded by the sea, had the highest cloud cover for both day and night.

6 78 T. V. Omotosho et al. Height (m) Cloud liquid water kg m Critical humidity (unitless) Relative humidity (unitless) Cloud liquid water/relative humidity/critical humidity (x-axis on the same scale) Figure 5. A typical radiosonde profile of cloud liquid water density on 5 August 97 at Kuala Lumpur obtained from radiosonde measurements. (a) Frequency kg m Daily liquid water content (kg m ) between 97 and 97 (76 days) (b) Integrated liquid water content (kg m ) Percentage of time ordinate exceeded (%) LWC Model for the whole period at Kuala Lumpur compared with ITU-RP data base Radiosonde data ITU-RP model Figure 6. Frequency and cumulative distributions of Liquid Water Content (LWC) obtained from radiosonde measurements at Kuala Lumpur. (a) Daily liquid water content (kg m ) between 97 and 97 (76 days). (b) LWC Model for the whole period at Kuala Lumpur compared with ITU-RP data base. The monthly and seasonal variation (97 98) of rain height in relation to C isotherm height for Kuala Lumpur is presented in Figure 4. It is seen that the monthly C isotherm height varies from 4.6 to 4.8 km and was found to be maximum in April and minimum in August. Seasonally, the C isotherm height peaks in March, April, May (MAM) and is very low in June, July, August (JJA). This seasonal pattern is seen in Figure for the three stations: low cloud cover minimum occurred also in JJA, this corresponds to the southwest monsoon season when low rainfall is recorded in Malaysia. The value of the C isotherm height assigned by the ITU radio propagation study group digital data for Kuala Lumpur is 4.96 km, while the annual average value from from the present work is 4.68 km: the difference

7 Cloud liquid water and cloud attenuation at Ka and V bands 78 Cloud liquid water kg m.7 (a) DJF MAM JJA SON 'Seasons variation' Seasonal pattern of cloud liquid water obtained from radiosonde measurement for Kuala Lumpur Liquid water density g m (b) Jan feb mar apr may jun jul aug sep oct nov dec 'Months' Monthly average cloud liquid water density from TRMM TMI Satellite 998 Figure 7. (a) Seasonal variation of Liquid Water Content (LWC) at a site (Kuala Lumpur) close to Station A. (b) Monthly variation of liquid water density from TRMM TMI for Kuala Lumpur the nearest site to Station A. between the two values is about.8 km. The Cisotherm is an important climatic input parameter in modelling rain attenuation for any location. Figure 5 presents the profile of liquid water density obtained from the Salonen model using radiosonde data on 5 August 97. In the figure the cloud layer is shown to have formed in the height range for which the relative humidity exceeds the critical humidity. The relative humidity is referred to here as water vapour saturation with respect to liquid water, which corresponds to the rapid spatial variations of the concentration of cloud drops and the cloud liquid content. The total liquid water content (LWC) has been obtained by integrating the liquid water density profile. The frequency distribution and cumulative distribution of LWC obtained using the Salonen model from the radiosonde data for Kuala Lumpur along with that obtained from the ITU-R model (ITU-R P. 84-4, 9), are presented in Figure 6. It is seen that the LWC at Kuala Lumpur peaked at.49 kg m with a frequency of 55. The cumulative distribution of LWC is somewhat underestimated by the ITU-R model for Kuala Lumpur. Figure 7(a) presents the seasonal variation of total LWC, seasonally between July 97 and December 97: 44 radiosondes were launched, of which 98 were launched at GMT, 5 at GMT and at GMT. An average of radiosondes were launched in every month, which is a good statistical sample. LWC maximum occurred in March, April May (MAM), and is a minimum in December, January, February, (DJF) and September, October, November (SON). LWC maximum occurred at the same season with a C isotherm maximum as seen in Figure 4 and when LWC was averaged for the two monsoon seasons (northeast, October to March) and (southeast, April to September) the values were.5 and.8 kg m, respectively. It is seen that LWC is higher during the Northeast Monsoon season, which corresponds to the season of high percentage of cloud cover over Kuala Lumpur as shown in Figure. The results suggest that during the season of heavy rainfall, October to March, total integrated cloud liquid water content is very high, and there could be a greater loss of radio wave signal due to clouds. The cloud temperature during the Northeast Monsoon was around 66 K while during the Southeast Monsoon months the temperature is around 67 K (MY NASA, ). Figure 7(b) also presents the cloud liquid water retrieved from the Tropical Rainfall Measurement Mission (TRMM) Microwave Imager (TMI) sensor from 998 to. The monthly average of cloud liquid water from the surface to 8 km above Kuala Lumpur was a minimum in January and a maximum in December. During the Northeast Monsoon it was approximately.9 g m while during the Southeast Monsoon months the value is.4 g m. This result confirms again that during the Northeast Monsoon season (of heavy rainfall, October to March) cloud water particle density is higher, due to the higher percentage of cloud cover, than the Southeast Monsoon season

8 784 T. V. Omotosho et al. (low rainfall, April to September) and there could be greater loss of radio wave signal due to clouds during the season. The results of specific attenuation due to cloud for Kuala Lumpur covering the Ku, Ka and V band frequencies are presented in Table for the two monsoon seasons. The results of the specific attenuation showed that cloud attenuation varies with the two seasons and is.6 db km at GHz and increases up to.6 db km at GHz. The cloud attenuation of.6 db km at GHz corresponds to the rain attenuation at rainfall rate of mm h. The results presented in Table and Figures compared well with the 5 years in situ measurement (Mandeep and Hassan, 8) at Penang (5. N,.7 E) about 4 km from the present station at Kuala Lumpur. Cloud attenuation over an Earth-space path was determined from the profile of liquid water density estimated in Figure 6 at two frequencies, and 5 GHz in the zenith direction (θ = 9 ), for the whole year. However, since the total liquid water content is the more abundantly available parameter, it is useful to relate the cloud attenuation to LWC. To indicate the relation, the cloud attenuation has been plotted against LWC. Figure 8 presents the curve generated from the ITU-R model compared with the present work: the ITU-R model underestimates the actual cloud attenuation at and 5 GHz. Also, Figure 9 gives the cumulative distributions at and 5 GHz attenuation at Kuala Lumpur with the Salonen Table. Specific attenuation of radio wave due to cloud at different frequencies. Northeast Monsoon Southeast Monsoon Frequency Band Specific attenuation Specific attenuation (GHz) name (db km ) (db km ) Ku Ku.. Ka K V V W W.6.99 model from radiosonde measurements for the whole period, along with the ITU-R model-generated distribution for the station. It is to be noted that although the radiosonde-generated LWC occurrences at Kuala Lumpur (a site close to station A) is higher than those obtained from the ITU-R model, as indicated in Figure 6, the attenuation distributions give a comparable picture. It can be seen again that the ITU-R model underestimates the actual cloud attenuation at and 5 GHz, up to about. db, at Ka and.4 db at V band, particularly when LWC is high. Attenuation (db) Attenuation (db) y =.7769x + E Liquid water content (kg m ) y =.9x + E- Ka_Radiosonde data ITU_Model Fitted linear curve Liquid water content (kg m ) V_Radiosonde data ITU_Model Fitted linear curve Figure 8. Cloud attenuation variation at and 5 GHz against Liquid Water Content (LWC) obtained at Kuala Lumpur from radiosonde data during the whole period. Also shown are the linear curve fitted to the data and the ITU-R model generated linear curve.

9 Cloud liquid water and cloud attenuation at Ka and V bands Kuala-Lumpur_Ka db ITU_Ka db Attenuation (db) Percentage of time ordinate exceed (%) Attenuation (db) Kuala-Lumpur_V db ITU_V db... Percentage of time ordinate exceed (%) Figure 9. Cumulative distributions of cloud attenuation at and 5 GHz obtained from radiosonde measurements at Kuala Lumpur for the whole period. Also shown are the data obtained from the ITU-R model for Kuala Lumpur. 5. Conclusion Cloud cover statistics, and cloud temperature, have been obtained from in situ (climatic atlas of clouds) and satellite measurements (NASA ISCCP, and TRMM TMI), respectively. Cloud liquid water contents, C isotherm height and cloud attenuation have also been obtained from radiosonde measurement and the Salonen model, respectively, for equatorial stations. The results show a strong variation of cloud cover and cloud liquid water content (LWC) in the two monsoon seasons. It is seen that LWC obtained from radiosonde and TRMM TMI is higher during the northeast monsoon season (October to March), which corresponds to the season of higher percentage of cloud cover and high rainfall accumulation. The ITU-R model underestimates LWC values at the present station. The relationship of the cloud attenuation, derived from the profiles of liquid water density and LWC shows some departure from the data obtained from the ITU-R model at Ka ( GHz) and V (5 GHz) bands. The cloud attenuation at Kuala Lumpur is somewhat underestimated by the ITU-R model. The results of the specific attenuation can be used for the estimation of cloud attenuation at microwave and millimetre wavelengths over Earth-space paths. The present data are important in view of the upcoming satellite communications at Ka and V-band in the equatorial region. Acknowledgements The authors wish to acknowledge the Institute of Space Science, Universiti Kebangsaan Malaysia. This research is jointly funded by the Malaysian Government through Universiti Kebangsaan Malaysia under the Science Fund 4-- -SF599, UKM-GUP.NBT and UKM-DLP--. References Bouchard P. 8. Observations of attenuation due to liquid bearing stratocumulus clouds over Ottawa using a ground-based profiling radiometer. th Specialist Meeting on Microwave Radiometry and Remote Sensing of the Environment, 4 March 8, Florence, Italy. Durre I, Russell SV, David BW. 6. Overview of the integrated global radiosonde archive. J. Clim. 9: Geleyn JF. 98. Some diagnostics of the cloud/radiation interaction in ECMWF forecasting model. ECMWF Workshop on Radiation and Cloud-Radiation Interaction in Numerical Modelling, 5 7 October 98, Reading, UK. International Satellite Cloud Climatology Project (ISCCP).. (accessed August ). ITU-R Recommendation P Attenuation due to cloud and fog. Kummerow C, Barnes W The Tropical Rainfall Measuring Mission (TRMM) sensor package. J. Atmos. Oceanic Technol. 5: Manabe TH, Liebe HJ, Hufford G Complex permittivity of water between and THz. International Conference on Infrared and Millimeter Waves, 4 8 December 987, Lake Buena Vista, Florida. Mandeep JS, Hassan SIS. 8. Cloud attenuation in millimeter wave and microwave frequencies for satellite applications over equatorial climate. Int J. Infrared Milli Waves 9: 6. Mandeep JS, Ojo JS, Emiliani LD. 9. Statistics of annual and diurnal cloud attenuation over equatorial climate. IET Commun. : Mattioli V, Basili P, Bonafoni S, Ciotti P, Westwater ER. 9. Analysis and improvements of cloud models for propagation studies. Radio Sci. 44: RS5, pp. My NASA Data Live Access Server.. nasa.gov/las/ (accessed March ). Salonen E, Uppala W. 99. New prediction method of cloud attenuation. Electron. Lett. 7: 6 8. Slobin SD. 98. Microwave noise temperature and attenuation of clouds: statistics of these effects at various sites in the United States, Alaska and Hawaii. Radio Sci. 7: Warren SG, Eastman RM, Hahn CJ. 7. A survey of changes in cloud cover and cloud types over land from surface observations, J. Clim. :