Heavy rainfall prediction applying satellite‐based cloud data assimilation over land

Abstract

To optimize flood management, it is crucial to determine whether rain will fall within a river basin. This requires very fine precision in prediction of rainfall areas. Cloud data assimilation has great potential to improve the prediction of precipitation area because it can directly obtain information on locations of rain systems. Clouds can be observed globally by satellite‐based microwave remote sensing. Microwave observation also includes information of latent heat and water vapor associated with cloud amount, which enables the assimilation of not only cloud itself but also the cloud‐affected atmosphere. However, it is difficult to observe clouds over land using satellite microwave remote sensing, because their emissivity is much lower than that of the land surface. To overcome this challenge, we need appropriate representation of heterogeneous land emissivity. We developed a coupled atmosphere and land data assimilation system with the Weather Research and Forecasting Model (CALDAS‐WRF), which can assimilate soil moisture, vertically integrated cloud water content over land, and heat and moisture within clouds simultaneously. We applied this system to heavy rain events in Japan. Results show that the system effectively assimilated cloud signals and produced very accurate cloud and precipitation distributions. The system also accurately formed a consistent atmospheric field around the cloud. Precipitation intensity was also substantially improved by appropriately representing the local atmospheric field. Furthermore, combination of the method and operationally analyzed dynamical and moisture fields improved prediction of precipitation duration. The results demonstrate the method's promise in dramatically improving predictions of heavy rain and consequent flooding.

Representing Cloud Water Content of Extensive Cloud Systems Over Land Using Satellite‐Based Passive Microwave Observations With a Coupled Land and Atmosphere Assimilation Method

“…The number of ensemble members for land data assimilation was set to 50, which was sufficient for threelayer, one-dimensional assimilation. Thus, we did not inflate the error covariance as described by Seto et al (2016). The observational error covariance matrix was empirically set to the diagonal matrix, 2:0 0 0 3:0 .…”

“…(2) According to the vertical distribution of CLWC, cloud area was defined as the threshold of CLWC exceeding 0.01 g/kg (Korolev et al, 2007;Seto et al, 2016).…”

“…(3) Air within the cloud area became saturated, and the potential temperature was modified according to the CLWC amount (Seto et al, 2016). (4) Higher-frequency TB at TOA was calculated using a four-stream fast model.…”

“…This regeneration began after only one or two time steps (about 1 min) of integration and thus had little influence on subsequent cloud and rain representations. The boundaries of vertically integrated CLWC data used for assimilation were set to the same values used by Seto et al (2016), of 0-15 kg/m 2 . CWC-only assimilation cannot represent the physical environment of cloud formation, that is, a saturated and unstable environment, which is typical of cloud regions and is a necessary condition for cloud development and maintenance.…”

“…They obtained good performance and effective all-sky assimilation with respect to representing large-scale temperature and humidity patterns but did not report the reproducibility of their cloud and rain system results. Seto et al (2016) overcame the difficulty of capturing cloud signals over land from satellite-based passive microwave brightness temperatures (TBs) by developing their Coupled Atmosphere and Land Data Assimilation System using the Weather Research and Forecasting Model (CALDAS-WRF). CALDAS-WRF estimates the optimum distribution of land states by assimilating lower-frequency microwave TBs and using a radiative transfer model (RTM) and a physical model for land.…”