ESRL/PSD Seminar Series

Abstract

---

As a primary component of the land-surface water budget, evapotranspiration (ET) varies significantly at small time- and space-scales. The estimation of ET in operational models of NWS river forecast centers (RFCs) is extremely crude, based on inputs of fixed monthly, spatially lumped climatologic pan evaporation (Epan) to meter atmospheric evaporative demand (E0). In river forecasting, this crude ET treatment introduces streamflow simulation and forecasting errors at many time-scales, affecting both flood and drought products and services. As water scarcity becomes more common in parts of the US, RFCs face increasing interest in predictions not only for water supply, but also for water demand, and in assessing the sensitivity of streamflow to climate change and variability, RFCs cannot meet these needs without improvements to the current ET treatment in RFC models and better datasets describing the time-space variability of ET and E0. There is currently no high spatio-temporal resolution, unbiased, CONUS-wide, real-time dataset of E0 with a real-time latency (lag from present time) to support the monitoring of drought development, and the absence of such predictive datasets restricts avenues for drought forecasting. The goals of this ongoing project are primarily two-fold: 1. The Colorado Basin River Forecast Center (CBRFC) seeks to improve the current treatment of ET in the Sacramento Soil Moisture Accounting (Sac-SMA) model that underpins their river forecast operations, and thereby to improve streamflow forecast skill at daily operational and seasonal time-scales. 2. A second goal, posed by various NWS Weather Forecast Offices (WFO) and their stakeholders is to provide scientifically sound, web-disseminated, fine-resolution, accurate, daily-to-weekly forecasts of reference crop ET (ETrc), and to develop a 30-year climatology to add value to these forecasts. For the first goal, we are examining driving the Sac-SMA model with seven formulations of E0-five physically based and two temperature-based. We have generated 30-year reanalyses (1979 to the near-present) of each E0 formulation forced by North American Land Data Assimilation System (NLDAS) data at six-hourly and daily time-steps, CONUS-wide at a 1/8th-degree spatial resolution. To address both goals, we have established the first ETrc forecast/reanalysis system at an RFC-operational scale and resolution. Forecasts are made at WFOs on their existing forecast platform with no new forecast fields required. The system is currently running on an experimental basis at four WFOs in CA, OR, and MT. Daily and weekly forecasts are issued daily at a 2.5-km resolution, and are driven by existing forecasts from the National Digital Forecast Database. The forecasts are web-disseminated on graphical and tabular bases. CONUS-wide coverage is anticipated. Within CBRFC operations, the improved representation of the physics of ET should improve water supply forecast skill, both in daily operations (e.g., forecasting low flows, drought, peak-flow estimates and timings) and in seasonal forecasts (for water supply). Treating the final piece of the hydrologic cycle in a temporally dynamic, physically rigorous manner will allow climate change analyses across the Colorado River basin within the RFC model infrastructure. Beyond CBRFC operations, potential end-users of the E0 reanalyses and ETrc forecasts are numerous at various temporal scales, and may include the aforementioned irrigation scheduling; monitoring ongoing drought; forecasting drought development; examining historical drought trends; demand-planning to facilitate seasonal operations at utility districts and by industry; and reservoir operations.

---