BAITSSSThe scientific responsibility of BAITSSS has two equal parts: understanding how a result should be traced, evaluated, and interpreted, and understanding the physical model that produces that result. Scientific stewardship comes first here because the meaning of a model result depends on the evidence, assumptions, uncertainty, and context carried with it.
A model result is scientifically useful only when its origin, configuration, evidence, uncertainty, limitations, and interpretation remain visible. The following stewardship story comes before the model formulation for that reason.
The BAITSSS model was developed to represent a continuously changing field system. Weather changes every hour. Vegetation develops through the season. Soil water is depleted and replenished. Energy exchange changes with atmospheric conditions, crop state, and water availability. Because each hour depends on the state carried from the hour before it, a final result is inseparable from the data, assumptions, and decisions that produced that trajectory.
That is why scientific stewardship is different from simply describing the equations. A scientifically responsible analysis must preserve the connection between what was observed, what was supplied, what the BAITSSS model calculated, what BAITSSS Desktop recorded, and what the researcher finally concluded.
As the earlier research implementation evolved into BAITSSS Desktop, this became increasingly important. A desktop system can make complex simulation easier to run, inspect, repeat, and export. That convenience is valuable, but it also increases the need to keep the scientific meaning of each quantity visible. A successful software run is not, by itself, scientific validation. A visually convincing map is not, by itself, evidence of accuracy. A modeled irrigation requirement is not automatically the irrigation actually applied in the field.
Many interpretation problems begin when different kinds of information are blended together. BAITSSS work is clearer when the scientific record remains separated into four classes.
Satellite measurements, field observations, weather measurements or forcing products, irrigation records, lysimeter data, eddy covariance, thermal observations, and other independent records retain their own measurement basis and uncertainty.
AOI, dates, soil properties, vegetation information, rooting depth, management assumptions, initial conditions, parameters, and other configured values define the simulation being asked of the BAITSSS model.
ET components, surface and root-zone soil water, modeled irrigation, resistance behavior, and modeled soil and canopy temperatures are quantities calculated by the BAITSSS model.
Statements about agreement, disagreement, field condition, management, performance, transferability, or applicability are conclusions. They are not direct model outputs.
The distinction matters because each class carries a different authority. An observation is not made more accurate because it is used by a model. A supplied parameter does not become an observation because it influences the result. A modeled variable does not become measured simply because it resembles a field value. And a scientific conclusion should never be presented as though it came directly from the model calculation.
A BAITSSS model result carries the influence of its forcing, vegetation representation, soil properties, initial water state, management assumptions, spatial scale, temporal matching, and model structure.
Weather forcing controls the atmospheric environment in which energy and water exchange are calculated. Satellite-derived vegetation information helps represent the evolving canopy. Soil hydraulic properties influence depletion, storage, evaporation, root-zone behavior, and irrigation response. Initial soil moisture determines where the simulated water balance begins. Management settings such as rooting depth and management allowable depletion alter how the modeled system responds to water stress and irrigation thresholds.
Those influences are not errors in themselves. They are part of the scientific representation. Stewardship means identifying them, preserving them, and testing their importance where the research question requires it.
Scale is equally important. A 30 m BAITSSS model pixel, a field average, a point soil-water measurement, an eddy-covariance footprint, a lysimeter, and a regional ET product do not represent exactly the same spatial support. Likewise, hourly, daily, seasonal, and annual quantities answer different questions. A comparison becomes stronger when those differences are acknowledged rather than hidden.
Evapotranspiration. BAITSSS model ET comes from the coupled water and energy balance. Agreement with an independent ET observation or product is evidence for that particular comparison. It is not a universal accuracy statement.
Soil water. Surface and root-zone soil-water states reflect the soil representation, initialization, precipitation, irrigation, evaporation, transpiration, and model structure used in that simulation. Their interpretation is therefore conditional on those choices.
Irrigation. When the BAITSSS model simulates irrigation, the value represents irrigation required under the configured model rules. It should not be described as farmer-applied irrigation unless independent irrigation records establish that correspondence.
Temperature. Soil and canopy temperatures are calculated internally by the BAITSSS model. Independent thermal measurements can be used for comparison, but measured thermal information and modeled temperature remain different evidence classes.
The BAITSSS model has been evaluated in peer-reviewed studies using independent field observations and comparison datasets. Those studies address scientific questions: how the model behaved at a defined site, during a defined period, under particular crops, climates, measurements, and input conditions.
BAITSSS Desktop introduces another responsibility. The software must also be shown to execute the intended science correctly. Numerical equivalence testing asks whether a changed execution route preserves the accepted scientific behavior. Software verification asks whether project state, data readiness, run identity, Stop and Resume, outputs, provenance, packaging, and related functions behave as intended.
These evidence classes complement each other, but they are not interchangeable. Software verification cannot substitute for field validation, and field validation cannot prove that every software pathway is implemented correctly.
BAITSSS Desktop is not the scientific model. It is the maintained software environment that runs the BAITSSS model and keeps the project, data, configuration, execution, Results, and provenance connected.
That distinction is important because reproducibility depends on more than saving an output file. A defensible result should remain tied to the project and AOI, simulation period, BAITSSS Desktop build or version, model settings, data sources, input provenance, and the exact run that produced it.
Traceability allows a result to be revisited. It allows a collaborator to understand what was done. It allows numerical or scientific differences to be investigated instead of guessed at. It also makes clear when two simulations differ because of changed science, changed data, changed configuration, or changed software behavior.
For research, education, collaboration, validation, or professional reporting, the scientific record should make the path from input to conclusion visible. At minimum, the study should identify:
The purpose is not to burden every analysis with unnecessary detail. It is to preserve enough detail that the result can be understood, reproduced where possible, challenged where necessary, and improved when better data or methods become available.
The longer-term BAITSSS direction is broader than a single model run. Additional forcing routes, uncertainty analysis, forecasting, new observations, management applications, and decision-support functions may eventually make the scientific environment more useful. But each new capability changes the evidence chain and therefore creates new responsibilities for testing, validation, documentation, and interpretation.
For that reason, future development should not be judged only by whether BAITSSS Desktop can produce more outputs. It should also be judged by whether those outputs remain traceable to the BAITSSS model science, whether new assumptions are visible, whether changed execution is numerically verified where appropriate, and whether scientific claims remain proportional to the evidence.
The BAITSSS model is supported by peer-reviewed research. View study evidence.
BAITSSS means Backward-Averaged Iterative Two-Source Surface temperature and energy balance Solution. The BAITSSS model is a two-source resistance-based biophysical surface-energy and soil-water balance model.
The BAITSSS model links a 30 m Landsat based field grid with vegetation state, hourly meteorological forcing, a two-source energy balance, a two-layer soil-water balance, modeled irrigation, and hour to hour state progression to produce spatial and time-series Results.
Landsat 7, 8, and 9, NDVI, LAI, and the 30 m model grid establish vegetation and spatial context.
Meteorological forcing and reference ET preparation provide the changing atmospheric conditions for each model hour.
Soil and canopy exchange are represented separately, partitioning evaporation and transpiration within total ET.
Surface and root-zone water states are carried forward and coupled to the hourly energy balance.
Configured vegetation, soil-water, and management conditions determine modeled irrigation response.
ET, soil water, irrigation, thermal behavior, spatial variability, time series, and selected 30 m pixel histories remain tied to the run.
The BAITSSS model solves the surface energy balance using separate soil and canopy source terms and resistance-based exchange. Latent heat flux is partitioned into soil evaporation and canopy transpiration.
The BAITSSS model uses aerodynamic equations for sensible and latent heat exchange, a Jarvis-type canopy-resistance formulation for transpiration, and soil-surface resistance for evaporation.
Surface roughness, canopy conditions, meteorological forcing, and soil-water state affect the exchange calculations.
Weather supplies the forcing while vegetation affects radiation and exchange. Energy balance determines fluxes and modeled temperatures. Evaporation and transpiration deplete soil-water stores. Precipitation and configured irrigation replenish water, and the resulting state carries into the next hour.
NDVI and LAI are prepared from Landsat imagery within the workflow.
The BAITSSS model calculates soil and canopy temperature states internally as part of the two-source energy-balance solution.
Satellite and UAV thermal observations remain a research area and are not part of the current Phase I assimilation route.
The BAITSSS model couples hourly soil-water accounting with vegetation development. Irrigation is triggered when the configured root-zone condition reaches its threshold, subject to crop-development rules.
The standard configuration estimates the irrigation needed to avoid substantial modeled crop-water stress.
A user may configure a different crop-water target. That scenario should be identified clearly.
Without observed irrigation-event data, the BAITSSS model calculates modeled irrigation requirement. It does not report the exact amount applied by a producer.
Resistance-based surface energy-balance models can become numerically sensitive under weak atmospheric coupling, particularly at very low wind speed. The BAITSSS model was developed with a backward-averaged iterative two-source solution specifically to reduce these short-term instabilities; see the peer-reviewed iterative convergence study.
Some hours can require many iterations, and under difficult conditions additional iteration may no longer materially stabilize the solution. Abrupt hourly changes can therefore occur in modeled temperature, resistance, ET, or irrigation response. These effects are most important when interpreting individual hourly values; over longer accumulation periods, opposing short-term deviations can partly offset, so cumulative ET may differ much less than individual hourly values.
Future development will continue to refine this behavior while preserving the underlying hourly water- and energy-balance structure.