Project work commenced by assembling the databases for the Chinese and Australian study sites. Initially for the Chinese site, a digital land-use map from 1993, and average annual of rainfall and potential evapotranspiration surfaces were developed. Having access to these data allowed the programming of a prototype modelling tool. This was developed using 'The Invisible Modelling Environment' (TIME) - a modelling framework developed by the Cooperative Research Centre for Catchment Hydrology. Having a bilingual working prototype meant allowed the visual and interactive illustration of a relevant 'modelling tool'; which formed the basis for direct communication with key stakeholders at the start-up meeting.
A workshop with senior policy makers involved in managing the middle reaches of the Yellow River was held in August, attended by the Chief Engineer, Bureau of the Upper and Middle Reaches of the Yellow River Committee, the Soil Conservation Bureau Section Leader, Yellow River Conservation Commission, and the Planning Section Leader, Bureau of the Upper and Middle Reaches of the Yellow River Committee). The managers were impressed by the prototype modelling tool, however they urged the project team to include sedimentation in the model outputs. Given the current slow and expensive Web access of their host organisations, the three key stakeholders strongly suggested that the most viable form of delivery was to develop a 'static' Web site (containing an example advertising the availability of the modelling tool), with a CD version of the dynamic modelling tool being sent to pre-determined relevant users, and all others who request the tool from the Yangling ACIAR project Web-site.
Meetings were also held with the Yellow River Conservation Commission in Zhengzhou. These introduced the project staff to potential high-level managers of the tool. Again stakeholders were interested in the outputs and were impressed by its usability. These key stakeholders also suggested that sedimentation be included in the tool and confirmed that a CD-based GIS modelling tool would be more widely used than a Web-based GIS modelling tool as internet access for the targeted users in China is currently too slow and expensive. These meetings provided an insight into some of the engineering data (specifically location, size and date of construction of earth dams and terraces) managed by the Yellow River Conservation Commission. Access to this data would need to be negotiated before modelling is undertaken of any sediment movement in the Coarse Sandy Hilly Region (CSHR).
Development of the required temporally varying databases for both sites is progressing well to date. For the Chinese site, particularly good progress has been made constructing the meteorological database. Monthly (rather than annual) meteorological data evenly covering the entire CSHR has been incorporated, allowing the water yield modelling to be driven at a monthly (not annual) time-step for the entire CSHR, which is important given snow-melt. Some precipitation (falling as snow in late November/December) will register as water yield the following spring thaw around March the following year. Monthly water yield data (not annual) will be used to calibrate and validate the model; having been acquired at the 39 sites that still record these data in the CSHR. Land-use data for 1983, 1987, 1993 and 1997, and static Geographic Information System (GIS) datasets, specifically a digital elevation model (DEM) and a soils map are all available for the study site. Quality control of these regional datasets is being undertaken. Staff members from Yangling have completed the equilibrium soil moisture data collection during two intensive field trips; the first was conducted before the summer rains, from 8th to 13th June, and the second was conducted from 6th to 15th October (after the summer rains). In both field trips, over thirty 5-m soils cores were measured. All cores were spatially clustered, and within one cluster soils from different land-covers were sampled.
For the Australian site, regional databases of meteorological, water yield and GIS data-layers (DEM, land-cover and soils) have been developed. More than 30 meteorological stations are in the Bureau of Meteorology (BoM)/Qld Department of Natural Resources and Mines (QDNRM) SILO 'Point Patched Database' where more than 95 per cent of the data are recorded at the BoM stations (and not the result of spatially interpolating isolated points by QDNRM). These data are daily from 1 January 1980 to the present. Water yield data have also been extracted for 25 stations from the New South Wales Department of Infrastructure Planning and Natural Resources (DIPNR) PINNEENA database, and are also daily from 1 January 1980 to the present. Currently, full radiometric corrections, including atmospheric correction and bidirectional reflectance distribution normalisation are being completed for the time series of regional remotely sensed data (specifically Advanced Very High Resolution Radiometer data) daily from April 1992 onwards by the CSIRO Earth Observation Centre.
In May, the project successfully applied for Associate Project status with the Cooperative Research Centre for Catchment Hydrology (CRC_CH), giving LWR1/2002/018 exposure to Australian stakeholders.

In Year 2, over 90% of the project team's (Chinese and Australian) time was spent performing quality control on the Chinese databases, and spatially interpolating the point based meteorological data. Five main databases have been checked and developed: (1) contour, spot height and river data used to develop a hydrologically correct Digital Elevation Model (DEM); (2) soils data used in vegetation suitability modelling; (3) time series (1980, 1986, 1993 and 1997) land-use data; (4) monthly meteorological database from Jan. 1980 to Dec. 2000 (21 years); and (5) monthly hydrological database, again from Jan. 1980 to Dec. 2000 (21 years). The importance of each is briefly discussed.

(1) Developing a hydrologically correct DEM: A previous DEM, created using a triangular irregular network, was not hydrologically correct (i.e., a stream network and associated contributing areas could not be accurately calculated). To generate a hydrologically correct DEM, we used the ANUDEM algorithm. However, the input data for ANUDEM (contours, spot heights and river networks) required much time identifying and correcting various errors. Additionally, effort was spent optimising key selected ANUDEM parameters to ensure highest quality output. Given the controlling influence of landform on all water (and soil) processes occurring in the Loess Plateau, having a hydrologically correct DEM was a high priority. The DEM was used in 4 fundamental ways: (A) defining the sub-catchments that comprise the study area; (B) defining the area contributing runoff to each hydrological station; (C) as a covariate from which to perform the spatial interpolation of the meteorological database; and (D) input to calculate solar radiation taking slopes and aspect into account.

(2) Correcting digital soils map: The soils database was carefully assessed for errors; this included ensuring that all digital map codes were identical to the paper map. We identified and corrected 1963 soil polygons that were incorrectly digitised. This improved the overall coherence of the digital data set, and meant that subsequent vegetation suitability modelling would be more reliable.

(3) Assessing the validity of the time series of land-use maps: Four land-use maps were available (1980, 1986, 1993 and 1997) and from these data we expected to be able to monitor the land-use change from 1980 to 1997 as a result of the re-vegetation program. Given that the maps came from different sources, we had to combine the different classes into one common classification system. Upon careful checking of the common class data we found that there was no logical explanation for the progression of land-uses for many individual pixels, and therefore we could not perform the monitoring on a spatial (per-pixel basis). For example, some individual pixels progressed from pasture to water to forest to pasture again during the four dates; this progression of land-use was not real, but simply due to the errors and differences in the maps used. Next, we assessed whether the four land-use maps could be compared on a per-catchment basis. However, the illogical progression of forest area eliminated this option (1993 was anomalously low). Use of high resolution remote sensing to re-label 1997 polygons was inspected, but rejected as 90% of the 96,000 polygons would require redigitising. The 1986 data was deemed by ISWC as the only dataset reliable enough to use.

(4) Checking, and spatially interpolating the monthly meteorological database: The meteorological data measured at the isolated stations in and around the CSHC required quality control checks prior to monthly surfaces being spatially interpolated (using ANUSPLIN). These quality control analyses were performed systematically using computer programs; as a result of this process, the input data was of much higher quality. The complex topography of the CSHC, however, meant that the spatial interpolation was more difficult than previously expected (based on experience from relatively flat Australia). We also needed to extend a program called SRAD program to deal with a DEM of the size needed for the study site. SRAD allows the calculation of net radiation in areas of complex topography (i.e., differing slopes and aspects are accounted for).

(5) Checking the monthly hydrological database: The systematic quality control of the hydrological database was performed with many errors successfully corrected. This task was much more difficult than expected; there was little metadata for this dataset and problems included units being incorrectly labelled and sometimes changed part way through the time series for a given station! The result of this effort means that the final datasets are of high quality to perform the model validation, which is critical to project success.

Detailed hydrologic understanding has been gained from analysis of experimental and point based data. In addition to re-vegetation programs, other engineering works - mainly sediment trapping dams and terraces - have been pursued in the Loess Plateau. The sediment trapping dams are usually small earth dams built across gullies that slow the movement of the water allowing the soil to drop-out of solution. Over time, and due to the high erosion rates, dams fill with soil creating 'new' horizontal farmland. Given that both the dams and terraces slow water, there may be greater opportunity for water to evaporate back into the atmosphere or infiltrate into the soil. Both of these outcomes may change the pattern of the flow duration curve (FDC), as we are primarily assessing the impact of the re-vegetation change on regional hydrology we also wanted to have an idea of the relative magnitude of the impact of engineering programs on the FDC.

In Year 3, over 90% of the project team's (Chinese and Australian) time was spent completing the Chinese study, focused on the Coarse Sandy Hilly Catchments (CSHC) of the Loess Plateau, which drain the main south flowing branch of the Yellow River. Specifically we finished: (1) calibrating a global steady-state land-use / water-balance framework to local conditions; (2) spatially modelled the suitability of 38 species of trees and shrubs used in the re-vegetation program; (3) determining the location of target and priority areas for undertaking re-vegetation activities; (4) developing the bi-lingual decision support system (DSS) called ReVegIH (Re-Vegetation Impacts on Hydrology), including writing a bilingual user-guide; and (5) fully documenting the processing of the datasets underpinning ReVegIH. The importance of each is briefly discussed.

(1) Locally calibrating the steady-state land-use / water-balance framework: To facilitate scenario planning of the impact of implementing China's re-vegetation program on water resources in the study area we had to ensure that a previously implemented land-use / water-balance framework was locally calibrated to local conditions. The framework was previously developed using a global dataset, and when used to estimate run-off at our study site, results were poor when validated against real streamflow data. To improve this we regionalised the framework parameters using the database of 36 hydrology stations that went through a quality control process in Year 2. We then had to ensure we had a suitable relationship to extend this relationship past the geographic area of the 36 stations to cover the entire CSHC. This was performed using meteorological datasets developed in Year 2. The result of this was a locally calibrated land-use / water-balance framework that could be used to illustrate the steady-state impact on water resources following implementation of the re-vegetation program, i.e., for scenario planning.

(2) Spatially modelling vegetation suitability: To ensure that ReVegIH was useful to managers involved with implementing the re-vegetation in the CSHC, we developed suitability maps for 38 species covering the site. This was implemented considering 5 variables that govern plant suitability (precipitation, air temperature, landform, and soil pH and nitrogen). This mapping heavily utilised the spatial datasets developed in Year 2 of the project.

(3) Determining target and priority areas: As the vegetation suitability mapping results were fairly general, we developed a series of rules to define target and priority areas for trees, shrubs and grasses in the landscape. Target areas are the locations where optimal growth of different vegetation types occurs, whereas the priority areas are those target areas that have the highest potential to reduce soil erosion. Priority areas are adjacent to, and lower than the steep slopes and gullies that characterise the CSHC. By re-vegetating the priority areas first, this focuses the initial planting to a smaller area (thereby minimising the reduction in run-off) that is expected to most effectively reduce the amount of soil entering the stream network.

(4) Completing the bi-lingual decision support system called ReVegIH: We developed a bilingual computer simulation tool that could be distributed to environmental managers in the CSHC free of charge on one CD. The tool, called ReVegIH provides a means for users to: (a) determine where priority (and target) re-vegetation activities should be undertaken; (b) ascertain what species are suitable for a specific location; and (c) simulate the related hydrological impact. The spatial resolution of the first two functions is provided at 100 m, while the third is at the catchment (or county) level for the 113,000 km2 CSHC. ReVegIH has an easy-to-use bilingual interface that includes some basic GIS functions. All electronic help is bilingual, as is the user guide which fully documents ReVegIH and also provides answers to frequently asked questions.

(5) Documenting the underpinning datasets: The systematic quality control and development of the databases underpinning ReVegIH was performed with many errors successfully corrected and best-practice methods used. The project team ensured that the datasets and underlying processing was fully documented, thereby recording the lineage of the data and ensuring that all processing could be performed independently by all organisations involved in the project. Completing this component of the project successfully means that the final datasets are of high quality to perform the model validation underpinning ReVegIH (which is critical to project success), and that both partners have the skills to generate these datasets.

In Year 4, much of the project team's (Chinese and Australian) time was spent finalising the decision support tool for the Chinese study (focused on the Coarse Sandy Hilly Catchments - CSHC of the Loess Plateau draining the main south flowing branch of the Yellow River) and ensuring that scientific publications from the CSHC were accepted in international peer-reviewed journals. Specifically we: (1) completed several moderate changes to our bilingual decision support tool called ReVegIH (Re-Vegetation Impact on Hydrology) that were requested by the project review team and key stakeholders during our final review meeting held in Oct 2005; (2) training computer programmers from China to successfully make changes to ReVegIH; (3) successfully responded to external reviewer comments for our paper documenting the development of the hydrologically correct Digital Elevation Model (DEM); (4) submitted, and successfully responded to the external reviewer's comments, for an internationally peer-reviewed journal paper documenting design principles and use of ReVegIH bilingual decision support tool; (5) submitted, and successfully responded to the external reviewer's comments, for internationally peer-reviewed journal paper documenting methods to spatial distribute key hydrological variables for the study site; and (6) submitted an internationally peer-reviewed journal paper documenting methods to calibrate a global steady-state land-use / water-balance framework to the local conditions of the CSHC. The importance of each is briefly discussed.

(1) Incorporating feedback / comments from stakeholders and reviewers into: A near final prototype of the bilingual computer simulation tool called ReVegIH was shown to our project reviewers and key Chinese stakeholders during our final review meeting that was held in China in October 2005. At that meeting several enhancements increasing the functionality, and hence utility of ReVegIH, were suggested. To maximise uptake of the tool by the stakeholders we implemented all changes early in 2006. We ensured that the final version of ReVegIH can be distributed to environmental managers in the CSHC free of charge on one CD. ReVegIH provides a means for users to: (a) determine where priority (and target) re-vegetation activities should be undertaken; (b) ascertain what species are suitable for re-vegetation at specific locations; and (c) simulate the related mean annual hydrological impact.

(2) Training Chinese computer scientists to make changes to ReVegIH: To ensure that the Chinese research agency involved in this project have the skills to maintain and update the ReVegIH bilingual decision support tool, two computer programmers travelled to Australia for one month and received training specific to the computing language and system architecture that ReVegIH was developed in. This means that the Chinese research agency now has the skills to make changes to the decision support tool as they wish, without requiring involvement of Australian scientists. This training constitutes a major capability enhancement of the Chinese research agency.

(3) Addressing external reviewer comments on our DEM journal paper: To ensure that only high quality scientific papers are published, journals require that all papers submitted be scrutinised by anonymous scientific peers, who are also specialists in the topic of the paper, called external reviewers. The external reviewers usually have comments / questions that need to be addressed before the editor of a journal accepts the paper for publication (the external reviewers also may suggest the paper be rejected for publication). To ensure our paper documenting the development of a hydrologically correct DEM was accepted into the International Journal of Applied Earth Observation and Geoinformation we had to make moderate changes to the paper.

(4) Submitting and addressing external reviewer comments on our design of ReVegIH journal paper: Our paper reviewing the literature on the effect that land-use change has on regional hydrology in the Loess Plateau (hence identifying the need for ReVegIH), and documenting the ethos of ReVegIH's design, and providing examples of its functionality was submitted and accepted, after minor changes, to a Special Issue of Forest Ecology and Management.

(5) Submitting and addressing external reviewer comments on spatial distribution of required hydrometeorological variables paper: Our paper discussing the generation of surfaces of crop reference evapotranspiration, pan evaporation, the so-called pan coefficient (and the forcing hydrometeorological data needed) was submitted to Journal of Hydrology in August. Minor changes were implemented to address the limited questions raised by the external reviewers.

(6) Submitting our paper discussing local calibration of the land use-hydrology framework: Our paper discussing generation of the local calibration of the previously globally developed land-use hydrology framework (that underpins the scenario modelling capacity in ReVegIH) was submitted to Hydrological Processes in September. At the time of writing this report we have just received the two external reviews for that paper and given minor changes to our paper this should be accepted into this journal.