The overall goal of this research was to improve simulation of soil phosphorus (P) transport and transformations in GLEAMS 3.0, a non-point source model that simulates edge-of-field and bottom-of-root-zone loadings of nutrients from climate-soil-management interactions to assess management alternatives. The objectives of this research were to identify the state of the science for P transport and transformations, determine appropriate relationships for inclusion in GLEAMS, and determine if modifications to GLEAMS improved predictions of P loss in runoff, sediment, and leachate.

The state of the science review revealed numerous equations available to predict dissolved P loss in runoff and leachate from a soil's nutrient status. These equations use a single variable to predict P loss and were developed for site-specific conditions based on empirical data. Use of these equations in GLEAMS is not reasonable as transport factors must also be considered when predicting P loss.

Results from the sensitivity analysis showed that GLEAMS prediction of leached P were extremely sensitive to changes in the P partitioning coefficient (CPKD). Runoff PO₄-P output was slightly to moderately sensitive, sediment PO₄-P was moderately sensitive to sensitive, and sediment organic P was moderately sensitive to changes in CPKD whereas plant uptake of P was insensitive to slightly sensitive. The weakness of GLEAMS to estimate CPKD has been documented. Upon further investigation, it was determined that CPKD was highly over-estimated in GLEAMS as compared to measured values found during the literature review. Furthermore, this over-estimation caused under-estimation of the P extraction coefficient (BETA P); the value of BETA P remained constant at 0.10 and did not vary over the simulation period.

Expressions for CPKD and BETA P were modified in GLEAMS. Data from three published studies (Belle Mina, Gilbert Farm, and Watkinsville) were used in the analyses of three modifications to GLEAMS: GLEAMS BETA P, GLEAMS CPKD, and GLEAMS BETA P+CPKD. GLEAMS BETA P investigated the change in BETA P as a function of soil clay content, GLEAMS CPKD attempted to improve GLEAMS' estimation of CPKD, and GLEAMS BETA P+CPKD assessed the combined effects of changes to BETA P and CPKD.

Over the respective study periods, GLEAMS over predicted runoff PO₄-P for Belle Mina by 193 to 238% while under-predicting runoff PO₄-P at Gilbert Farm by 41% and Watkinsville by 81%. Sediment P was over-predicted by GLEAMS for Belle Mina by 225 to 233% and Gilbert Farm by 560%, while sediment P was under-predicted by 62% at Watkinsville. Leached PO₄-P was both over- and under-predicted by GLEAMS; Belle Mina was the only data set with observed leached P values.

Simulation results from the model changes were inconclusive. There was no clear evidence supporting use of one model over another. Modifications increased predicted dissolved P in runoff and leachate, while decreasing predicted sediment-bound P in runoff. The original GLEAMS model best predicted runoff and leached PO₄-P at the Belle Mina sites. GLEAMS CPKD was the best predictor of runoff PO₄-P and sediment P at Gilbert Farm. GLEAMS BETA P+CPKD best predicted runoff PO₄-P at Watkinsville. Overall, the proposed improvements to GLEAMS did not improve GLEAMS predictions.

In conclusion, GLEAMS should not be used for quantitative estimates of hydrology, sediment, and nutrient loss for specific management practices. As recommended by the GLEAMS model developers, GLEAMS should only be used to predict relative differences in alternative management systems. It is recommended that future research focus on developing a better correlation between CPKD, clay mineralogy and content, and organic matter content, as CPKD has been identified as a vital component of the GLEAMS P sub-model that requires further examination.