Search Results

You are looking at 1 - 10 of 43 items for

  • Author or Editor: Sujay Kumar x
  • Refine by Access: All Content x
Clear All Modify Search
Soni Yatheendradas
and
Sujay Kumar

Abstract

Satellite-based remotely sensed observations of snow cover fraction (SCF) can have data gaps in spatially distributed coverage from sensor and orbital limitations. We mitigate these limitations in the example fine-resolution Moderate Resolution Imaging Spectroradiometer (MODIS) data by gap-filling using auxiliary 1-km datasets that either aid in downscaling from coarser-resolution (5 km) MODIS SCF wherever not fully covered by clouds, or else by themselves via regression wherever fully cloud covered. This study’s prototype predicts a 1-km version of the 500-m MOD10A1 SCF target. Due to noncollocatedness of spatial gaps even across input and auxiliary datasets, we consider a recent gap-agnostic advancement of partial convolution in computer vision for both training and predictive gap-filling. Partial convolution accommodates spatially consistent gaps across the input images, effectively implementing a two-dimensional masking. To overcome reduced usable data from noncollocated spatial gaps across inputs, we innovate a fully generalized three-dimensional masking in this partial convolution. This enables a valid output value at a pixel even if only a single valid input variable and its value exist in the neighborhood covered by the convolutional filter zone centered around that pixel. Thus, our gap-agnostic technique can use significantly more examples for training (∼67%) and prediction (∼100%), instead of only less than 10% for the previous partial convolution. We train an example simple three-layer legacy super-resolution convolutional neural network (SRCNN) to obtain downscaling and regression component performances that are better than baseline values of either climatology or MOD10C1 SCF as relevant. Our generalized partial convolution can enable multiple Earth science applications like downscaling, regression, classification, and segmentation that were hindered by data gaps.

Full access
Wade T. Crow
,
Hyunglok Kim
, and
Sujay Kumar

Abstract

Due to recent advances in the development of land data assimilation systems (LDAS) and the availability of high-quality, satellite-based surface soil moisture (SSM) retrieval products, we now have unambiguous evidence that the assimilation of SSM retrievals, or their proxy, can improve the precision (i.e., correlation versus truth) of surface state estimates provided by a land surface model (LSM). However, this clarity does not yet extend to the estimation of LSM surface water fluxes that are key to hydrologic and numerical weather forecasting applications. Here, we hypothesize that a key obstacle to extrapolating realized improvements in water state precision into comparable improvements in water flux accuracy (i.e., mean absolute error) is the presence of water state–water flux coupling strength biases existing in LSMs. To test this hypothesis, we conduct a series of synthetic fraternal twin data assimilation experiments where realistic levels of state–flux coupling strength bias—involving both evapotranspiration and runoff—are systematically introduced into an assimilation LSM. Results show that the accuracy of the resulting water flux analysis is sharply reduced by the presence of such bias, even in cases where the precision of soil moisture state estimates (e.g., SSM) is improved. The rescaling of SSM observations prior to their assimilation (i.e., the most common approach for addressing systematic differences between LSMs and assimilated observations) is not always a robust strategy for addressing these errors and can, in certain circumstances, degrade water flux accuracy. Overall, results underscore the critical need to assess, and correct for, LSM water state–water flux coupling strength biases during the operation of an LDAS.

Significance Statement

Land data assimilation is the process by which land surface model estimates of water states (e.g., soil moisture) and water fluxes (e.g., runoff and evapotranspiration) are improved via the incorporation of observations. Over the past decade, substantial improvements have been made in the precision of land surface model states via the assimilation of satellite-based soil moisture information. However, to date, these improvements have not yet been extended into water flux estimates like runoff and evapotranspiration. This is a critical shortcoming since advances in important weather and hydrologic forecasting applications are dependent on the improved estimation of such fluxes. We demonstrate that this shortcoming is linked to the inability of existing land surface models to accurately describe the impact of variations in water states on water fluxes and propose strategies for overcoming this issue in future land data assimilation systems.

Restricted access
Soni Yatheendradas
,
David M. Mocko
,
Christa Peters-Lidard
, and
Sujay Kumar

Abstract

Using information theory, our study quantifies the importance of selected indicators for the U.S. Drought Monitor (USDM) maps. We use the technique of mutual information (MI) to measure the importance of any indicator to the USDM, and because MI is derived solely from the data, our findings are independent of any model structure (conceptual, physically based, or empirical). We also compare these MIs against the drought representation effectiveness ratings in the North America Drought Indices and Indicators Assessment (NADIIA) survey for Köppen climate zones. This reveals 1) agreement between some ratings and our MI values [high for example indicators like standardized precipitation evapotranspiration index (SPEI)]; 2) some divergences (e.g., soil moisture has high ratings but near-zero MIs for ESA Climate Change Initiative (CCI) soil moisture in the Western United States, indicating the need of another remotely sensed soil moisture source); and 3) new insights into the importance of variables such as snow water equivalent (SWE) that are not included in sources like NADIIA. Further analysis of the MI results yields findings related to 1) hydrological mechanisms (summertime SWE domination during individual drought events through snowmelt into the water-scarce soil); 2) hydroclimatic types (the top pair of inputs in the Western and non-Western regions are SPEIs and soil moistures, respectively); and 3) predictability (high for the California 2012–17 event, with longer-time scale indicators dominating). Finally, the high MIs between multiple indicators jointly and the USDM indicate potentially high drought forecasting accuracies achievable using only model-based inputs, and the potential for global drought monitoring using only remotely sensed inputs, especially for locations having insufficient in situ observations.

Significance Statement

Drought maps from the U.S. Drought Monitor and the Objective Short- and Long-Term Drought Indicator Blends and Blend Equivalents are integrated information sources of the different types of drought. Multiple indicators go into creation of these maps, yet it is usually not clear to both public and private stakeholders like local agencies and insurance companies about the importance of any indicator in any region and season to the drought maps. Our study provides such objective information to enable understanding the mechanism and type of drought occurring at a location, season, and possibly event of interest, as well as to potentially aid in better drought monitoring and forecasting using smaller custom sets of indicators.

Restricted access
Joseph A. Santanello Jr.
,
Christa D. Peters-Lidard
, and
Sujay V. Kumar

Abstract

The inherent coupled nature of earth’s energy and water cycles places significant importance on the proper representation and diagnosis of land–atmosphere (LA) interactions in hydrometeorological prediction models. However, the precise nature of the soil moisture–precipitation relationship at the local scale is largely determined by a series of nonlinear processes and feedbacks that are difficult to quantify. To quantify the strength of the local LA coupling (LoCo), this process chain must be considered both in full and as individual components through their relationships and sensitivities. To address this, recent modeling and diagnostic studies have been extended to 1) quantify the processes governing LoCo utilizing the thermodynamic properties of mixing diagrams, and 2) diagnose the sensitivity of coupled systems, including clouds and moist processes, to perturbations in soil moisture. This work employs NASA’s Land Information System (LIS) coupled to the Weather Research and Forecasting (WRF) mesoscale model and simulations performed over the U.S. Southern Great Plains. The behavior of different planetary boundary layers (PBL) and land surface scheme couplings in LIS–WRF are examined in the context of the evolution of thermodynamic quantities that link the surface soil moisture condition to the PBL regime, clouds, and precipitation. Specifically, the tendency toward saturation in the PBL is quantified by the lifting condensation level (LCL) deficit and addressed as a function of time and space. The sensitivity of the LCL deficit to the soil moisture condition is indicative of the strength of LoCo, where both positive and negative feedbacks can be identified. Overall, this methodology can be applied to any model or observations and is a crucial step toward improved evaluation and quantification of LoCo within models, particularly given the advent of next-generation satellite measurements of PBL and land surface properties along with advances in data assimilation schemes.

Full access
Sujay V. Kumar
,
Christa D. Peters-Lidard
,
David Mocko
, and
Yudong Tian

Abstract

The downwelling shortwave radiation on the earth’s land surface is affected by the terrain characteristics of slope and aspect. These adjustments, in turn, impact the evolution of snow over such terrain. This article presents a multiscale evaluation of the impact of terrain-based adjustments to incident shortwave radiation on snow simulations over two midlatitude regions using two versions of the Noah land surface model (LSM). The evaluation is performed by comparing the snow cover simulations against the 500-m Moderate Resolution Imaging Spectroradiometer (MODIS) snow cover product. The model simulations are evaluated using categorical measures, such as the probability of detection of “yes” events (PODy), which measure the fraction of snow cover presence that was correctly simulated, and false alarm ratio (FAR), which measures the fraction of no-snow events that was incorrectly simulated. The results indicate that the terrain-based correction of radiation leads to systematic improvements in the snow cover estimates in both domains and in both LSM versions (with roughly 12% overall improvement in PODy and 5% improvement in FAR), with larger improvements observed during snow accumulation and melt periods. Increased contribution to PODy and FAR improvements is observed over the north- and south-facing slopes, when the overall improvements are stratified to the four cardinal aspect categories. A two-dimensional discrete Haar wavelet analysis for the two study areas indicates that the PODy improvements in snow cover estimation drop to below 10% at scales coarser than 16 km, whereas the FAR improvements are below 10% at scales coarser than 4 km.

Full access
Joseph A. Santanello Jr.
,
Patricia Lawston
,
Sujay Kumar
, and
Eli Dennis

Abstract

The role of soil moisture in NWP has gained more attention in recent years, as studies have demonstrated impacts of land surface states on ambient weather from diurnal to seasonal scales. However, soil moisture initialization approaches in coupled models remain quite diverse in terms of their complexity and observational roots, while assessment using bulk forecast statistics can be simplistic and misleading. In this study, a suite of soil moisture initialization approaches is used to generate short-term coupled forecasts over the U.S. Southern Great Plains using NASA’s Land Information System (LIS) and NASA Unified WRF (NU-WRF) modeling systems. This includes a wide range of currently used initialization approaches, including soil moisture derived from “off the shelf” products such as atmospheric models and land data assimilation systems, high-resolution land surface model spinups, and satellite-based soil moisture products from SMAP. Results indicate that the spread across initialization approaches can be quite large in terms of soil moisture conditions and spatial resolution, and that SMAP performs well in terms of heterogeneity and temporal dynamics when compared against high-resolution land surface model and in situ soil moisture estimates. Case studies are analyzed using the local land–atmosphere coupling (LoCo) framework that relies on integrated assessment of soil moisture, surface flux, boundary layer, and ambient weather, with results highlighting the critical role of inherent model background biases. In addition, simultaneous assessment of land versus atmospheric initial conditions in an integrated, process-level fashion can help address the question of whether improvements in traditional NWP verification statistics are achieved for the right reasons.

Full access
Patricia Lawston-Parker
,
Joseph A. Santanello Jr.
, and
Sujay V. Kumar

Abstract

Accurately representing land–atmosphere (LA) interactions and coupling in NWP systems remains a challenge. New observations, incorporated into models via assimilation or calibration, hold the promise of improved forecast skill, but erroneous model coupling can hinder the benefits of such activities. To better understand model representation of coupled interactions and feedbacks, this study demonstrates a novel framework for coupled calibration of the single column model (SCM) capability of the NASA Unified Weather Research and Forecasting (NU-WRF) system coupled to NASA’s Land Information System (LIS). The local land–atmosphere coupling (LoCo) process chain paradigm is used to assess the processes and connections revealed by calibration experiments. Two summer case studies in the U.S. Southern Great Plains are simulated in which LSM parameters are calibrated to diurnal observations of LoCo process chain components including 2-m temperature, 2-m humidity, surface fluxes (Bowen ratio), and PBL height. Results show a wide range of soil moisture and hydraulic parameter solutions depending on which LA variable (i.e., observation) is used for calibration, highlighting that improvement in either soil hydraulic parameters or initial soil moisture when not in tandem with the other can provide undesirable results. Overall, this work demonstrates that a process chain calibration approach can be used to assess LA connections, feedbacks, strengths, and deficiencies in coupled models, as well as quantify the potential impact of new sources of observations of land–PBL variables on coupled prediction.

Full access
Bailing Li
,
Matthew Rodell
,
Christa Peters-Lidard
,
Jessica Erlingis
,
Sujay Kumar
, and
David Mocko

Abstract

Estimating diffuse recharge of precipitation is fundamental to assessing groundwater sustainability. Diffuse recharge is also the process through which climate and climate change directly affect groundwater. In this study, we evaluated diffuse recharge over the conterminous United States simulated by a suite of land surface models (LSMs) that were forced using a common set of meteorological input data. Simulated annual recharge exhibited spatial patterns that were similar among the LSMs, with the highest values in the eastern United States and Pacific Northwest. However, the magnitudes of annual recharge varied significantly among the models and were associated with differences in simulated ET, runoff, and snow. Evaluation against two independent datasets did not answer the question of whether the ensemble mean performs the best, due to inconsistency between those datasets. The amplitude and timing of seasonal maximum recharge differed among the models, influenced strongly by model physics governing deep soil moisture drainage rates and, in cold regions, snowmelt. Evaluation using in situ soil moisture observations suggested that true recharge peaks 1–3 months later than simulated recharge, indicating systematic biases in simulating deep soil moisture. However, recharge from lateral flows and through preferential flows cannot be inferred from soil moisture data, and the seasonal cycle of simulated groundwater storage actually compared well with in situ groundwater observations. Long-term trends in recharge were not consistently correlated with either precipitation trends or temperature trends. This study highlights the need to employ dynamic flow models in LSMs, among other improvements, to enable more accurate simulation of recharge.

Full access
Fadji Z. Maina
,
Sujay V. Kumar
,
Ishrat Jahan Dollan
, and
Viviana Maggioni

Abstract

Precipitation estimates are highly uncertain in complex regions such as High Mountain Asia (HMA), where ground measurements are very difficult to obtain and atmospheric dynamics poorly understood. Though gridded products derived from satellite-based observations and/or reanalysis can provide temporally and spatially distributed estimates of precipitation, there are significant inconsistencies in these products. As such, to date, there is little agreement in the community on the best and most accurate gridded precipitation product in HMA, which is likely area dependent because of HMA’s strong heterogeneities and complex orography. Targeting these gaps, this article presents the development of a consensus ensemble precipitation product using three gridded precipitation datasets [the Integrated Multi-satellitE Retrievals for Global Precipitation Measurement (IMERG), the Climate Hazards Group Infrared Precipitation with Station data (CHIRPS), and the ECMWF reanalysis ERA5] with a localized probability matched mean (LPM) approach. We evaluate the performance of the LPM estimate along with a simple ensemble mean (EM) estimate to overcome the differences and disparities of the three selected constituent products on long-term averages and trends in HMA. Our analysis demonstrates that LPM reduces the high biases embedded in the ensemble members and provides more realistic spatial patterns compared to EM. LPM is also a good alternative for merging data products with different spatiotemporal resolutions. By filtering disparities among the individual ensemble members, LPM overcomes the problem of a certain product performing well only in a particular area and provides a consensus estimate with plausible temporal trends.

Free access
Rachel T. Pinker
,
Wen Chen
,
Yingtao Ma
,
Sujay Kumar
,
Jerry Wegiel
, and
Eric Kemp

Abstract

We present a global-scale evaluation of surface shortwave (SW↓) radiative fluxes as derived with cloud amount information from the U.S. Air Force (USAF) Cloud Depiction Forecast System (CDFS) II World-Wide Merged Cloud Analysis (WWMCA) and implemented in the framework of the NASA Land Information System (LIS). Evaluation of this product is done against ground observations, a satellite-based product from the Moderate Resolution Imaging Spectroradiometer (MODIS), and several reanalysis outputs. While the LIS/USAF product tends to overestimate the SW↓ fluxes when compared to ground observations and satellite estimates, its performance is comparable or better than the following reanalysis products: ERA5, CFSR, and MERRA-2. Results are presented using all available observations over the globe and independently for several regional domains of interest. When evaluated against ground observations over the globe, the bias in the LIS/USAF product at daily time scale was about 9.34 W m−2 and the RMS was 29.20 W m−2 while over the United States the bias was about 10.65 W m−2 and the RMS was 35.31 W m−2. The sample sizes used were not uniform over the different regions, and the quality of both ground truth and the outputs of the other products may vary regionally. It is important to note that the LIS/USAF is a near-real-time (NRT) product of interest for potential users and as such fills a need that is not met by most products. Due to latency issues, the level of observational inputs in the NRT product is less than in the reanalysis data.

Significance Statement

We evaluate a current scheme to produce surface radiative fluxes in the NASA Land Information System (LIS) framework as driven with cloud amount information from the U.S. Air Force (USAF) Cloud Depiction Forecast System (CDFS) II World-Wide Merged Cloud Analysis (WWMCA). The LIS/USAF product is provided at near–real time and as such, fills a need that is not met by most products. Information used for evaluation are ground observations, MODIS satellite-based estimates, and independent outputs from several reanalysis. Since the various LIS products are used by the hydrometeorology community, this manuscript should be of interest to the users of the LIS/USAF information on surface radiative fluxes.

Free access