Search Results
You are looking at 1 - 3 of 3 items for :
- Author or Editor: Takmeng Wong x
- Journal of Atmospheric and Oceanic Technology x
- Refine by Access: All Content x
Determination of CERES TOA Fluxes Using Machine Learning Algorithms. Part I: Classification and Retrieval of CERES Cloudy and Clear ScenesAbstract
Continuous monitoring of the earth radiation budget (ERB) is critical to the understanding of Earth’s climate and its variability with time. The Clouds and the Earth’s Radiant Energy System (CERES) instrument is able to provide a long record of ERB for such scientific studies. This manuscript, which is the first of a two-part paper, describes the new CERES algorithm for improving the clear/cloudy scene classification without the use of coincident cloud imager data. This new CERES algorithm is based on a subset of the modern artificial intelligence (AI) paradigm called machine learning (ML) algorithms. This paper describes the development and application of the ML algorithm known as random forests (RF), which is used to classify CERES broadband footprint measurements into clear and cloudy scenes. Results from the RF analysis carried using the CERES Single Scanner Footprint (SSF) data for January and July are presented in the manuscript. The daytime RF misclassification rate (MCR) shows relatively large values (>30%) for snow, sea ice, and bright desert surface types, while lower values (<10%) for the forest surface type. MCR values observed for the nighttime data in general show relatively larger values for most of the surface types compared to the daytime MCR values. The modified MCR values show lower values (<4%) for most surface types after thin cloud data are excluded from the analysis. Sensitivity analysis shows that the number of input variables and decision trees used in the RF analysis has a substantial influence on determining the classification error.
Abstract
Continuous monitoring of the earth radiation budget (ERB) is critical to the understanding of Earth’s climate and its variability with time. The Clouds and the Earth’s Radiant Energy System (CERES) instrument is able to provide a long record of ERB for such scientific studies. This manuscript, which is the first of a two-part paper, describes the new CERES algorithm for improving the clear/cloudy scene classification without the use of coincident cloud imager data. This new CERES algorithm is based on a subset of the modern artificial intelligence (AI) paradigm called machine learning (ML) algorithms. This paper describes the development and application of the ML algorithm known as random forests (RF), which is used to classify CERES broadband footprint measurements into clear and cloudy scenes. Results from the RF analysis carried using the CERES Single Scanner Footprint (SSF) data for January and July are presented in the manuscript. The daytime RF misclassification rate (MCR) shows relatively large values (>30%) for snow, sea ice, and bright desert surface types, while lower values (<10%) for the forest surface type. MCR values observed for the nighttime data in general show relatively larger values for most of the surface types compared to the daytime MCR values. The modified MCR values show lower values (<4%) for most surface types after thin cloud data are excluded from the analysis. Sensitivity analysis shows that the number of input variables and decision trees used in the RF analysis has a substantial influence on determining the classification error.
Abstract
Clouds and the Earth’s Radiant Energy System (CERES) is a NASA spaceborne measurement program for monitoring the radiation environment of the earth–atmosphere system. The first CERES instrument is scheduled to be launched on board the Tropical Rainfall Measuring Mission (TRMM) satellite in late 1997. In addition to gathering traditional cross-track fixed azimuth measurements for calculating monthly mean radiation fields, this single CERES scanner instrument will also be required to collect angular radiance data using a rotating azimuth configuration for developing new angular dependence models (ADMs). Since the TRMM single CERES instrument can only be run in either one of these two configurations at any one time, it will need to be operated in a cyclical pattern between these two scan modes to achieve the intended measurement goals. To minimize the errors in the derived monthly mean radiation field due to missing cross-track scanner measurements during this satellite mission, determination of the optimal scan mode sequence for the TRMM single CERES instrument is carried out. The Earth Radiation Budget Experiment S-4 daily mean cross-track scanner data product for April and July 1985 and January 1986 is used with a simple temporal sampling scheme to produce simulated daily mean cross-track scanner measurements under different TRMM CERES operational scan mode sequences. Error analysis is performed on the monthly mean radiation fields derived from these simulated datasets. It is found that the best monthly mean result occurred when the cross-track scanner is operated on a “2 days on and 1 day off” mode. This scan mode sequence will effectively allow for 2 consecutive days of cross-track scanner data and 1 day of angular radiance measurement for each 3-day period. The root-mean-square errors for the monthly mean all-sky (clear sky) longwave and shortwave radiation field, due to missing cross-track scanner measurements for this particular case, are expected to be less than 2.5 (0.5) and 5.0 (1.5) W m−2, respectively.
Abstract
Clouds and the Earth’s Radiant Energy System (CERES) is a NASA spaceborne measurement program for monitoring the radiation environment of the earth–atmosphere system. The first CERES instrument is scheduled to be launched on board the Tropical Rainfall Measuring Mission (TRMM) satellite in late 1997. In addition to gathering traditional cross-track fixed azimuth measurements for calculating monthly mean radiation fields, this single CERES scanner instrument will also be required to collect angular radiance data using a rotating azimuth configuration for developing new angular dependence models (ADMs). Since the TRMM single CERES instrument can only be run in either one of these two configurations at any one time, it will need to be operated in a cyclical pattern between these two scan modes to achieve the intended measurement goals. To minimize the errors in the derived monthly mean radiation field due to missing cross-track scanner measurements during this satellite mission, determination of the optimal scan mode sequence for the TRMM single CERES instrument is carried out. The Earth Radiation Budget Experiment S-4 daily mean cross-track scanner data product for April and July 1985 and January 1986 is used with a simple temporal sampling scheme to produce simulated daily mean cross-track scanner measurements under different TRMM CERES operational scan mode sequences. Error analysis is performed on the monthly mean radiation fields derived from these simulated datasets. It is found that the best monthly mean result occurred when the cross-track scanner is operated on a “2 days on and 1 day off” mode. This scan mode sequence will effectively allow for 2 consecutive days of cross-track scanner data and 1 day of angular radiance measurement for each 3-day period. The root-mean-square errors for the monthly mean all-sky (clear sky) longwave and shortwave radiation field, due to missing cross-track scanner measurements for this particular case, are expected to be less than 2.5 (0.5) and 5.0 (1.5) W m−2, respectively.
Abstract
The NOAA-9 Earth Radiation Budget Experiment (ERBE) scanner measured broadband shortwave, longwave, and total radiances from February 1985 through January 1987. These scanner radiances are reprocessed using the more recent Clouds and the Earth’s Radiant Energy System (CERES) unfiltering algorithm. The scene information, including cloud properties, required for reprocessing is derived using Advanced Very High Resolution Radiometer (AVHRR) data on board NOAA-9, while no imager data were used in the original ERBE unfiltering. The reprocessing increases the NOAA-9 ERBE scanner unfiltered longwave radiances by 1.4%–2.0% during daytime and 0.2%–0.3% during nighttime relative to those derived from the ERBE unfiltering algorithm. Similarly, the scanner unfiltered shortwave radiances increase by ~1% for clear ocean and land and decrease for all-sky ocean, land, and snow/ice by ~1%. The resulting NOAA-9 ERBE scanner unfiltered radiances are then compared with NOAA-9 nonscanner irradiances by integrating the ERBE scanner radiance over the nonscanner field of view. The comparison indicates that the integrated scanner radiances are larger by 0.9% for shortwave and 0.7% smaller for longwave. A sensitivity study shows that the one-standard-deviation uncertainties in the agreement are ±2.5%, ±1.2%, and ±1.8% for the shortwave, nighttime longwave, and daytime longwave irradiances, respectively. The NOAA-9 and ERBS nonscanner irradiances are also compared using 2 years of data. The comparison indicates that the NOAA-9 nonscanner shortwave, nighttime longwave, and daytime longwave irradiances are 0.3% larger, 0.6% smaller, and 0.4% larger, respectively. The longer observational record provided by the ERBS nonscanner plays a critical role in tying the CERES-like NOAA-9 ERBE scanner dataset from the mid-1980s to the present-day CERES scanner data record.
Abstract
The NOAA-9 Earth Radiation Budget Experiment (ERBE) scanner measured broadband shortwave, longwave, and total radiances from February 1985 through January 1987. These scanner radiances are reprocessed using the more recent Clouds and the Earth’s Radiant Energy System (CERES) unfiltering algorithm. The scene information, including cloud properties, required for reprocessing is derived using Advanced Very High Resolution Radiometer (AVHRR) data on board NOAA-9, while no imager data were used in the original ERBE unfiltering. The reprocessing increases the NOAA-9 ERBE scanner unfiltered longwave radiances by 1.4%–2.0% during daytime and 0.2%–0.3% during nighttime relative to those derived from the ERBE unfiltering algorithm. Similarly, the scanner unfiltered shortwave radiances increase by ~1% for clear ocean and land and decrease for all-sky ocean, land, and snow/ice by ~1%. The resulting NOAA-9 ERBE scanner unfiltered radiances are then compared with NOAA-9 nonscanner irradiances by integrating the ERBE scanner radiance over the nonscanner field of view. The comparison indicates that the integrated scanner radiances are larger by 0.9% for shortwave and 0.7% smaller for longwave. A sensitivity study shows that the one-standard-deviation uncertainties in the agreement are ±2.5%, ±1.2%, and ±1.8% for the shortwave, nighttime longwave, and daytime longwave irradiances, respectively. The NOAA-9 and ERBS nonscanner irradiances are also compared using 2 years of data. The comparison indicates that the NOAA-9 nonscanner shortwave, nighttime longwave, and daytime longwave irradiances are 0.3% larger, 0.6% smaller, and 0.4% larger, respectively. The longer observational record provided by the ERBS nonscanner plays a critical role in tying the CERES-like NOAA-9 ERBE scanner dataset from the mid-1980s to the present-day CERES scanner data record.
