A backward‐averaged iterative two‐source surface temperature and energy balance solution (BAITSSS) algorithm was developed to estimate evapotranspiration (ET) during the period between Landsat satellite overpass dates. The METRIC (mapping evapotranspiration at high resolution with internalized calibration) model was used to estimate ET using short wave and thermal data from the Landsat images. METRIC generated ET was used to define initial surface characteristics, soil water conditions and initialize the soil water content for the surface and root zone at the start of the simulation period, and to adjust the results from BAITSSS at the next satellite overpass date if needed. North American Regional Reanalysis (NARR) weather data were used to estimate the surface energy balance components in between the satellite overpasses. The fraction of vegetation cover (fc) was used to partition surface energy balance components, as defined by the normalized difference vegetation index. Soil surface resistance (rss) and Jarvis‐model‐type canopy resistance (rsc) were used to calculate latent heat flux using the aerodynamic equations. A water balance was implemented to track the water content at the surface and root zone. An irrigation sub‐model was developed to consider the role of irrigation for known irrigated agricultural fields, which is critical when computing ET in an agriculture‐dominant area. Any mismatch between the estimated and METRIC ET at the next satellite overpass can be adjusted back over the simulation period with a time‐based linear correction to increase the accuracy and reduce computation time. In this study, BAITSSS results from southern Idaho and northern California are presented for 3 h time steps.

A surface energy balance was conducted to calculate the latent heat flux (λE) using aerodynamic methods and the Penman-Monteith (PM) method. Computations were based on gridded weather and Landsat satellite reflected and thermal data. The surface energy balance facilitated a comparison of impacts of different parameterizations and assumptions, while calculating λE over large areas through the use of remote sensing. The first part of the study compares the full aerodynamic method for estimating latent heat flux against the appropriately parameterized PM method with calculation of bulk surface resistance (r s ). The second part of the study compares the appropriately parameterized PM method against the PM method, with various relaxations on parameters. This study emphasizes the use of separate aerodynamic equations (latent heat flux and sensible heat flux) against the combined Penman-Monteith equation to calculate λE when surface temperature (T s ) is much warmer than air temperature (T a ), as will occur under water stressed conditions. The study was conducted in southern Idaho for a 1000-km 2 area over a range of land use classes and for two Landsat satellite overpass dates. The results show discrepancies in latent heat flux (λE) values when the PM method is used with simplifications and relaxations, compared to the appropriately parameterized PM method and full aerodynamic method. Errors were particularly significant in areas of sparse vegetation where differences between T s and T a were high. The maximum RMSD between the correct PM method and simplified PM methods was about 56 W/m 2 in sparsely vegetated sagebrush desert where the same surface resistance was applied.

Abstract:Groundwater depletion in the face of growth is a well-known problem, particularly in those areas that have grown to become dependent on a declining resource. This research comprises a broad synthesis of existing water resources data, to understand the long-term implications of continued growth in water demand on groundwater dominant water resources, and to develop a tool for sustainable water management. The Palouse region of Washington and Idaho, USA. (approximately 60,000 people in a rural setting) is entirely dependent on groundwater from two basalt aquifers for potable water. Using the systems dynamics approach and a water balance that considered the entire hydrologic cycle, a hydrologic model of these aquifers was developed, tested and applied to simulate their behavior over a 150 year time period assuming the current infrastructure does not change. With 1% population growth and current water extraction rates, the results indicated the upper aquifer use may be sustainable, while the lower aquifer use is likely unsustainable in the long term. This study also shows that uncertainties in key aspects of the system create limitations to groundwater management.

The backward‐averaged iterative two‐source surface temperature and energy balance solution (BAITSSS) model was developed to calculate evapotranspiration (ET) at point to regional scales. The BAITSSS model is driven by micrometeorological data and vegetation indices and simulates the water and energy balance of the soil and canopy sources separately, using the Jarvis model to calculate canopy resistance. The BAITSSS model has undergone limited testing in Idaho, United States. We conducted a blind test of the BAITSSS model without prior calibration for ET against weighing lysimeter measurements, net radiation, and surface temperature of drought‐tolerant corn (Zea mays L. cv. PIO 1151) in a semiarid, advective climate (Bushland, Texas, United States) in 2016. Later in the season (20 days), BAITSSS consistently overestimated ET by up to 3 mm d−1. For the entire growing season (127 days), simulated versus measured ET resulted in a 7% error in cumulative ET, RMSE = 0.13 mm h−1, and 1.70 mm d−1; r2 = 0.66 (daily) and r2 = 0.84 (hourly); MAE = 0.08 mm h−1 and 1.24 mm d−1; and MBE = 0.02 mm h−1 and 0.58 mm d−1. The results were comparable with thermally driven instantaneous ET models that required some calibration. Next, the initial soil water boundary condition was reduced, and model revisions were made to resistance terms related to incomplete cover and assumption of canopy senescence. The revisions reduced discrepancies between measured and modelled ET resulting in <1% error in cumulative ET, RMSE = 0.1 mm h−1, and 1.09 mm d−1; r2 = 0.86 (daily) and r2 = 0.90 (hourly); MAE = 0.06 mm h−1 and 0.79 mm d−1; and MBE = 0.0 mm h−1 and 0.17 mm d−1 and generally mitigated the previous overestimation. The advancement in ET modelling with BAITSSS assists to minimize uncertainties in crop ET modelling in a time series.

No abstract