This issue summarizes the various research outcomes obtained through the project known as “Asian Precipitation – Highly Resolved Observational Data Integration Towards Evaluation of Water Resources” (or “APHRODITE’s water resources” for short). Nowadays, most environmental and social development issues are linked with climate change, i.e., the global warming issue.Hence, high-resolution climate models are being developed and their outputs are being used to assess the impact of climate change on disaster risk, agricultural production and local hydrological resources. However, there has been no reliable observed dataset over Asia to verify climate model simulations and/or downscale their outputs to local hydrological application studies.

Hence we started the APHRODITE project in 1996. Using various connections of the members at the Research Institute for Humanity and Nature (RIHN) and Meteorological Research Institute (MRI) of the Japan Meteorological Agency (JMA), we negotiated with national meteorological and hydrological services (NMHs) of many countries to obtain their observed daily precipitation data. Finally, we got more than 10,000 station data figures over Asia as shown in Fig. 1. Then we reformatted the data, and applied a quality control (QC) scheme. We developed an interpolation algorithm to represent the orographic precipitation pattern more realistically. Details of the algorithm are described in Yatagai et al. (2009, 2012). I am happy to include a paper which described the QC scheme in this volume. It will provide suggestions to the user of the product as well as beginners at data processing.

The papers submitted to this volume appear by region, since the orientation of the application studies differ from among regions. First, the product for Japan, APHRO_JP was released in 0.05 degree resolution (Kamiguchi et al., 2010) and it has already been used in studies on extreme events in relation to the global warming issue. A paper written by a specialist on statistical downscaling for agricultural application is included.

South and Southeast Asia are regions where we see large differences in estimates of quantitative rainfall among the products. Hence, a comparison study among the products and a study for developing a new satellite-gauge-based algorithm are shown. In addition, hydrological and agricultural application studies are included in this volume.

In the Middle East – the APHRODITE project was initiated by a feasibility study to develop a rain-gauge-based product for the Middle East. I am convinced that a quantitative and high-spatial resolution product would be beneficial to understanding in paleo-climate studies by combining climate model simulations and proxy data. In this issue, we show how representative the Fertile Crescent is, using a high-resolution model and our precipitation estimates. In northern Eurasia, unlike Monsoon Asia, differences among the products are not that large. On the contrary, cold regions including the high-altitude region in Monsoon Asia and the Middle East do have an important issue with regard to global warming – namely, the existence of snow and ice. We developed a temperature dataset and added discrimination between snowfall and rainfall, including these in this paper together with their application in a run-off study.

On behalf of the project members, I express the sincere hope that our APHRODITE product will be used in many climate change studies. We anticipate feedback from the users.

The authors constructed a dataset of historical extreme precipitation indices in Japan named APHRO_JP_EX. This dataset contains more than 40 annual indices derived from daily precipitation. Although the spatial coverage is limited to the land of Japan, the uniqueness of these data with their long period (1900-2009) and high resolution (0.05 x 0.05 degrees) is useful not only for climate research but also for application to studies such as risk assessment of hydrological disasters. Using this dataset, we analyzed trends and annual variations of these indices to quantify the effects of global warming on local extreme precipitation in Japan.

The number of wet days has significantly decreased all over Japan, especially in the northeastern part, where about 20 wet days per year have been lost during the past century. The annual mean precipitation has slightly decreased in most regions. Looking at the regional average of all of Japan, increased heavy precipitation intensity is distinct. The local changes in heavy precipitation are larger in the western part; however, they are not significant in many areas due to large annual variations. When the former and the latter halves of the 20th century are compared, increases in the annual variations of heavy precipitation indices are remarkable in the latter period. While indices of wet spells and dry spells show clear regionality in their trends, wet-spell indices have been decreasing on the Japan Sea side and dry-spell indices have been increasing on the Pacific Ocean side, though both results indicate a progression of meteorological dryness.

After the beginning of the 20th century, in which surface temperatures were rising, precipitation in Japan has become more extreme overall; however, we found regional differences in these signals.

In this study, we compared the daily precipitation indices derived from two types of statistical down scaling simulations in Japan. One simulation was conducted by applying a cumulative distribution func tion-based downscaling method (CDFDM) to a state-of-the-art gridded daily precipitation dataset for Japan, APHRO_JP (Kamiguchi et al., 2010). The other simulation was conducted by applying the same downscaling method to a simple dataset, namely a set of linearly interpolated rain gauge data. We then highlighted the added values for the coarse-resolution general circulation model (GCM) outputs achieved by the incorporation of the APHRO_JP dataset into the statistical downscaling simulations, relative to the simulations using the simple dataset. The evaluated indices were mean precipitation, number of wet days, 90th percentile of daily precipitation, and maximum number of consecutive dry days. A comparison of the two approaches shows that the relative improvements achieved using the APHRO_JP dataset vs. the simple dataset are most pronounced for mean precipitation in mountainous areas, precipitation frequency on the leeward side of mountains, and intensity and frequency of heavy precipitation. These relative improvements over the regional climate change scenarios derived from the CDFDM demonstrate a significant benefit from combining advanced gridded observation datasets with statistical downscaling methods. On a regional scale, the bias of the daily precipitation indices for the simple dataset, relative to the APHRO_JP dataset, is occasionally comparable in amplitude to the projected change. This finding suggests the importance of higher-quality gridded observation datasets in assessing climate impacts in various fields, particularly, hydrological regimes, irrigation planning, and risk assessments of water-related disasters.

The authors conducted an intercomparison of the elevation dependency of precipitation over the Asian summer monsoon region among 16 gridded precipitation datasets that consisted of rain-gauge-based (GAUs), satellite-based (SATs), and gauge-satellite combination datasets (MRGs). The elevation dependencies of the mean terrestrial precipitation in the gridded precipitation datasets were qualitatively consistent. The SATs were generally less than the GAUs at low and middle altitudes, but comparable at high altitudes. The scatter of the differences among the GAUs was smaller than that of the SATs. The elevation dependency of the mean intensity of the heavy terrestrial precipitation, defined as the 95th percentile of daily precipitation, was similar to that of the mean precipitation. The scatter of the differences in the elevation dependency among the SATs was greater than that among the GAUs. These results imply that differences in spatial interpolation methods have little impact on the differences among GAUs but that differences in the rainfall retrieval algorithm have a large impact on the quantitative differences among SATs.

In this study we analyzed the rate of change of forest coverage in Savannakhet Province, in central south Lao PDR (People’s Democrotic Republic), and the causes of this change. In considering reasons for the transition of forest coverage, we took into account human activities such as large-scale development, expansion of agricultural land and shifting cultivation as well as effects of changes in rainfall. To survey changes in land coverage, including forests, we used AVNIR-2 images with high spatial resolution and MODIS images with high time resolution. To supplement the insufficient network of rainfall gauges in the study area, data from the APHRODITE Project were used for analysis. As a result of this analysis, we estimated that the total forest coverage decreased by nearly 40,000 ha in Savannakhet Province, in the ten years between 2000 and 2009. Deforestation occurred both in the open forest in the plains (dry dipterocarp forest) and in isolated masses of dense forest (mixed deciduous forest) as the result of expansion of agricultural areas, mainly in the western part of the province. We could find no long-term trend in rainfall, but recently there has been a tendency of rainfall to decline where deforestation has occurred.

The present study focuses on estimating rainfall (daily accumulated) using gauge and satellite observations over land and ocean regions of South Asia (30°S-50°N, 40°E-120°E) at a 0.25°× 0.25° spatial resolution for the period 2007-2010 (four years). The study utilizes observations from rain gauges, the Special Sensor Microwave/Imager (SSM/I) onboard the Defense Meteorological Satellite Program (DMSP), Precipitation Radar (PR) onboard the Tropical Rainfall Measuring Mission (TRMM) and geo-stationary satellite Meteosat from Eumetsat. The present study makes use of rainfall estimates by synergistic use of multi-satellite sensors using Meteosat Infrared and Water Vapor absorption channels and PR observations (Mishra et al., 2009b, 2010) and SSM/I-derived microwave estimates using a regional scattering index developed by Mishra et al., (2009a). The pixels over the land portion of the study area are filled with available rain gauge observations over the southern part (around 14°N and 78°E) of the study area, which has a dense network of Automatic Weather Station (AWS) rain gauges operated by the Indian Space Research Organization (ISRO). Pixels over the other part of the study area are filled with available microwave observations first, but if microwave observations are unavailable then these pixels are filled with microwave-calibrated infrared observations over the land and oceanic regions of the study area. The precipitation estimates from the present approach are validated against rain gauge observations and other available standard rainfall products like TRMM-3B42V6 and Global Precipitation Climatology Project (GPCP) version 2.1. The validation results show that the present approach of precipitation estimation is able to estimate rainfall with high accuracy.

We investigated the simulation performance of two versions of the atmospheric general circulation model (MRI-AGCM3) in reproducing precipitation climatology over the Middle East in comparison with high-resolution gridded precipitation datasets. Both models overestimated precipitation amounts compared with the observations over this region. The new version, MRI-AGCM3.2, overestimated precipitation much more than the old version, MRI-AGCM3.1, particularly in the winter and spring rainy seasons, but the new version was better than the old version in the summer dry season. We also investigated resolution dependency. The higher the resolution the more the precipitation indicated, but the higher resolution version showed better performance in reproducing the spatial distribution of annual and seasonal mean precipitation, probably due to better representation of orography. These results indicate difficulty in climate model development in terms of regional climate reproducibility.

The author presents a precipitation climatology and its trend over the mountainous areas of the Middle East using the dense network rain-gauge observation data gathered by the APHRODITE project. Coastal and orographically induced precipitation zones are found in semi-arid regions, and those patterns are consistent with those observed by the Precipitation Radar of the Tropical Rainfall Measuring Mission.

The author then uses 34-year monthly precipitation data over Turkey, Israel and Iran to show seasonal precipitation trends. There is a strong and significant decreasing trend over Turkey and Israel in winter, especially in January. Trends over Turkey differ from month to month, while general trends over Iran differ little from season to season. On the other hand, Iran has orographically characterized trends. Trends are affected by the mountain ridges and slopes of the Taurus, Anatolian, Zagros, Erboz and other ranges.

Then she compares the trends at thirteen high stations paired with low stations under the same synoptic conditions for the winter season. Most station pairs showed decreasing trends in the ratio of precipitation at the high stations to that at the low stations. Exceptional cases, which showed increasing trends in the ratio, were observed in the easternmost part of the Taurus Mountains (north of Syria) and in the northwestern part of Turkey.

In order to achieve risk management of water-related disasters such as droughts and floods in Iran, it is essential to obtain knowledge of the exact quantities of spatial and temporal precipitation over the long term. In this regard, this paper presents the results of monitoring precipitation on monthly, seasonal and annual timescales during the period of 1986-2006 (21 years) over Iran using a product of 0.5-degree-gridded data sets provided by the Asian Precipitation-Highly-Resolved Observational Data Integration (APHRODITE) Towards Evaluation of the Water Resources project, namely APHRO_ME_V003R1. Spatial distribution of mean seasonal and annual precipitation shows two rainfall peaks that extend from the northwest to the southeast along the Zagros Mountains and along the Caspian Sea and Alborz Mountains in the northern part of the country.

With respect to a seasonal spatial analysis of rainfall over Iran, it is concluded that in winter, DJF (December- January- February), and spring, MAM (March-April-May), most precipitation occurs in the west, northwest and southwest along the Zagros Mountains, but in summer. JJA (Jun-July-August), and autumn, SON (September-October-November), most precipitation occurs along the southern Caspian Sea in the north.

Annual and seasonal drought condition using APHRODITE data have showed that mean annual precipitation amount were low (drought condition) over Iran in 1989, 1990, 1991, 1992, 1996, 1998, 1999, 2000, 2001 and they are in accordance with Yazdani et al (2011)'s study about drought analysis over Iran.

We created a daily mean gridded temperature dataset of monsoon Asia (15˚S-55˚N, 60˚E-155˚E) for the period of 1973-2007, with a 0.50 x 0.50 degree grid. We analyzed this dataset based on station observations collected and a quality control and interpolation system developed through the activities of the Asian Precipitation – Highly Resolved Observational Data Integration Towards Evaluation of Water Resources (APHRODITE) project (Yatagai et al., 2009, 2012). The number of stations is up to 1.5-3 times the number of stations based on the Global Telecommunication System (GTS), which have been used to obtain other gridded temperature products. The monthly means and climatology of our product are comparable to the monthly means of those products. This is the only product in Asia that has high resolution both temporally and spatially as the APHRODITE precipitation product is known for.

The ability to discriminate between rain and snow is added to the APHRODITE daily precipitation product by using daily mean temperature and relative humidity (RH). Relative humidity is derived from a reanalysis product, because RH observation data are collected insufficiently for interpolation. We found use of the temperature product of this study and the RH derived from the reanalysis product to be adequate for determining whether precipitation was rain or snow. Our estimated solid precipitation amount using rain/snow discrimination for late fall to early spring (October to March) is consistent with satellite observations.

This dataset is available on the APHRODITE website. The combination of daily mean temperature, precipitation and rain/snow information in this high- resolution gridded format would be useful as input to river-flow models, crop models and many other situations where water resources must be estimated.

Recent studies show that dissolved iron is a factor essential to the biological productivity of the Sea of Okhotsk. It is also highly probable that wetlands in the Amur River basin play an important role in providing dissolved iron to the Sea of Okhotsk. During the last century, large areas of wetlands within the Amur River catchment have been cultivated, which, in turn, may have had a great impact on the production of dissolved iron. To assess the impact of this conversion of land cover on the production of dissolved iron – and thereby primary production in the Sea of Okhotsk – we have been constructed a numerical model to simulate dissolved iron production and transport in the Amur River.

Though the accuracy of hydrological models heavily depends upon the precipitation input, there has been no observation-based gridded precipitation dataset covering a continental-scale region with high spatio-temporal resolution. Recently, however, datasets such as APHRODITE have been released. Thus, in this article we evaluate how hydrological simulation results may be improved by these datasets. By comparing the results from the APHRODITE precipitation dataset with those from the NCEP2 precipitation dataset, we show that the precipitation amount from NCEP2 around the Amur River basin is overestimated. Moreover, the closeness of the fit evaluated by the Nash-Sutcliffe criteria indicates that the results from APHRODITE are significantly better than those from NCEP2. The results indicate that the precipitation of the APHRODITE dataset for the Amur River basin has good accuracy.

An automated quality control (QC) system was developed for detecting errors in daily rain-gauge data. This QC system was basically designed to detect obvious errors such as clerical errors automatically and objectively. A total of 14 components have been developed for daily rain-gauge data, but many of the components can easily be applied to other weather elements such as temperature with appropriate changes in parameters. The authors applied the QC system to a continental-scale rain-gauge network. The results are illustrated herein with examples with discussions of possible causes for each kind of error. While most of the errors were found in data which had not been subjected to thorough QC, many basic errors were also found in this widely used global/regional dataset. The results given by newly proposed QC components, which use multiple data records at the same station but from different data sources, show that such comparison tests are important and work well for detecting errors such as unit misconversions.