Estimation of Monthly Rainfall over Japan and Surrounding Waters from a Combination of Low-Orbit Microwave and Geosynchronous IR Data

View More View Less
  • a NASA/Goddard Space Flight Center, Greenbelt, Maryland
  • | b Science Systems and Applications, Inc., Seabrook, Maryland
  • | c General Sciences Corporation, Laurel, Maryland
© Get Permissions
Full access

Abstract

This paper describes a method to combine geosynchronous IR and low-orbit microwave data to estimate mean monthly rainfall useful for climate studies. The IR data have the advantage of high time resolution (important for rapidly changing precipitation patterns and for the detection of diurnal signals) but lack a strong physical connection between the remotely sensed signal and the surface rainfall. The microwave data provide a stronger relation between the radiance and the rainfall but provide poor time sampling of the rainfall signal.

The microwave technique uses the brightness temperature at 37 and 86 GHz from the Special Sensor Microwave/Imager instrument on board the Defense Meteorological Satellite Program (DMSP) satellite to define raining areas over water and land and uses the 86-GHz scattering signal to assign rain rate based on cloud model-microwave calculations. The microwave results are generally good for both individual swaths and monthly totals, except for a glaring underestimation of shallow, orographic rain systems over the southern coast of Japan. The IR techniques used are the GOES precipitation index of Arkin and Meisner and the convective-stratiform technique of Adler and Negri.

Initially the IR estimates are computed separately using hourly data from the Japanese Geostationary Meteorological Satellite. Calibration or adjustment factors are derived by dividing the microwave monthly estimate by a second IR estimate (made with the microwave sampling that simulates the observations from an IR radiometer on board the DMSP satellite). The spatial array of coefficients are then multiplied by the original IR monthly estimates (produced from all the hourly data) to produce the merged IR-Microwave monthly estimates. The results show that in areas where the base (microwave) technique performs well, that is, has a relatively small bias, the combined microwave-IR monthly total estimates have better error statistics than either the microwave or IR techniques individually.

Abstract

This paper describes a method to combine geosynchronous IR and low-orbit microwave data to estimate mean monthly rainfall useful for climate studies. The IR data have the advantage of high time resolution (important for rapidly changing precipitation patterns and for the detection of diurnal signals) but lack a strong physical connection between the remotely sensed signal and the surface rainfall. The microwave data provide a stronger relation between the radiance and the rainfall but provide poor time sampling of the rainfall signal.

The microwave technique uses the brightness temperature at 37 and 86 GHz from the Special Sensor Microwave/Imager instrument on board the Defense Meteorological Satellite Program (DMSP) satellite to define raining areas over water and land and uses the 86-GHz scattering signal to assign rain rate based on cloud model-microwave calculations. The microwave results are generally good for both individual swaths and monthly totals, except for a glaring underestimation of shallow, orographic rain systems over the southern coast of Japan. The IR techniques used are the GOES precipitation index of Arkin and Meisner and the convective-stratiform technique of Adler and Negri.

Initially the IR estimates are computed separately using hourly data from the Japanese Geostationary Meteorological Satellite. Calibration or adjustment factors are derived by dividing the microwave monthly estimate by a second IR estimate (made with the microwave sampling that simulates the observations from an IR radiometer on board the DMSP satellite). The spatial array of coefficients are then multiplied by the original IR monthly estimates (produced from all the hourly data) to produce the merged IR-Microwave monthly estimates. The results show that in areas where the base (microwave) technique performs well, that is, has a relatively small bias, the combined microwave-IR monthly total estimates have better error statistics than either the microwave or IR techniques individually.

Save