Study On Assessment Of Nutrients Loss Risk And Management Of Nutrients Balance In Agricultural Land

Nutrients (N and P) loss from the agricultural land to river is getting more and more serious in southwest Australia, which leads to water eutrophication in downstream areas. This study focused on the agricultural land near Jacup in southern Western Australia (WA). The first baseline monitoring data for stream water quality was provided after a three-year monitoring period (2005-2007). Results showed nitrogen and phosphorus concentrations in stream water had seriously exceeded. In this study, for the storm events happened from time to time and runoff phenomenon they caused in WA, soil loss was selected as the main driving force of the nutrients loss. An assessment of nutrient loss (set P as an example) was carried out to indentify higher loss risk areas. Meanwhile, the wheat which was the main crop in the study area was taken as an example to extract the planting area and estimate yield. P loss caused by soil loss was considered to estimate P balance status using remote sensing data and ground stastical materials. The main results obtained are summarized as follows:1. Water quality analysis of the study areaContinually data of five water quality monitoring sites was used to measure and analyze parameters such as conductivity, pH value and total suspended solids, and N & P concentrations in different forms. And this was the first measure and analysis for baseline water quality in WA. The concentrations of total nitrogen and total phosphorus (3.14mg/L and 0.60mg/L) were both exceeded the ANZECC trigger values (3.0mg/L and 0.30mg/L). The water quality was classified as poor grade. Stream water was rich in N and P. Regression models were built to search for a good correlationship between ordinary water quality parameters (such as EC or TSS) and different forms of N & P concentrations to assess the water quality more easily and economical. But most regression models got poor correlationship and could not be widely use. Thus it's more important to assess the nutrients loss driven by soil loss and analyze balance of nutrients in agricultural lands.2. Estimation of soil loss in study areaRevised Universal Soil Loss Equation (RUSLE) was applied to estimate soil loss for the period of 2005-2007 in study area. It was found that rainfall made most contribution to soil loss. According to twelve-year historical rainfall data, monthly rainfall was basicly accord with cosine function distribution. During the year of 2005-2007 rainfall in the same month varied much, the main reason may mostly came to the individual storm events. Especially in dry summer, sometimes rainfall from one storm could be similar with whole rainfall quality in that month, and led to serious soil loss. Thus the rainfall erosivity factor was calculated based on daily rainfall data to reflect the characteristic of rainfall and its influence to soil loss in WA. An Australian-modified equation was used for calculating the Soil erodibility factor. The percentage of particles with diameter less than 0.125mm, percent organic carbon, and soil permeability rating were the input. Soil texture data and SOILOSS were also used. Slope length factor and slope-steepness factor were calculated in GIS software. Results showed the terrain slopes gently. The percent slope was less than 9% or almost all the area. The terrain there didn't affect the soil loss much. But the terrain of Fitzgerald National Park to the southeast of the study area has a undulating terrain, the elevation discrepancy there would become a factor to cause soil loss. The cover and management factor was calculated using soil loss ratio (SLR), which was suggested monthly according to the location of study area and crop types in SOILOSS. It was found that the study area had a good vegetation cover all year round and had good conservation of water and soil.Soil loss during the year of 2005 and 2007 were 3.04t ha-1 yr-1,1.62 t ha-1yr-1, and 1.19t ha-1 yr, which could be classified as low to moderate grade according to the Australian soil loss guideline. The extent distribution of soil loss was uneven. There was severe soil loss in some regions. The soil loss of some region could even come up to 31.54t ha-1 yr-1.3. Assessement of nutrients loss riskAccording to the relationship between particulate and dissolved phosphorus in different soil available phosphorus (Colwell P) concentrations, and soil loss in study area, an assessment of phosphorus loss risk was carried out using matrix method.A spatial distribution for soil available phosphorus was firstly calculated. Box-Cox method was selected for data transfering to pass the Kolmogorov-Smirnov test. The exponential model reasonably fitted the Box-Cox transformed data. Ordinary Kriging interpolation was used to obtain the spatial distribution map of Colwell P. High values of Colwell P occurred mainly in the western part of the study area. This spatial distribution of soil P may be caused by differences in soil type, but is more likely due to the longer cropping history and longer accumulation of P from fertilisers in the western part of the study area.Areas with low to moderate risk of potential P loss were mainly in the eastern and southern parts of the study area, while corresponding areas with higher risks were in the middle and northwest part (approximately 6.5% of the total area). The distribution also showed regionalization. Comparison of water quality across areas with risk of P loss showed that P concentrations of some monitoring site had similar trend with loss risk, but was imperfect. This was simply because in this initial study we only focused on erosion loss of particulate materials. Additional pathways are likely because dissolved and organic forms of P were important contributors of the stream P load and the broader range of transport mechanisms need to be taken into account.4. Yield estimation and nutrients balance research on wheatA methodology was put forward based on remote sensing images and ground stastical collection. The methodology included preprocess and classification of images, extraction of crop planting area, calculation of vegetation index, building yield estimating models, estimation of yield, calculation of nutrients'input and output, and analysis of nutrients balance.Landsat-5 TM images were selected as data source. Supervised (Maximum likelihood and SVM) and unsupervised (ISODATA) classification were carried out in study area. For the plant of wheat, the sensitivity and specificity were similar between SVM and ISODATA, and were corresponding low with maximum likelihood method. The ISODATA was chosen for better feasibility. An error matrix was used to test the classification results again to show the accuracy of ISODATA. The area of wheat planting was extracted and validated to meet the demand of study. After building regression models between NDVI and yield data, it was found that it's better to use the planting area extracted from classification rather than the planting area from the stastical data. And quadratic polynomial regression gave the best results. The paddocks which were not used in building the models were then used for validation. The precision of total yield (t) and per unit yield (t/ha) were 87.76% and 91.15%.The fertilizers were considered as main input of nutrients. Meanwhile the nutrients that harvest took away (including grains and straws) and transported by soil loss were the main output. The discrepency of input and output gave balance status. It was concluded from the results that usage of P fertilizers and P enrichment in soil made huge amount of P input, although the harvest and soil loss caused comparatively output. P was generally excessive in the study area. Total P surplus was 2.59 ton in study area in the year of 2003. Measures should be made to prevent further P accumulation in soil. And protection of soil loss should be reinforced to decline output in high P loss risk years.