Satellite Rainfall Applications for Surface Hydrology

The Tropical Rainfall Measuring Mission (TRMM) Multi-satellite Precipitation Analysis (TMPA) is intended to provide a “best” estimate of quasi-global precipitation from the wide variety of modern satellite-borne precipitation-related sensors. Estimates are provided at relatively fine scales (0.25° × 0.25°, 3-h) in both real and post-real time to accommodate a wide range of researchers. However, the errors inherent in the finest scale estimates are large. The most successful use of the TMPA data is when the analysis takes advantage of the fine-scale data to create time/space averages appropriate to the user’s application. We review the conceptual basis for the TMPA, summarize the processing sequence, and focus on two new activities. First, a recent upgrade for the real-time version incorporates several additional satellite data sources and employs monthly climatological adjustments to approximate the bias characteristics of the research quality post-real-time product. Second, an upgrade for the research quality post-real-time TMPA from Versions 6 to 7 (in beta test at press time) is designed to provide a variety of improvements that increase the list of input data sets and correct several issues. Future enhancements for the TMPA will include improved error estimation, extension to higher latitudes, and a shift to a Lagrangian time interpolation scheme.

The CMORPH technique was developed to synergize the most desirable aspects of passive microwave (high quality) and infrared (spatial and temporal resolution) data. CMORPH is a global (in longitude; 60°N–60°S) high-resolution (∼0.10° latitude/longitude, 1/2-hourly) precipitation analysis technique that uses motion vectors derived from half-hourly geostationary satellite IR imagery to propagate precipitation estimates derived from passive microwave data. Multi-hour precipitation totals derived via the CMORPH methodology are an improvement over both simple averaging of all available microwave-derived precipitation estimates and over other merging techniques that blend microwave and infrared information but which derive estimates of precipitation directly from infrared data when passive microwave data are not available.

The Self-Calibrating Multivariate Precipitation Retrieval (SCaMPR) is an algorithm for retrieving rainfall rates using visible (VIS)/infrared (IR) and microwave-frequency data from Earth-orbiting satellites. Rainfall rates derived from microwave-frequency data are used as a calibration target for an algorithm framework that both selects the optimal VIS/IR predictors and determines their optimal calibration coefficients in real time. This algorithm is highly flexible and its short data latency makes it well-suited for rapidly-changing heavy rainfall situations that trigger flash flooding.

The need for frequent observations of precipitation is critical to many hydrological applications. The recently developed high resolution satellite-based precipitation algorithms that generate precipitation estimates at sub-daily scale provide a great potential for such purpose. This chapter describes the concept of developing high resolution Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks-Cloud Classification System (PERSIANN-CCS). Evaluation of PERSIANN-CCS precipitation is demonstrated through the extreme precipitation events from two hurricanes: Ernesto in 2006 and Katrina in 2005. Finally, the global near real-time precipitation data service through the UNESCO G-WADI data server is introduced. The query functions for viewing and accessing the data are included in the chapter.

The retrieval of satellite rainfall estimates from multi-platform Earth observations has received much attention over the last decade. The Passive Microwave – InfraRed algorithm, developed at the University of Birmingham, has been operating in a quasi-operational mode since 2002. The algorithm combines the temporally-rich information from the infrared geostationary observations with the more quantitative, but less frequent, rainfall information from the passive microwave polar-orbiting satellites. Co-located infrared and passive microwave information is entered into a database which is used to generate the relationship between the surface rainfall and infrared cloud top temperatures at a centred-weighted 5舁×舁5 scale. The technique produces rainfall estimates at a temporal resolution of 30 min and a spatial resolution of 0.1舁×舁0.1: the user can then aggregate these results to suit their requirements.

In this chapter, we discuss the basic workings of the NRL-Blend high-resolution precipitation product, followed by a validation experiment. We employ satellite omissions to the existing (late 2008) constellation of low Earth orbiting satellite platforms to examine the impact of several proxy Global Precipitation Mission (GPM) satellite constellation configurations when used to initialize land surface models (LSM). The emphasis is on how high resolution precipitation products such as the NRL-Blend are affected by such factors as sensor type (conical or across-track scanning) and nodal crossing time, using a collection of GPM proxy datasets gathered over the continental United States. We present results which examine how soil moisture states simulated by the two state-of-the-art land surface models are impacted when forced with the various precipitation datasets, each corresponding to a different proxy GPM constellation configuration.

GSMaP (Global Satellite Mapping of Precipitation) is a project aiming (1) to produce high-precision and high-resolution global precipitation maps using satellite-borne microwave radiometer data, (2) to develop reliable microwave radiometer algorithms, and (3) to establish precipitation map techniques using multi-satellite data for the coming GPM era. The GSMaP_MVK system uses a Kalman filter model to estimate precipitation rate at each 0.1° with 1-h resolution on a global basis. The input data sets are precipitation rates retrieved from the microwave radiometers and infrared images to compute the moving vector fields. Based on the moving vector fields calculated from successive IR images, precipitation fields are propagated and refined on the Kalman filter model, which uses the relationship between infrared brightness temperature and surface precipitation rate. This Kalman filter – based method shows better performance than the moving vector – only method, and the GSMaP_MVK system shows a comparable score compared with other high-resolution precipitation systems.

Abstract For continual refinement of error models and their promotion in prototyping satellite-based hydrologic monitoring systems, a practical user guide that readers can refer to, is useful. In this chapter, we provide our readers with one such practical guide on a space-time stochastic error model called SREM2D (A Two Dimensional Satellite Rainfall Error Model) developed by Hossain and Anagnostou (IEEE Transactions on Remote Sensing and Geosciences, 44(6), pp. 1511–1522, 2006). Our guide first provides an overview of the philosophy behind SREM2D and emphasizes the need to flexibly interpret the error model as a collection of modifiable concepts always under refinement rather than a final tool. Users are encouraged to verify that the complexity and assumptions of error modeling are compatible with the intended application. The current limitations on the use of the error model as well as the various data quality control issues that need to be addressed prior to error modeling are also highlighted. Our motivation behind the compilation of this practical guide is that readers will learn to apply SREM2D by recognizing the strengths and limitations simultaneously and thereby minimize any black-box or unrealistic applications for surface hydrology.

This chapter concerns the validation of high resolution (mostly 0.25°, daily and three-hourly) precipitation products over the United States. A synthesis of relevant studies is followed by comparisons of high resolution estimates based on satellites and models, with in situ ground validation data over the US. All the comparisons use multiple satellite estimates as well as model data (from the NCEP GFS). First, daily results are shown from the ongoing, web-based, real-time International Precipitation Working Group validation activity over the US. Next, validation data from 15 sub-daily gauges over Kansas and Oklahoma are used to assess the performance of three-hourly precipitation estimates, with attention to the distribution of precipitation. Finally, results from the comparison of several products against data collected from the North American Monsoon Experiment are given. Results show that existing high resolution products have great skill over many areas of the US, even at the three-hourly time-scale. Significant issues still exist over orography and results are seasonally dependent.

Two satellite rainfall estimation algorithms, CMORPH and TMPA, are evaluated over two mountainous regions at daily accumulation and spatial resolution 0.25°. The evaluated TMPA products are TRMM-3B42 and TRMM-3B42RT. The first of the two validations region is located over the Ethiopian highlands in the Horn of Africa. The second is located over the highlands of Columbia in South America. Both sites are characterized by a very complex terrain. Relatively dense station networks over the two sites are used to validate the satellite products. The correlation coefficients between the reference gauge data and the satellite products were found to be low. Besides, the products underestimate both the occurrence and amount of rainfall over both validation sites. These were attributed, at least partly, to orographic warm rain process over the two regions. The performance over Colombia was better compared to that for Ethiopia. And CMORPH has exhibited better performance as compared to the two TRMM products.

Evaluation of satellite rainfall products was conducted using ground-based daily rainfall measurements at 22 locations within a grid of 5×5 km collected during summer monsoon 2008 in a very complex terrain and humid tropical region in Ethiopia. Two high-resolution satellite rainfall products, namely, PERSIANN-CCS available at 1-h and 0.04° resolution, and CMORPH available at 30-min and 0.08° resolution. Both remotely-sensed products underestimated heavy events by about 50%, and so caution must be exercised when using CMORPH and PERSIANN-CCS as input for flood forecasting, as this could underestimate large flood events. The underestimation in monthly total rainfall was significant (32% for CMORPH, 49% for PERSIANN-CCS), and this error level needs to be acknowledged in applications that require monthly analyses. PERSIANN-CCS failed to detect half of the light events, and consistently those under 1.6 mm/day, indicating clearly that PERSIANN-CCS has difficulty detecting light rainfall events in complex terrain.

The study presented in this chapter evaluates the use of satellite rainfall for flood prediction applications in complex terrain basins. It focuses on a major flood event that occurred in October 1996 in a complex terrain basin of the northeastern region of Italy. A satellite rainfall error model is calibrated and used to generate rainfall ensembles based on two different satellite products and spatio-temporal resolutions. The generated ensembles are propagated through a distributed hydrologic model to simulate the hydrologic response. The resulted hydrographs are compared against the hydrograph obtained by using high-resolution radar-rainfall as the “reference” rainfall input. The error propagation of rainfall to stream runoff is evaluated for a number of basin scales that range from 100 to 1200 km². The results from this study show that (i) use of satellite-rainfall for flood prediction depends strongly on the scale of application (catchment area) and the satellite product resolution, (ii) different satellite products perform differently in terms of hydrologic error propagation and (iii) the propagation of error depends on the basin size; for example, this study shows that small watersheds (< 400 km²) exhibit a higher ability in dampening the error from rainfall-to-runoff than larger size watersheds.

Satellite-based precipitation retrieval techniques and algorithms have been developed to estimate precipitation from satellite observation. The realistic characterization of uncertainty in satellite precipitation estimate and the corresponding uncertain hydrologic response can better aid water resources managers in their decision making. In this study, the standard error of satellite-based PERSIANN-CCS rainfall estimates conditioning on the assumed true field (i.e. radar rainfall) is obtained according to a multivariate function considering the spatial and temporal scales. Accepting the multiplicative nature of this error, the Monte Carlo simulation is used to generate the ensemble of precipitation and propagate them into a conceptual hydrologic model to investigate the impact of input error on streamflow simulation. The statistical assessment of the results through probabilistic measures explores the more in-depth quality and reliability of the hydrologic response resulted from input error characterization.

To offer a cost-effective solution to the ultimate challenge of building flood alert systems for the data-sparse regions of the world, this chapter describes a modular-structured Global Flood Monitoring (GFM) framework that incorporates satellite-based near real-time rainfall flux into a cost-effective hydrological model for flood modeling quasi-globally. This framework includes four major components: TRMM-based real-time precipitation, a global land surface database, a distributed hydrological model, and an open-access web interface. Retrospective simulations for 1998–2006 demonstrate that the GFM performs consistently at catchment levels. The interactive GFM website shows close-up maps of the flood risks overlaid on topography/population or integrated with the Google-Earth visualization tool. One additional capability, which extends forecast lead-time by assimilating QPF into the GFM, also will be implemented in the future.

Global precipitation, temperature, soil moisture, vegetation health, and lake water heights data sets are several operational data sets continuously monitored by crop analysts from the Foreign Agricultural Service (FAS) of USDA to identify global weather and vegetation health anomalies that may affect national crop yield and production in foreign countries. Three relatively new satellite-derived precipitation data sets were recently introduced into the USDA/FAS crop monitoring system, along with two new soil moisture products that utilize passive microwave (PMW). Comparison results indicate no operational global precipitation data set is correct at all times for all geographic areas and those that combine station gauge (SG), polar-orbiting passive microwave and geo-stationary infrared (IR) data tend to perform better.

A multi-institutional partnership, the US Agency for International Development’s Famine Early Warning System Network (FEWS NET) provides routine monitoring of climatic, agricultural, market, and socioeconomic conditions in over 20 countries. FEWS NET supports and informs disaster relief decisions that impact millions of people and involve billions of dollars. In this chapter, we focus on some of FEWS NET’s hydrologic monitoring tools, with a specific emphasis on combining “low frequency” and “high frequency” assessment tools. Low frequency assessment tools, tied to water and food balance estimates, enable us to evaluate and map long-term tendencies in food security. High frequency assessments are supported by agrohydrologic models driven by satellite rainfall estimates, such as the Water Requirement Satisfaction Index (WRSI). Focusing on eastern Africa, we suggest that both these high and low frequency approaches are necessary to capture the interaction of slow variations in vulnerability and the relatively rapid onset of climatic shocks.