(see detailed comparisons) (High resolution rainfall maps)

rainfall estimates

Methods to estimate rainfall

Measuring the amount of rainfall reaching the ground is not as simple as it might seem. Many techniques have been developed but it is still difficult to make accurate estimates by any one of them. Rain gauge are the traditional way to measure the rainfall at a point. Quite apart from the obvious impossibility of deploying enough rain gauges over deserts, mountains and oceans, the adequacy of point measurements to represent a larger area is a major issue. Since satellite data became available, techniques of relating the area of cold clouds to surface rainfall have had some success (Joyce and Arkin 1997). However, infrared techniques suffer from inevitable underestimation of warm rain, and frequent false alarms for certain anvil and thick cirrus clouds with cold IR brightness temperatures.

Microwave sensors and precipitation radar are the other tools that have been used increasingly in recent years, with the potential to improve precipitation estimates from the surface and from space. Unlike IR, these techniques directly sense precipitation particles rather than cloud tops. However, significant difficulties remain. ?Rainfall retrieved from the TRMM precipitation radar suffers from uncertain attenuation correction, problems over complex terrain and the limit of minimum detectable signal (Iguchi et al. 2000). Microwave retrievals over the ocean are thought to rival radar retrievals for accuracy, but retrievals over land are compromised because of variations of the surface emissivity (Spencer et al 1989; Kummerow et al 2001).

As shown in the figure, rainfall estimated from different measuring techniques have some similarities globally, but there are also differences among them. For example, over central Africa, techniques based on microwave data estimate much heavier rainfall than those using precipitation radar or rain gauge data. Microwave estimates are also unrealistic over the Tibetan plateau. The differences among the rainfall estimates from different techniques vary seasonally as well as regionally (see detailed comparisons). More detailed comparisons can be found at climate rainfall data center.

Rainfall estimates in different types of precipitation

Because different rainfall retrieval algorithms are using different raw observations, and relying on different physical detection methods , they perform differently for various precipitation types. For example, it is difficult to detect some weak warm rainfall over ocean by the TRMM PR because radar echoes from them may be less than the PRí»s minimum detectable signal. However, they may be detectable with the emission signal at low frequency microwave channels from TMI. This may help to explain the underestimation of the total rainfall over the east Pacific ITCZ by the PR (Berg et al 2006). Comparing the rainfall from different types of precipitation systems and different seasons retrieved by different techniques can help to validate the algorithms (see a discussion regarding the TMI land algorithm), also may help to improve understanding of the precipitation systems themselves. (Detail - see rainfall from different precipitation systems)

Rainfall from Rain Gauges The rain gauge rainfall product used in this website is the GPCC Monitoring Product. The GPCC Monitoring Product of monthly precipitation is based on SYNOP and monthly CLIMAT reports received near-realtime via GTS from about 7,500 stations. Origin of the products can be found here.

Rainfall from Microwave Sensors The microwave rainfall product used in this website is the 2A12 product (Kummerow et al 2001) from the TRMM TMI. The gridded monthly 2A12 product is called 3A12. The rain retrieval algorithm includes distinctly different methods for land and ocean. Over ocean, the algorithm derives the surface rainfall by using the TMI measured microwave radiances at 10,19,22, 37, and 85 GHz and a database of hydrometeor profiles from cloud resolving model simulations. Over land, it derives rainfall mainly from an empirical relationship based on the ice scattering signature at 85 GHz after a rain/no rain determination using a complex surface screening method (Ferraro et al. 1998).

Rainfall from Space-borne Radar Rainfall from Space-borne radar is not a simple Z-R (radar reflectivity vs. rainfall rate) relationship because it is well known that there is no universal Z-R relationship for different precipitation types. Also, the PR operates at a highly attenuating frequency, so the algorithm for near-surface rain rate uses a complex method to attempt a reasonable attenuation correction.? The TRMM PR rain retrieval algorithm is described by Iguchi et al (2000). The PR rainfall product used here is 2A25 and the gridded montly product is 3A25.

Rainfall from Infrared images The rainfall product from infrared observations used here is from the Global Precipitation Index (GPI). The algorithm was developed by Joyce and Arkin (1997). It is known that GPI overestimates the rainfall from the large amount of cold clouds over the West Pacific Ocean (Liu et al 2007), and underestimates certain kinds of rainfall, but may do quite well in certain regimes.

ChallengesOne of the main challenges we are facing currently, which will become especially important in the GPM era, is to estimate the amount of precipitation falling as snow, especially over snow covered land. This is especially difficult for microwave methods because it is hard to separate the scattering and emission signature of the snow in the air from the signature of snow on the ground. For space-borne radar, it is difficult because precipitation in the form of snow is often below the minimum detectible signal. ?This requires a radar with a higher sensitivity.

Contact Us | © Texas A&M University at Corpus Christi

l